Dev Sci. Author manuscript; available in PMC 2012 Jan 1.
Someone in Omaha bought a Presidential Floor Clock Door Key With Tassel| 399868 Howard Miller for $ 19.00 3 days ago. Someone in Omaha bought a Grandfather Clock Brass Door Key| 238912 Howard Miller for $ 15.00 3 days ago.
- Obituaries for the last 7 days on Your Life Moments.
- With ongoing investment in switchgear development, the Socomec Automatic Transfer Switch (ATyS) monitors the power supply to a critical system and – in case of failure – manages the transfer.
Published in final edited form as:
doi: 10.1111/j.1467-7687.2010.00960.x
NIHMSID: NIHMS191035
The publisher's final edited version of this article is available at Dev Sci
See other articles in PMC that cite the published article.
Abstract
Deaf children with cochlear implants (CIs) represent an intriguing opportunity to study neurocognitive plasticity and reorganization when sound is introduced following a period of auditory deprivation early in development. Although it is common to consider deafness as affecting hearing alone, it may be the case that auditory deprivation leads to more global changes in neurocognitive function. In this paper, we investigate implicit sequence learning abilities in deaf children with CIs using a novel task that measured learning through improvement to immediate serial recall for statistically-consistent visual sequences. The results demonstrated two key findings. First, the deaf children with CIs showed disturbances in their visual sequence learning abilities relative to the typically-developing normal-hearing children. Second, sequence learning was significantly correlated with a standardized measure of language outcome in the CI children. These findings suggest that a period of auditory deprivation has secondary effects related to general sequencing deficits, and that disturbances in sequence learning may at least partially explain why some deaf children still struggle with language following cochlear implantation.
Keywords: implicit sequence learning, deafness, cochlear implants, language development
Deaf children with cochlear implants (CIs) provide a unique opportunity to study brain plasticity and neural reorganization. In some sense, this research effort can be thought of as the modern equivalent of the so-called “forbidden experiment” in the field of language development: it provides an ethical research opportunity to study the effects of the introduction of sound and spoken language on cognitive and linguistic development after a period of auditory deprivation. Whereas most previous work with this population has investigated the development of auditory perception, speech perception, and spoken language development, relatively few studies have examined more global learning and cognitive capabilities.
There is in fact some indication that a period of auditory deprivation occurring early in development may have secondary cognitive and neural ramifications in addition to the obvious hearing-related effects. Specifically, because sound by its very nature is a temporally-arrayed signal, a lack of experience with sound may affect how well one is able to encode, process, and learn serial patterns in any modality (Marschark, 2006; Rileigh & Odom, 1972; Todman & Seedhouse, 1994). Exposure to sound may provide a kind of “auditory scaffolding” in which a child gains vital experience and practice with learning and representing sequential patterns in the environment (). If true, then a lack of experience with sound may delay the development of domain-general processing skills that rely on the encoding and learning of temporal or sequential patterns, even for non-auditory input. In fact, previous findings do suggest that profound deafness may result in disturbances to non-auditory abilities related to processing temporal or serial order information (e.g., ; ; Pisoni & Cleary, 2004; Rileigh & Odom, 1972). Even so, the ability to learn complex, non-auditory sequential patterns has not been explored in children who are profoundly deaf.
Fundamental learning abilities related to acquiring complex probabilistic patterns – i.e., implicit, statistical, or sequential learning – have been argued to be important for cognitive development, especially in regards to successful language acquisition and processing (; ; ; ; ; ). At its most fundamental level, spoken language consists of a series of sounds (phonemes, syllables, words) occurring sequentially (Lashley, 1951). Language acquisition in part likely involves general learning mechanisms that are used to extract and process regularities in any complex sequential domain, be it linguistic or not. There are many published examples of infants (), children (), adults (), neural networks (Elman, 1990), and even nonhumans () demonstrating robust implicit sequence learning capabilities. These “existence proofs” have proven beyond a doubt that the human (and possibly non-human) organism, at least under typical developmental conditions, is equipped with relatively powerful, raw learning capabilities for acquiring complex, probabilistic sequential patterns.
In the case of profound deafness, although a CI provides the means to successfully develop age-appropriate speech and language abilities, it is well known that some children obtain little language benefit other than the awareness of sound from their implant (American Speech-Language-Hearing Association, 2004). Some of this variation in outcome has been shown to be due to certain demographic factors, such as age at implantation and length of deafness (; ). However, these demographic variables leave a large amount of variance unexplained. It is likely that intrinsic cognitive factors, especially fundamental learning and memory abilities, contribute to language outcomes following implantation (). Disturbances in implicit sequence learning specifically may hold the key to understanding the enormous range of variation in language development in this population (Pisoni, Conway, Kronenberger, Horn, Karpicke, & Henning, 2008).
In this paper, we explore these issues by examining implicit sequence learning in deaf children with CIs compared to an age-matched group of normal-hearing children. The aims are twofold: to assess the effects that a period of auditory deprivation and language delay may have on domain-general sequence learning skills; to investigate the possible role that sequence learning plays in language outcomes following cochlear implantation. Our hypothesis is that deaf children with CIs may show disturbances in visual implicit sequence learning as a result of their relative lack of experience with sequential (auditory) patterns early on in development. Furthermore, if implicit sequence learning is important for successful language acquisition, then we would expect that sequence learning performance will be associated with measures of language development, with better sequence learners showing the best language outcomes post-implantation.
Experiment
Two groups of children participated, deaf children with CIs, and an age-matched group of typically-developing, normal-hearing (NH) children. All children were tested on an implicit sequence learning task. We also collected a clinical measure of language outcome for the CI children. We reasoned that if language development is based in part on general, fundamental learning abilities, then it ought to be possible to observe empirical associations between performance on the implicit sequence learning task and a measure of language development. Several additional measures of memory and language were also collected from all participants in order to rule out alternative mediating variables – such as vocabulary knowledge or immediate memory span -- responsible for any observed correlations. Observing a correlation between the two tasks even after partialing out the common sources of variance associated with these other measures would provide converging support for the conclusion that implicit learning is directly associated with spoken language development, rather than being mediated by a third underlying factor.
The sequence learning task used here is based on the Milton Bradley game “Simon” and was developed to be used specifically with hearing-impaired children (see ; Pisoni & Cleary, 2004). In this task, visual color sequences are presented. After each sequence presentation, the child is asked to reproduce it by pressing the panels of a touch-sensitive screen. Unbeknownst to the participants, in the initial phase, all sequences are generated from an underlying artificial grammar that dictates the order in which particular colors can occur in the sequence (). Learning is assessed by the extent to which immediate serial recall improves for novel sequences having the same underlying structure (i.e., conforming to the artificial grammar) compared to novel sequences that are not consistent with the grammar. Several recent studies (; ; ; ) as well as a number of older, classic studies (; Reber, 1967) have looked at improvements to immediate serial recall as a measure of implicit learning. As argued by Redington and Chater (2002), this indirect method for measuring sequence learning is arguably superior to that typically used in most artificial grammar learning studies, explicit grammaticality judgments, which likely depend on metacognitive awareness. Especially considering the age of the participants, using an indirect measure of learning that does not depend on explicit or consciously-controlled strategies seems ideal.
Method
Participants
Deaf children with CIs
Twenty-five prelingually, profoundly deaf children were recruited through the DeVault Otologic Research Laboratory at the Indiana University School of Medicine, Indianapolis. Inclusion criteria included chronological age 5–10 years, onset of profound bilateral hearing loss (90-dB or greater) by age 2, had received a cochlear implant by age 4, had used their implant for a minimum of 3 years, and were native speakers of English. Except for two children with bilateral implants and one child who had a hearing aid in the non-implanted ear, all children had a single implant. For the three children with bilateral hearing, testing was conducted with only one CI activated (the original implant). Although several of the children had been exposed to Signed Exact English, none of the children relied exclusively on sign or gesture, and all children were tested using oral-only procedures. Aside from hearing loss, there were no other known cognitive, motor, or sensory impairments. For their time and effort, the children’s parents/caregivers received monetary compensation.
Normal-hearing children
Twenty-seven typically developing, NH children were recruited through Indiana University’s “Kid Information Database” and through the Life Education and Resource Home Schooling Network of Bloomington, IN. Inclusion criteria included chronological age 5–10 years and native speakers of English. Parental reports indicated no history of a hearing loss, speech impairment, or cognitive or motor disorder. For their participation, children received a small toy and their parents received monetary compensation.
Exclusion criteria
For both groups of participants, children’s data were excluded based on the following criteria. First, if any child refused to participate in portions of the tasks and/or displayed inattention or lack of motivation, their data was excluded. This criterion resulted in 2 of the CI children being excluded from subsequent analyses. Second, if any child’s performance on the primary experimental measure, the Simon learning game (described below), was more than 2 standard deviations from the group mean, their data was excluded. This criterion resulted in 1 of the NH children being excluded. In total, 2 CI children and 1 NH child were excluded, resulting in a total of 23 CI children and 26 NH children included in all analyses subsequently reported.
316 Serial Key Transfer Panels Kirk Franklin
Participant characteristics
Table 1 summarizes the demographic characteristics of the 23 CI children and 26 NH children included in all analyses. In addition to chronological age and two measures specific to the CI group (age at implantation and duration of CI use), five additional measures were collected in order to provide a comparison between the groups on both verbal and non-verbal abilities. These are described next.
Table 1
CI Childrena | NH Childrenb | |||||||
---|---|---|---|---|---|---|---|---|
Measure | M | SD | Range | M | SD | Range | t(47) | p |
Age | 90.1 | 19.9 | 61–118 | 87.8 | 11.8 | 65–104 | 0.5 | 0.60 |
Age Implant | 21.2 | 8.3 | 10–39 | -- | -- | -- | -- | -- |
CI Duration | 68.9 | 19.4 | 36–98 | -- | -- | -- | -- | -- |
FD | 4.9 | 1.6 | 2–8 | 7.0 | 1.9 | 5–12 | −4.3 | 0.0001 |
BD | 2.5 | 1.5 | 0–5 | 3.7 | 1.0 | 2–6 | −3.5 | 0.001 |
PPVT | 85.9 | 12.2 | 59–107 | 114.3 | 12.7 | 90–139 | −7.9 | 0.0001 |
CMS-T | 11.4 | 3.2 | 6–16 | 11.5 | 3.0 | 5–15 | −0.1 | 0.90 |
CMS-D | 10.7 | 2.9 | 5–15 | 11.6 | 2.0 | 5–14 | −1.2 | 0.22 |
Note. Age is given in months; Age Implant, age at cochlear implantation (in months); CI Duration, duration of cochlear implant use (in months); FD, forward digit span score; BD, backward digit span score; PPVT, Peabody Picture Vocabulary Test scaled score; CMS-T, Children’s Memory Scale (Total) scaled score; CMS-D, Children’s Memory Scale (Delayed test) scaled score.
bn=26
Verbal ability was assessed through the forward and backward digit span tasks of the WISC-III intelligence scale (Wechsler, 1991) as well as the Peabody Picture Vocabulary Test (PPVT) (3rd edition). In the digit span tasks, digits were played through loudspeakers and the child’s task was to repeat the digits back in correct order. Subjects received a point for each list that they correctly recalled in each digit span task. Generally, the forward digit span task is thought to reflect the involvement of processes that maintain and store verbal items in short-term memory for a brief period of time, whereas the backward digit span task reflects the operation of controlled attention and higher-level executive processes that manipulate and process the verbal items held in immediate memory (Rosen & Engle, 1997). The PPVT is a standard measure of vocabulary development (Dunn & Dunn, 1997). In this task, participants are shown four pictures on a single trial. They are prompted with a particular English word and then asked to pick the picture that most accurately depicts the word. For each child, a scaled score is derived based on comparison with a large normative sample.
Nonverbal ability was assessed through the “dot locations” subtest of the Children’s Memory Scale (CMS; Cohen, 1997). In this test of nonverbal visual-spatial learning and memory, the children were shown a picture of six blue dots inside a large white background. The dot pattern was presented to the child for five seconds before being taken out of sight. The child was then asked to reproduce the dot pattern from memory by placing six blue chips onto a 3×4 grid. This occurred a total of three times using the same dot pattern each trial. Next, a trial of red dots was presented and the child was asked to reproduce it. The red dot trial was not scored, but rather served as a distracter. The child was then asked to recall from memory the initial blue dot pattern that had been presented three times (“short delay” trial). At the conclusion of the experiment (after a delay of approximately 30 minutes), the child was asked once more to reproduce the blue dot pattern from memory (“long delay” trial). The child’s pattern reproductions were scored based on total number of chips placed correctly on the grid. As per standard procedure, the raw scores were converted into scaled scores, taking into account the age of the child, resulting in two scaled scores: visual-spatial learning “total” score (sum of scores on trials 1–3 plus the short delay trial); and visual-spatial “long delay” score (score on the long delay trial).
As Table 1 shows, the children were well-matched in regards to both chronological age and visual (nonverbal) memory abilities as assessed by the dot pattern subtest of the CMS. However, the NH children exceeded the CI children on their forward and backward digit spans and receptive vocabulary scores, a finding that is consistent with previous research using this population (; Pisoni & Cleary, 2004).
Apparatus
A Magic Touch® touch-sensitive monitor displayed the visual stimuli and recorded participant responses for the sequence learning task.
Stimulus Materials
For the sequence learning task, we used two artificial grammars to generate the stimuli (c.f., ). These grammars, depicted in Table 2, specify the probability of a particular element (color) occurring given the preceding element. For each stimulus sequence, the starting element (1–4) was randomly determined and then the listed probabilities were used to determine each subsequent element, until the desired length was reached. Grammar A was used to generate 16 unique sequences for the learning phase (6 of length 2 and 5 each of lengths 3 and 4) and 12 sequences for the test phase (4 each of lengths 3–5), hereafter referred to as the “grammatical test sequences”. Grammar B was used to generate 12 sequences for the test phase as well (4 each of lengths 3–5), hereafter referred to as the “ungrammatical test sequences”. All learning and test phase sequences are listed in Appendix A.
Table 2
Grammars A and B used in the Implicit Learning Task
Grammar A (n+1) | Grammar B (n+1) | |||||||
---|---|---|---|---|---|---|---|---|
Colors/locations (n) | 1 | 2 | 3 | 4 | 1 | 2 | 3 | 4 |
1 | 0.0 | 0.5 | 0.5 | 0.0 | 0.0 | 0.0 | 0.0 | 1.0 |
2 | 0.0 | 0.0 | 1.0 | 0.0 | 0.5 | 0.0 | 0.0 | 0.5 |
3 | 0.5 | 0.0 | 0.0 | 0.5 | 0.0 | 1.0 | 0.0 | 0.0 |
4 | 1.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.5 | 0.5 | 0.0 |
Note. Grammars show transition probabilities from position n of a sequence to position n+1 of a sequence for four colors labeled 1–4.
Appendix A
Learning and Test Sequences used in the Implicit Learning Task
Sequence Length | Learning Sequence | Test Sequence (A) | Test Sequence (B) |
---|---|---|---|
2 | 4-1 | ||
2 | 3-1 | ||
2 | 1-3 | ||
2 | 2-3 | ||
2 | 1-2 | ||
2 | 3-4 | ||
3 | 4-1-3 | 2-3-4 | 3-2-1 |
3 | 2-3-1 | 1-3-1 | 2-4-2 |
3 | 1-2-3 | 4-1-2 | 4-2-4 |
3 | 1-3-4 | 3-1-3 | 2-4-3 |
3 | 3-4-1 | ||
4 | 1-2-3-4 | 1-3-1-3 | 3-2-4-2 |
4 | 3-1-2-3 | 3-4-1-2 | 1-4-2-4 |
4 | 1-2-3-1 | 4-1-2-3 | 4-2-1-4 |
4 | 4-1-3-1 | 3-1-3-4 | 2-1-4-3 |
4 | 2-3-1-3 | ||
5 | 1-2-3-1-2 | 4-3-2-1-4 | |
5 | 4-1-3-4-1 | 1-4-3-2-4 | |
5 | 3-1-2-3-1 | 3-2-1-4-2 | |
5 | 1-2-3-4-1 | 4-2-4-2-4 |
Procedure
The deaf children with CI’s were tested by a trained Speech Language Pathologist at the Devault Otologic Research Laboratory, Department of Otolaryngology, Indiana University School of Medicine, Indianapolis. The NH children were tested in a sound-attenuated booth in the Speech Research Laboratory at Indiana University, Bloomington. All testing procedures were identical for both groups of children. This was accomplished by creating a written protocol manual specifying the procedures and language used for the experiments with both groups of children. Both experimenters followed this manual precisely. In addition, the two experimenters met regularly with the first author to discuss matters related to procedure as well as occasionally observed one another’s test sessions in order to keep testing procedures as close as possible for all children. For both groups of children, the study consisted of 10 tasks in a session lasting 60–90 minutes, with breaks provided as needed. However, data from only four of the tasks are reported here (sequence learning, digit spans, PPVT, and CMS).
Before beginning the experiment, all NH children received and passed a brief pure-tone audiometric screening assessment in both ears. Both groups of children were also given a brief color screening, which consisted of presenting four blocks to the children, each of a different color (blue, green, red, yellow), and asking them to point to each and name the color. This was done to ensure that the children could perceive and name each of the four colors used in the implicit learning task. All children passed this screening. Following the screenings, all children were given the sequence learning task, followed by the digit span tasks and the vocabulary test.
In addition, for the deaf children with CIs, we included a standardized clinical measure of language outcome. As part of the children’s regular visits to the Department of Otolaryngology, 17 of the 23 children were assessed on three core subtests of the Clinical Evaluation of Language Fundamentals, 4th Edition (CELF-4), an assessment tool for diagnosing language disorders in children (Semel, Wiig, & Secord, 2003). These three subtests measure aspects of general language ability: Concepts and Following Directions (C&FD), Formulated Sentences (FS), and Recalling Sentences (RS). A brief description of these three subtests is provided in Table 3 (see Paslawski, 2005, for a review and description of all subtests).
Table 3
Subtest | Description |
---|---|
C&FD | Measures auditory comprehension and recall of utterances of increasing length and complexity |
WS | Assesses morphology and pronoun use |
FS | The child is given a word or words and must generate spoken sentences in reference to a picture cue |
Note. C&FD, Concepts and Following Directions; FS, Formulated Sentences; RS, Recalling Sentences.
Sequence learning task
For the visual sequence learning task, participants were given the following instructions:
You are going to see some squares of different colors on this computer screen. The squares will flash on the screen in a pattern. Your job is to try to remember the pattern of colors that you see on the screen. After each pattern, you will see all four colors on the screen. You need to touch the colors in the same pattern that you just saw. For example, if you saw the pattern red-green-blue, you would touch the red square, then the green square, and then the blue square. Touch where it says ‘continue’ on the bottom of the screen when you’re finished. Always use your (preferred) hand to touch the screen and rest your arm on this gel pad.
After explaining the instructions to each child, the experimenter gave each child two practice trials to assess whether they understood the instructions. Only after successful demonstration that they understood the instructions did the experiment continue.
Unbeknownst to participants, the task actually consisted of two parts, a Learning Phase and a Test Phase. The procedures for both phases were identical and in fact from the perspective of the subject, there was no indication of separate phases at all. The only difference between the two phases was which sequences were used. In the Learning Phase, the 16 learning sequences were presented in three blocks: the 6 length-2 sequences first, then the 5 length-3 sequences, and finally the 5 length-4 sequences; within each block, sequences were presented in random order. After completing the sequence reproduction task for all of the learning sequences, the experiment seamlessly transitioned to the Test Phase, which used the 12 novel grammatical 12 novel ungrammatical test sequences. Test sequences were presented in three blocks: the 8 length-3 sequences first, the 8 length-4 sequences next, and finally the 8 length-5 sequences; within each block, sequences were presented in random order.
Sequence presentation consisted of colored squares appearing one at a time, in one of four possible positions on the touchscreen (upper left, upper right, lower left, lower right). The four elements (1–4) of each grammar were randomly mapped onto each of the four screen locations as well as four possible colors (red, blue, yellow, green). The assignment of stimulus element to position/color was randomly determined for each subject; however, for each subject, the mapping always remained consistent across all trials.
After a colored square appeared for 700 msec, the screen was blank for 500 msec, and then the next color of the sequence appeared. After the entire sequence had been presented, there was a 500 msec delay and then the four panels appeared on the touch screen that were the same-sized and same-colored as the four locations that were used to display each sequence. The subject’s task was to watch a sequence presentation and then to reproduce the sequence they saw by pressing the appropriate buttons in the correct order as dictated by the sequence. When they entered their response, they were instructed to press a “Continue” button at the bottom of the screen, and then the next sequence was presented after a 3-sec delay. A schematic of the sequence learning task is shown in Figure 1.
Depiction of the visual implicit learning task used in Experiments 1 and 2, similar to that used in previous work (; ). Participants view a sequence of colored squares (700-msec duration, 500-msec ISI) appearing on the computer screen (top) and then, 500-msec after sequence presentation, they must attempt to reproduce the sequence by pressing the touch-panels in correct order (bottom). The next sequence occurs 3000-msec following their response.
Participants were not told that there was an underlying grammar for any of the learning or test sequences, nor that there were two types of sequences in the Test Phase. From the standpoint of the participant, the sequence task was solely one of observing and then reproducing a series of visual sequences.
Results
In the sequence learning task, a sequence was scored correct if the participant reproduced each test sequence correctly in its entirety. For each group, separate accuracy (% correct) scores were computed for the Learning and Test phases. Because of the relatively short duration of the Learning Phase, accuracy scores for this phase are not expected to reflect grammar learning per se; rather, performance in this phase presumably reflects children’s ability to accurately reproduce visual sequences from immediate memory. On the other hand, as is typical in artificial grammar learning studies, the Test Phase has been constructed such that it will indeed provide a way to measure the children’s sequence learning abilities for the artificial grammar. This is achieved by comparing recall performance for novel grammatical test sequences relative to ungrammatical test sequences. To the extent that sequence learning has occurred, one would expect recall for the grammatical patterns to exceed those for the ungrammatical ones (see ; ; ).
The Learning Phase results revealed no differences between the two groups in the number of sequences correctly reproduced: 76.19% vs. 72.56% for the NH and CI groups, respectively, t(47) = .61, p=.55. This suggests that the CI children can accurately reproduce visual sequences from immediate memory just as well as the NH children. Importantly, it also provides a very nice control, because the equivalent Learning Phase performances for the two groups suggest that both groups of children understood the task instructions equally well, ruling out the possibility that any differences occurring in the Test Phase results are due to such confounds.
While the Learning Phase results indicate no overall difference in the ability to immediately recall and reproduce visual sequences, the Test Phase results on the other hand did reveal group differences. As shown in Figure 2, the NH group correctly reproduced a significantly greater number of grammatical compared to ungrammatical test sequences (59.0% vs. 53.2%; t(25) = 2.25, p<.05). This finding demonstrates that as a group the NH children showed better immediate serial recall for novel test sequences having the same statistical/sequential structure as the ones from the Learning Phase. On the other hand, the CI group did not show a difference in performance for the grammatical compared to ungrammatical test sequences (50.0% vs. 52.5%), t(22)=−.77, p=.50. Thus, on average, the NH group showed evidence of implicit sequence learning on this task, whereas the CI children essentially showed no learning.
Average grammatical sequence (black) and ungrammatical sequence (white) % correct scores for CI children (left) and NH children (right). Error bars represent +/− 1 standard error.
For each subject we also calculated a learning score (LRN), the difference in accuracy between the grammatical and ungrammatical test sequences. The LRN score reflects the extent that sequence memory spans improved for sequences derived from the same grammar as in the Learning Phase and therefore is a quantifiable measure of implicit sequence learning. Consistent with the above analyses, the NH group’s learning score (5.8%1) was significantly greater than the CI group’s score (−2.5%), t(47) = −2.01, p<.05.
In addition, Figure 3 shows a comparison of the distribution of individual LRN scores for each of the two groups of children (NH group on the top and CI group on the lower panel). Whereas about half (53.8%; 14/26) of the NH children showed a sequence learning score greater than 0, only about one third (34.7%; 8/23) of the CI children did.
Implicit learning scores for each individual participant in the NH (top) and CI groups (bottom), ranked order from lowest to highest.
Finally, we computed partial correlations between sequence learning and age at implantation and duration of implant use, while controlling for chronological age. Sequence learning was negatively correlated with the age in which the child received their implant (r=−.410, p=.058, 2-tailed) and positively correlated with the duration of implant use (r=.410, p=.058, 2-tailed)2. That is, the longer the child was deprived of auditory stimulation early in development, the lower their visual sequence learning scores (see Figure 4); correspondingly, the longer the child had experience with sound via the implant, the higher their sequence learning scores.
Scatter plot showing the association between implicit learning and the age at which each child received their implant.
Consistent with the hypothesis that a period of deafness (and/or language delay) may cause secondary difficulties with domain-general sequencing skills, the present results reveal that deaf children with CI’s display atypical visual implicit sequence learning abilities. Moreover, the partial correlations suggest that both the length of auditory deprivation and the amount of exposure to sound via a cochlear implant has secondary consequences not directly associated with hearing or language development per se. The amount of experience with sound, or lack thereof, appears to affect the ability to implicitly learn complex visual sequential patterns.
Implicit learning and language outcomes in deaf children with CIs
The question we next turn to is whether individual differences in sequence learning are associated with language outcomes in the CI group. We conducted bivariate correlations between the LRN score and the three subtest scaled scores of the CELF-4. Sequence learning was positively and significantly correlated with two of the subtests of the CELF-4: Formulated Sentences (r=.571, p<.05, 2-tailed, see Figure 5), and Recalling Sentences (r=.540, p< .05, 2-tailed). Although not significant, the correlation with the third subtest, Concepts and Following Directions, was also positive (r=.469, p=.058, 2-tailed). For the most part, these correlations remained significant even after controlling for the common variance associated with duration of CI use, forward digit span, backward digit span, and vocabulary scores (PPVT) (see Table 4). Especially robust was the correlation between sequence learning and the Formulated Sentences subtest.
Scatter plot showing the association between a standardized measure of language outcome (Formulated Sentences from the CELF-4) and implicit learning for the deaf children with CIs.
Iobit uninstaller 7 serial key. IObit Uninstaller 7 Serial Key & License Key. IObit Uninstaller 7 Serial Key is one of best software that design to help user regarding removing unnecessary files or software from system. If your system giving low performance or contain many unnecessary files then you must clean them. IObit Uninstaller 8 Pro Crack Full Version Serial Key 2018 IObit Uninstaller 8 Pro Crack is another version of this popular Windows applications uninstaller that will be published by IOBit and also released a beta edition of Uninstaller 7 now; motive enough to select the program to. With the new added Toolbars Uninstallation Module, IObit Uninstaller Serial key helps you remove unwanted toolbars thoroughly. Subscribe to Blog via Email Enter your email address to subscribe to this blog and receive notifications of new posts by email.
Table 4
Partial Correlations between Implicit Learning and Language Outcome Measures in Deaf Children with CIs
Controlling | C/FD | FS | RS |
---|---|---|---|
AgeImp | .434 | .484 | .464 |
UseLength | .531* | .613* | .553* |
FWdigit | .407 | .498* | .461 |
BWdigit | .428 | .550* | .518* |
PPVT | .480 | .729** | .646** |
**p<.01,
2-tailed. LRN, implicit learning score; AgeImp, age at implantation; UseLength, length of CI use; C/FD, Concepts and Following Directions scaled score on the CELF-4; FS, Formulated Sentences scaled score on the CELF-4; RS, Recalling Sentences scaled score on the CELF-4; FWdigit, forward digit span; BWdigit, backward digit span; PPVT, Peabody Picture Vocabulary Test (scaled score).
In contrast, immediate serial recall of the sequential patterns, as measured by accuracy for the sequences from the Learning Phase, was not correlated with language outcomes. None of the correlations between Learning Phase performance and the three standardized language outcome scores were significant: Concepts and Following Directions (r=−.163, p=.531), Formulated Sentences (r=−.223, p=.391), and Recalling Sentences (r=−.227, p=.38). In fact, not only were the correlations non-significant, but they were in the opposite direction as one would expect to see if immediate serial recall was contributing to language outcomes. These findings are consistent with the idea that sequence learning specifically, separate from mere serial recall, is a critical factor contributing to language outcome in this population. This point is discussed in full in the General Discussion section.
In summary, visual sequence learning abilities were found to be associated with a standardized measure of language outcome in deaf children who have received a cochlear implant. Because the CELF-4 consists of a standardized score that takes into account the age of each child, the observed correlations are not merely due to chronological age. More importantly, the observed correlations between sequence learning and language outcomes also do not appear to be mediated by short-term/working memory or vocabulary knowledge. The present results demonstrate that children who show the best performance on the sequence learning task also are the ones displaying the best language outcomes. Taken together with the findings that the CI group showed impaired sequence learning, these findings raise the intriguing possibility that individual differences in domain-general sequence learning are partially responsible for the large range of variation in language outcomes following implantation.
General Discussion
The goals of this study were to assess 1) the visual sequence learning abilities in deaf children with CIs; and 2) whether individual differences in sequence learning can account for some of the enormous range of variability in language outcomes following cochlear implantation. The results showed that as a group, the CI children performed significantly worse on the visual sequence learning task compared to the age-matched group of NH children. Furthermore, implicit sequence learning was found to be significantly correlated with a standardized measure of language development. We discuss both of these findings in turn.
Effects of Auditory Deprivation on Sequence Learning
The artificial grammars used here were created such that it is possible to isolate exactly what is learned (not a trivial endeavor when using more “traditional” finite-state grammars). The two grammars are completely orthogonal to one another based on pair-wise transitions. For instance, in Grammar A, the “1” can only be followed by a “2” or “3”, whereas in Grammar B, the “1” can only be followed by the “4”. This is true of every pair-wise transition. Therefore, presumably to learn the “grammar” requires learning these pair-wise transitions. The fact that the CI children as a group essentially show no learning is quite striking, given that learning pair-wise transitions would appear to be a relatively fundamental form of sequence learning (see also Pisoni & Cleary, 2004).
The group differences in visual sequence learning are consistent with the hypothesis that a period of auditory deprivation may have major secondary effects on brain and cognition that are not specific to hearing or the processing of sound by the auditory modality. There is existing evidence that auditory learning and plasticity is reduced, even after cochlear implantation, due to a reorganization of auditory cortex following a period of auditory deprivation (). However, the present set of findings suggests that non-auditory implicit sequence learning may also be impaired.
Sound is unique among sensory input in several important ways. Compared to vision and touch, sound appears to be more attention-demanding (), especially early in development (). Sound is also intrinsically a temporal and sequential signal, one in which time and serial order are of primary importance (Hirsh, 1967). Indeed, previous work in healthy typical-developing adults suggests that auditory processing of time and serial order is superior to the other senses. Auditory advantages have been found in tasks involving temporal processing (Sherrick & Cholewiak, 1986), rhythm perception (; ), immediate serial recall (; ), sequential pattern perception (), and implicit sequence learning (; Conway & Christiansen, 2009). These findings suggest an intimate link between auditory cognition and the processing of temporal and sequential relations in the environment. In addition, some previous work has suggested that the profoundly deaf (including those with and without CIs) show disturbances in (non-auditory) functions related to time and serial order, including: rhythm perception (Rileigh & Odom, 1972); attention to serially-presented stimuli (; ; Quittner, Smith, Osberger, Mitchell, & Katz, 1994); immediate serial recall (Marschark, 2006; Pisoni & Cleary, 2004; Todman & Seedhouse, 1994); motor sequencing (); and aspects of executive function and cognitive control (Hauser, Lukomski, & Hillman, 2008; Pisoni et al., 2008). Furthermore, the introduction of sound via a cochlear implant appears to progressively improve certain sequencing abilities over time ().
It is possible that experience with sound and auditory patterns, which are complex, serially-arrayed signals, provides a child vital experience with perceiving and learning sequential patterns. Under this view, a period of deafness early in development deprives a child with the essential experience of dealing with complex sequential auditory input, which, it would appear, affects their ability to deal with sequential patterns in other sense modalities as well. Once hearing is introduced via the CI, a child begins for the first time to gain experience with auditory sequential input. The positive correlation between length of CI use and sequence learning scores which we found – obtained even when chronological age was partialed out -- suggests that experience with sound via a CI improves one’s ability to learn complex non-auditory sequential patterns. Thus, it is possible that given enough exposure to sound via a CI, a deaf child’s sequence learning abilities will eventually improve to age-appropriate levels. Alternatively, it may be that there is a sensitive period that significantly limits the time period in which auditory input can provide a scaffolding for sequence learning skill; in such a case, cochlear implantation in older children would do little to improve sequence learning.
An examination of the three CI children who had bilateral hearing (although they were tested only with one implant activated) shows an additional intriguing finding: these children were among the best performers of the group on the sequence learning task. Whereas the average sequence learning score for the CI group was negative (−2.5%), all three of these children had positive learning scores (8.3%, 16.7%, 25%), performing very similar to the best NH children. Thus, a greater experience with sound (via bilateral hearing) may have an even more beneficial effect on the development of sequence learning skill.
An important and intriguing population to explore in the future are profoundly deaf children without cochlear implants who are users of a gestural language such as American Sign Language (ASL). Arguably, ASL also contains a rich source of temporal and sequential information and therefore its use may alleviate some of the sequence learning disturbances seen in the present sample of children. On the other hand, signed languages, compared to spoken languages, have relatively limited sequential contrasts and instead rely heavily on nonlinear and simultaneous spatial expressions to convey information (). As such, it could be expected that deaf users of sign language also will show difficulties with sequential processing. In the case of immediate serial recall, this does appear to be the case ().
From a neurobiological standpoint, it is known that lack of auditory stimulation results in a decrease of myelination and fewer projections out of auditory cortex () – which presumably includes connectivity to the frontal lobe. The frontal lobe, and specifically the prefrontal cortex as well as Broca’s area, are believed to play an essential role in learning, planning, and executing sequences of thoughts and actions (; 1995). It is therefore possible that the lack of auditory input early on in development, and corresponding reduction of auditory-frontal activity, fundamentally alters the neural organization of the frontal lobe and connections to other brain circuits (), impacting the development of sequencing functions regardless of input modality. An alternative possibility (though not necessarily mutually exclusive) is that language experience, rather than sound per se, may affect sequence learning skills, a possibility raised in the next section.
Sequence Learning and Language Development
The second primary finding of this study was that sequence learning performance was significantly correlated with language outcomes in the CI group. Based on previous work with healthy adults (; ), we hypothesized that visual sequence learning abilities would be associated with language outcomes in deaf children with CIs. In support of this hypothesis, we found that the CI children’s sequence learning scores were positively and significantly correlated with a standardized clinical measure of language outcome. These correlations remained significant even after partialing out the effects of auditory digit spans and general vocabulary knowledge. The present findings suggest a close coupling between the development of general (non-auditory) sequence learning skills and spoken language.
Importantly, whereas sequence learning performance was correlated with language outcomes, performance during the Learning Phase of the task was not correlated with the language outcome measures (p’s > .38). This has an important implication regarding the role of verbal mediation on this task. It could be argued that because the Simon learning game presumably involves verbal rehearsal, the reason for the observed correlations between sequence learning and language is merely that both are indices of verbal abilities. It is certainly plausible that when a participant sees a color pattern, they may be covertly rehearsing the color names in order to help them reproduce the sequence (e.g., “blue – red – yellow – blue”). Adult participants do appear to covertly verbalize the color patterns, which means this task could be fruitfully considered a visual/verbal learning task (see ). On the other hand, there is evidence that children within the age ranges of our participants do not spontaneously engage in verbal rehearsal strategies (Ornstein, Naus, & Liberty, 1975; Naus, Ornstein, & Aivano, 1977). Even if some children were engaging in verbal rehearsal strategies, the fact that Learning Phase performance was not correlated with language outcomes indicates that only (verbal) sequence learning is associated with language development, not (verbal) immediate serial recall. Thus, these findings are still a novel contribution because they show that sequence learning, above and beyond sequence memory, is coupled with aspects of language development. Furthermore, verbal sequence learning may be neurocognitively distinct from (nonverbal) visuospatial sequence learning (; ), with only the former being related to language acquisition.
There are at least three explanations for the observed correlations between sequence learning and language outcomes: 1) sequence learning abilities may causally contribute to language development; 2) sequence learning and language processing may develop on a similar timescale but are independent of one another; 3) or, differences in language skill may affect sequence learning abilities, rather than the other way around. Possibility #2 is unlikely to be correct, based on other studies that have also found a link between sequential learning and language. For instance, several studies have found that domain-general implicit learning abilities may be disturbed in children and adults with specific language impairment (; ; Tomblin, Mainela-Arnold, & Zhang, 2007) and dyslexia (; ; ). Thus there appears to be a close coupling between language competence and domain-general sequence learning abilities; when one of these two abilities is disturbed, the other appears to be as well. Therefore, sequence learning and language do not appear to develop independently of one another.
Alternatives #1 and #3 both appear viable, based on our data and on previous findings and theory. It has been argued previously that language learning and processing depend in part on domain-general sequence learning mechanisms (; ; ; ). For instance, the neurocognitive mechanisms underlying the processing of both language and music appear to be somewhat co-extensive (e.g., Patel, 2003), with Broca’s area possibly being a “supramodal” sequence processor, especially for complex hierarchical sequences (; Greenfield, 1991). Therefore, sequential processing mechanisms may be recruited for language acquisition. However, it may also be possible that experience with the complexities of language, especially grammatical relations, promotes better learning of complex patterns more generally, be they linguistic or not. This, to our knowledge is an unexplored yet intriguing possibility. Interestingly, the processing of grammatical relations in language appears to be heavily influenced by language experience, whereas the learning of lexical/semantic categories is much less so (). Therefore, if grammatical processing is heavily experience-dependent, and if grammar learning is at least partially based on general sequential processing mechanisms, then one’s experience with language (specifically grammar), could possibly affect domain-general sequence learning.
To help tease apart the direction of causality between implicit sequence learning and language development, a longitudinal design would be useful. Such a design could help determine if implicit learning abilities predict subsequent language abilities assessed several years later, or vice-versa. For instance, this approach has been used to show that particular perceptual and cognitive abilities measured early in infancy or childhood, such as speech perception or working memory, have a measurable effect on subsequent language processing abilities assessed later (; Bernhardt, Kemp, & Werker, 2007; Gathercole & Baddeley, 1989; ; ).
One additional interesting implication of the current findings is that sequence learning as measured here does not appear to be completely specific to the sensory modality of the input. If it were, then hearing status ought not to impact visual sequence learning, nor would there be a correlation between visual sequence learning and spoken (auditory) language comprehension. Previous work has suggested a modality-specific locus to aspects of implicit sequence learning (e.g., ; ). For example, most neuroimaging studies of implicit learning have revealed modality-specific brain regions directly related to the learning process itself (e.g., ; ).
On the other hand, it has been argued that implicit learning results in knowledge that is abstract or amodal in nature, independent of the physical qualities of the stimulus (Altmann, Dienes, & Goode, 1995; Reber, 1993). Although these findings appear at odds with one another, it may be the case that implicit learning involves both stimulus-specific and domain-general processes (). Under this view, implicit sequence learning likely involves multiple levels of learning including learning simple stimulus-response associations or modality-specific patterns (that recruit unimodal brain regions) as well as higher-order forms of learning that could be considered more abstract or domain-general (that recruit the prefrontal cortex, Broca’s area, or striatum). The manner in which both modality-specific and domain-general implicit learning processes interact have yet to be fully specified. One interesting avenue for future research would be to examine both visual and auditory sequence learning in hearing-impaired populations, with the expectation that while both types of learning may be disturbed (due to domain-general effects of auditory deprivation on sequencing skills), auditory learning would be worst (due to modality-specific effects).
Aside from their theoretical importance, from a clinical standpoint, the current findings with CI children are important because they suggest that individual differences in basic sequence learning abilities may provide a principled explanation for why some deaf children with CIs achieve near-typical levels of speech and language outcomes whereas other children do not. Several recent studies have been devoted to understanding the nature of the enormous variation in language outcome in deaf children who receive a CI (e.g., ; ; ; Knutson, 2006; Pisoni & Cleary, 2004). The current results are clinically important because they may provide both the prediction of audiological benefit from a CI and the formulation of new intervention programs that specifically target the development of implicit sequence learning skills in deaf children who are doing poorly with their CIs. In particular, interventions focused on the training of cognitive sequencing skills and executive functions (e.g., ; ) may provide benefits above and beyond the standard audiological-based treatment strategies.
Conclusion
In sum, the present findings suggest that a period of auditory deprivation early in development may negatively impact implicit sequence learning abilities, which has profound implications for understanding variation in neurocognitive development and plasticity in both normal-hearing and deaf populations. In addition, these results revealed a direct empirical link between visual implicit sequence learning and language outcome, suggesting that basic cognitive learning abilities related to encoding sequential structure --- independent of immediate serial recall abilities -- may be an important foundational aspect of language development. In line with other recent findings (e.g., ; ; Tomblin, Mainela-Arnold, & Zhang, 2007), we suggest it will be fruitful to investigate implicit sequence learning in other populations with language delays or cognitive disorders, such as children with specific language impairment or autism.
Acknowledgments
This project was supported by the following grants from the National Institute on Deafness and Other Communication Disorders: R03DC009485, T32DC00012, and R01DC00111.
We wish to thank Linda Smith and Char Wozniak for their help with accessing the Kid Information Database. We also wish to thank Luis Hernandez for invaluable technical assistance, and Lauren Grove for her help in manuscript preparation. Finally, we are especially indebted to all of the children and parents who took part in this study.
Footnotes
1It could be objected that an increase of 5.8% for grammatical vs. ungrammatical sequences is a trivial gain. However, the magnitude of learning in much of the artificial grammar learning literature is often within the 5–10% range. Especially considering the age range of the participants and the extremely short period of exposure to the grammatical patterns, a 5.8% learning gain score is not insubstantial.
2As pointed out by an anonymous reviewer, because each child’s chronological age is the sum of age at implantation and duration of implant use, these two variables are complementary and not free to vary. Thus, it is not surprising that the partial correlations of sequence learning with these two variables are identical (though opposite in sign).
References
- Altmann GTM. Statistical learning in infants. Proceedings of the National Academy of Sciences. 2002;99:15250–15251.[PMC free article] [PubMed] [Google Scholar]
- Altmann GTM, Dienes Z, Goode A. Modality independence of implicitly learned grammatical knowledge. Journal of Experimental Psychology: Learning, Memory, & Cognition. 1995;21:899–912.[Google Scholar]
- American Speech-Language-Hearing Association. Cochlear implants [Technical report] 2004. Available from www.asha.org/policy.
- Baddeley A, Gathercole S, Papagno C. The phonological loop as a language learning device. Psychological Review. 1998;105:158–173. [PubMed] [Google Scholar]
- Bernhardt BM, Kemp N, Werker JF. Early word-object associations and later language development. First Language. 2007;27:315–328.[Google Scholar]
- Botvinick MM. Effects of domain-specific knowledge on memory for serial order. Cognition. 2005;97:135–151. [PubMed] [Google Scholar]
- Boutla MT, Supalla T, Newport EL, Bavelier D. Short-term memory span: Insights from sign language. Nature Neuroscience. 2004;7:997–1002.[PMC free article] [PubMed] [Google Scholar]
- Cleary M, Pisoni DB, Geers AE. Some measures of verbal and spatial working memory in eight- and nin-year-old hearing-impaired children with cochlear implants. Ear & Hearing. 2001;22:395–411.[PMC free article] [PubMed] [Google Scholar]
- Cleeremans A, Destrebecqz A, Boyer M. Implicit learning: News from the front. Trends in Cognitive Sciences. 1998;2:406–416. [PubMed] [Google Scholar]
- Cohen MJ. Children’s memory scale. San Antonio, TX: Psychological Corporation; 1997. [Google Scholar]
- Conway CM, Bauernschmidt A, Huang SS, Pisoni DB. Implicit statistical learning in language processing: Word predictability is the key. Cognition. (in press) [PMC free article] [PubMed] [Google Scholar]
- Conway CM, Christiansen MH. Seeing and hearing in space and time: Effects of modality and presentation rate on implicit statistical learning. European Journal of Cognitive Psychology. 2009;21:561–580.[Google Scholar]
- Conway CM, Christiansen MH. Modality-constrained statistical learning of tactile, visual, and auditory sequences. Journal of Experimental Psychology. 2005;31:24–39. [PubMed] [Google Scholar]
- Conway CM, Christiansen MH. Statistical learning within and between modalities: Pitting abstract against stimulus-specific representations. Psychological Science. 2006;17:905–912. [PubMed] [Google Scholar]
- Conway CM, Karpicke J, Pisoni DB. Contribution of implicit sequence learning to spoken language processing: Some preliminary findings with hearing adults. Journal of Deaf Studies and Deaf Education. 2007;12:317–334.[PMC free article] [PubMed] [Google Scholar]
- Conway CM, Pisoni DB. Neurocognitive basis of implicit learning of sequential structure and its relation to language processing. Annals of the New York Academy of Sciences. 2008;1145:113–131.[PMC free article] [PubMed] [Google Scholar]
- Conway CM, Pisoni DB, Kronenberger WG. The importance of sound for cognitive sequencing abilities: The auditory scaffolding hypothesis. Current Directions in Psychological Science. 2009;18:275–279.[PMC free article] [PubMed] [Google Scholar]
- Dawson PW, Busby PA, McKay CM, Clark GM. Short-term auditory memory in children using cochlear implants and its relevance to receptive language. Journal of Speech, Language, and Hearing Research. 2002;45:789–801. [PubMed] [Google Scholar]
- Dunn LM, Dunn LM. Peabody picture vocabulary test. 3rd edition. Circle Pines, MN: American Guidance Service; 1997. [Google Scholar]
- Emmorey K, Allen JS, Bruss J, Schenker N, Damasio H. A morphometric analysis of auditory brain regions in congentially deaf adults. Proceedings of the National Academy of Sciences. 2003;100:10049–10054.[PMC free article] [PubMed] [Google Scholar]
- Evans JL, Saffran JR, Robe-Torres K. Statistical learning in children with specific language impairment. Journal of Speech, Language, and Hearing Research. 2009;52:321–335.[PMC free article] [PubMed] [Google Scholar]
- Forkstam C, Hagoort P, Fernandez G, Ingvar M, Petersson KM. Neural correlates of artificial syntactic structure classification. NeuroImage. 2006;32:956–967. [PubMed] [Google Scholar]
- Fuster J. The prefrontal cortex-- an update: Time is of the essence. Neuron. 2001;30:319–333. [PubMed] [Google Scholar]
- Fuster J. Temporal processing. In: Grafman J, Holyoak KJ, Boller F, editors. Structure and functions of the human prefrontal cortex. New York: New York Academy of Sciences; 1995. pp. 173–181. [Google Scholar]
- Gathercole SE, Baddeley AD. Evaluation of the role of phonological STM in the development of vocabulary in children: A longitudinal study. Journal of Memory and Language. 1989;28:200–213.[Google Scholar]
- Glenberg AM, Swanson NG. A temporal distinctiveness theory of recency and modality effects. Journal of Experimental Psychology: Learning, Memory, & Cognition. 1986;12:3–15. [PubMed] [Google Scholar]
- Goschke T, Friederici AD, Kotz SA, van Kampen A. Procedural learning in Broca’s aphasia: Dissociation between the implicit acquisition of spatio-motor and phoneme sequences. Journal of Cognitive Neuroscience. 2001;13:370–388. [PubMed] [Google Scholar]
- Greenfield PM. Language, tools and brain: The ontogeny and phylogeny of hierarchically organized sequential behavior. Behavioral and Brain Sciences. 1991;14:531–595.[Google Scholar]
- Handel S, Buffardi L. Using several modalities to perceive one temporal pattern. Quarterly Journal of Experimental Psychology. 1969;21:256–266. [PubMed] [Google Scholar]
- Hauser PC, Lukomski J, Hillman T. Development of deaf and hard of hearing students’ executive function. In: Marschark M, Hauser P, editors. Deaf cognition: Foundations and outcomes. New York: Oxford University Press; 2008. pp. 286–308. [Google Scholar]
- Hauser MD, Newport EL, Aslin RN. Segmentation of the speech stream in a non-human primate: Statistical learning in cotton-top tamarins. Cognition. 2000;75:1–12. [PubMed] [Google Scholar]
- Hirsh IJ. Information processing in input channels for speech and language: The significance of serial order of stimuli. In: Darley FL, editor. Brain mechanisms underlying speech and language. New York: Grune & Stratton; 1967. pp. 21–38. [Google Scholar]
- Horn DL, Davis RAO, Pisoni DB, Miyamoto RT. Development of visual attention skills in prelingually deaf children who use cochlear implants. Ear & Hearing. 2005;26:389–408.[PMC free article] [PubMed] [Google Scholar]
- Horn DL, Pisoni DB, Miyamoto RT. Divergence of fine and gross motor skills in prelingually deaf children: Implications for cochlear implantation. Laryngoscope. 2006;116:1500–1506.[PMC free article] [PubMed] [Google Scholar]
- Howard JH, Jr, Howard DV, Japikse KC, Eden GF. Dyslexics are impaired on implicit higher-order sequence learning, but not on implicit spatial context learning. Neuropsychologia. 2006;44:1131–1144. [PubMed] [Google Scholar]
- Jaeggi SM, Buschkuehl M, Jonides J, Perrig WJ. Improving fluid intelligence with training on working memory. Proceedings of the National Academy of Sciences. 2008;105:6829–6833.[PMC free article] [PubMed] [Google Scholar]
- Jamieson RK, Mewhort DJK. The influence of grammatical, local, and organizational redundancy on implicit learning: An analysis using information theory. Journal of Experimental Psychology: Learning, Memory, & Cognition. 2005;31:9–23. [PubMed] [Google Scholar]
- Karpicke JD, Pisoni DB. Using immediate memory span to measure implicit learning. Memory & Cognition. 2004;32(6):956–964.[PMC free article] [PubMed] [Google Scholar]
- Keele SW, Ivry R, Mayr U, Hazeltine E, Heuer H. The cognitive and neural architecture of sequence representation. Psychological Review. 2003;110:316–339. [PubMed] [Google Scholar]
- Kirk KI, Miyamoto RT, Lento CL, Ying E, O’Neil T, Fears B. Effects of age at implantation in young children. Annals of Otology, Rhinology, & Laryngology. 2002;189:69–73. [PubMed] [Google Scholar]
- Klingberg T, Fernell E, Olesen PJ, Johnson M, Gustafsson P, Dahlstrom K, Gillberg CG, Forssberg H, Westerberg H. Computerized training of working memory in children with ADHD - A randomized, controlled trial. Journal of the American Academy of Child & Adolescent Psychiatry. 2005;44:177–186. [PubMed] [Google Scholar]
- Knutson JF. Psychological aspects of cochlear implantation. In: Cooper H, Craddock L, editors. Cochlear implants: A practical guide. 2nd Ed. London: Whurr Publishers; 2006. pp. 151–178. [Google Scholar]
- Knutson JF, Hinrichs JV, Tyler RS, Gantz BJ, Schartz HA, Woodworth G. Psychological predictors of audiological outcomes of multichannel cochlear implants. Annals of Otology, Rhinology, & Laryngology. 1991;100:817–822. [PubMed] [Google Scholar]
- Kolers PA, Brewster JM. Rhythms and responses. Journal of Experimental Psychology. 1985;11:150–167. [PubMed] [Google Scholar]
- Kral A, Eggermont JJ. What’s to lose and what’s to learn: Development under auditory deprivation, cochlear implants and limits of cortical plasticity. Brain Research Reviews. 2007;56:259–269. [PubMed] [Google Scholar]
- Lashley KS. The problem of serial order in behavior. In: Jeffress LA, editor. Cerebral mechanisms in behavior. New York: Wiley; 1951. pp. 112–146. [Google Scholar]
- Lieberman MD, Chang GY, Chiao J, Bookheimer SY, Knowlton BJ. An event-related fMRI study of artificial grammar learning in a balanced chunk strength design. Journal of Cognitive Neuroscience. 2004;16:427–438. [PubMed] [Google Scholar]
- Marschark M. Intellectual functioning of deaf adults and children: Answers and questions. European Journal of Cognitive Psychology. 2006;18:70–89.[Google Scholar]
- Menghini D, Hagberg GE, Caltagirone C, Petrosini L, Vicari S. Implicit learning deficits in dyslexic adults: An fMRI study. Neuroimage. 2006;33:1218–1226. [PubMed] [Google Scholar]
- Meulemans T, Van der Linden M. Implicit sequence learning in children. Journal of Experimental Child Psychology. 1998;69:199–221. [PubMed] [Google Scholar]
- Miller GA. Free recall of redundant strings of letters. Journal of Experimental Psychology. 1958;56:485–491. [PubMed] [Google Scholar]
- Naus MJ, Ornstein PA, Aivano S. Developmental changes in memory: The effects of processing time and rehearsal instructions. Journal of Experimental Child Psychology. 1977;23:237–251.[Google Scholar]
- Neville HJ, Mills D, Lawson D. Fractionating language: Different neural subsystems with different sensitive periods. Cerebral Cortex. 1992;2:244–258. [PubMed] [Google Scholar]
- Newman R, Bernstein Ratner N, Jusczyk AM, Jusczyk PW, Dow KA. Infants’ early ability to segment the conversational speech signal predicts later language development: A retrospective analysis. Developmental Psychology. 2006;42:643–655. [PubMed] [Google Scholar]
- Ornstein PA, Naus MJ, Liberty C. Rehearsal and organizational processes in children’s memory. Child Development. 1975;46:818–830.[Google Scholar]
- Paslawski T. The Clinical Evaluation of Language Fundamentals, Fourth Edition (CELF-4): A review. Canadian Journal of School Psychology. 2005;20:129–134.[Google Scholar]
- Penney CG. Modality effects and the structure of short-term verbal memory. Memory & Cognition. 1989;17:398–422. [PubMed] [Google Scholar]
- Pisoni D. Cognitive factors and cochlear implants: Some thoughts on perception, learning, and memory in speech perception. Ear and Hearing. 2000;21:70–78.[PMC free article] [PubMed] [Google Scholar]
- Pisoni DB, Cleary M. Learning, memory, and cognitive processes in deaf children following cochlear implantation. In: Zeng FG, Popper AN, Fay RR, editors. Springer handbook of auditory research: Auditory prosthesis, SHAR Volume X. 2004. pp. 377–426. [Google Scholar]
- Pisoni DB, Conway CM, Kronenberger WG, Horn DL, Karpicke J, Henning S. Efficacy and effectiveness of cochlear implants in deaf children. In: Marschark M, Hauser P, editors. Deaf cognition: Foundations and outcomes. New York: Oxford University Press; 2008. pp. 52–101. [Google Scholar]
- Plante E, Gomez R, Gerken L. Sensitivity to word order cues by normal and language/learning disabled adults. Journal of Communication Disorders. 2002;35:453–462. [PubMed] [Google Scholar]
- Posner MI, Nissen MJ, Klein RM. Visual dominance: An information-processing accounts of its origins and significance. Psychological Review. 1976;83:157–171. [PubMed] [Google Scholar]
- Quittner AL, Smith LB, Osberger MJ, Mitchell TV, Katz DB. The impact of audition on the development of visual attention. Psychological Science. 1994;5:347–353.[Google Scholar]
- Reber AS. Implicit learning of artificial grammars. Journal of Verbal Learning and Verbal Behavior. 1967;6:855–863.[Google Scholar]
- Reber AS. Implicit learning and tacit knowledge: An essay on the cognitive unconscious. Oxford, England: Oxford University Press; 1993. [Google Scholar]
- Redington M, Chater N. Knowledge representation and transfer in artificial grammar learning (AGL) In: French RM, Cleeremans A, editors. Implicit learning and consciousness: An empirical, philosophical, and computational consensus in the making. Hove, East Sussex: Psychology Press; 2002. pp. 121–143. [Google Scholar]
- Repp BH, Penel A. Auditory dominance in temporal processing: New evidence from synchronization with simultaneous visual and auditory sequences. Human Perception & Performance. 2002;28:1085–1099. [PubMed] [Google Scholar]
- Rileigh KK, Odom PB. Perception of rhythm by subjects with normal and deficient hearing. Developmental Psychology. 1972;7:54–61.[Google Scholar]
- Robinson CW, Sloutsky VM. Auditory dominance and its change in the course of development. Child Development. 2004;75:1387–1401. [PubMed] [Google Scholar]
- Rosen VM, Engle RW. Forward and backward serial recall. Intelligence. 1997;25:37–47.[Google Scholar]
- Saffran JR, Aslin RN, Newport EL. Statistical learning by 8-month-old infants. Science. 1996;274:1926–1928. [PubMed] [Google Scholar]
- Saffran JR, Senghas A, Trueswell JC. The acquisition of language by children. Proceedings of the National Academy of Sciences. 2001;98:12874–12875.[PMC free article] [PubMed] [Google Scholar]
- Semel E, Wiig EH, Secord WA. Clinical evaluation of language fundamentals, fourth edition (CELF-4) Toronto, Canada: The Psychological Corporation/A Harcourt Assessment Company; 2003. [Google Scholar]
- Sherrick CE, Cholewiak RW. Cutaneous sensitivity. In: Boff KR, Kaufman L, Thomas JP, editors. Handbook of perception and human performance, Vol. I: Sensory processes and perception. New York: Wiley; 1986. pp. 1–12. [Google Scholar]
- Skosnik PD, Mirza F, Gitelman DR, Parrish TB, Mesulam M-M, Reber PJ. Neural correlates of artificial grammar learning. NeuroImage. 2002;17:1306–1314. [PubMed] [Google Scholar]
- Todman J, Seedhouse E. Visual-action code processing by deaf and hearing children. Language & Cognitive Processes. 1994;9:129–141.[Google Scholar]
- Tomblin JB, Barker BA, Hubbs S. Developmental constraints on language development in children with cochlear implants. International Journal of Audiology. 2007;46:512–523. [PubMed] [Google Scholar]
- Tomblin JB, Mainela-Arnold E, Zhang X. Procedural learning in adolescents with and without specific language impairment. Language Learning and Development. 2007;3:269–293.[Google Scholar]
- Tsao F-M, Liu H-M, Kuhl PK. Speech perception in infancy predicts language development in the second year of life: A longitudinal study. Child Development. 2004;75:1067–1084. [PubMed] [Google Scholar]
- Ullman MT. Contributions of memory circuits to language: The declarative/procedural model. Cognition. 2004;92:231–270. [PubMed] [Google Scholar]
- Vicari S, Marotta L, Menghini D, Molinari M, Petrosini L. Implicit learning deficit in children with developmental dyslexia. Neuropsychologia. 2003;41:108–114. [PubMed] [Google Scholar]
- Wechsler D. Wechsler intelligence scale for children. Third edition. San Antonio, TX: The Psychological Corporation; 1991. [Google Scholar]
- Wilson M, Emmorey K. Working memory for sign language: A window into the architecture of the working memory system. Journal of Deaf Studies and Deaf Education. 1997;2:121–130. [PubMed] [Google Scholar]
- Wolff AB, Thatcher RW. Cortical reorganization in deaf children. Journal of Clinical and Experimental Neuropsychology. 1990;12:209–221. [PubMed] [Google Scholar]
Abstract
Increasingly, patients are receiving treatment at facilities other than hospitals, including long-term—health care facilities, assisted-living environments, rehabilitation facilities, and dialysis centers. As with hospital environments, nonhospital settings present their own unique risks of pneumonia. Traditionally, pneumonia in these facilities has been categorized as community-acquired pneumonia (CAP). However, the new designation for pneumonias acquired in these settings is health care—associated pneumonia (HCAP), which covers pneumonias acquired in health care environments outside of the traditional hospital setting and excludes hospital-acquired pneumonia (HAP), ventilator-associated pneumonia (VAP), and CAP. Although HCAP is currently treated with the same protocols as CAP, recent evidence indicates that HCAP differs from CAP with respect to pathogens and prognosis and, in fact, more closely resembles HAP and VAP. The HCAP Summit convened national infectious disease opinion leaders for the purpose of analyzing current literature, clinical trial data, diagnostic considerations, therapeutic options, and treatment guidelines related to HCAP. After an in-depth analysis of these areas, the infectious disease investigators participating in the summit were surveyed with regard to 10 clinical practice statements. The results were then compared with results of the same survey as completed by 744 Infectious Diseases Society of America members. The similarities and differences between those survey results are the basis of this publication.
Pneumonia is one of the most common infections requiring hospitalization. Changes in the location and manner in which health care is currently administered have resulted in the need to reassess the classification scheme employed for pneumonia. This is most evident when dealing with the increasing numbers of ambulatory and nonhospitalized individuals who are in regular contact with the health care system [1, 2]. Currently accepted classifications of pneumonia include community-acquired pneumonia (CAP), hospital-acquired pneumonia (HAP), ventilator-associated pneumonia (VAP), and nursing home—associated pneumonia (NHAP). The designation of health care—associated pneumonia (HCAP) was recently introduced to include an already-ill population of nursing-home residents, patients in long-term care, patients undergoing same-day procedures, patients receiving home- or hospital-based intravenous therapy, and patients undergoing dialysis [3]. The patient population at risk for HCAP is large and diverse, probably making up the largest single category of patients with pneumonia [2, 4, 5]. In general, patients who develop HCAP are more similar to hospitalized patients than to true community patients, in that they have a greater burden of comorbidities, including cancer, chronic kidney disease, heart disease, chronic obstructive lung disease, immunosuppression, dementia, and impaired mobility [2, 6, 7].
An important distinction of HCAP is that the pathogens are often multidrug-resistant (MDR) bacteria [2]. Therefore, the initial treatment of HCAP should be similar to that of HAP and VAP, which also differentiates it from CAP [3]. This is particularly important for clinicians working in first-response areas, such as emergency departments (EDs), to recognize, so that appropriate initial antimicrobial therapy is not delayed. Several studies have demonstrated that delaying the delivery of pathogen-appropriate antimicrobial therapy to patients with CAP and VAP results in excess mortality [8–10]. Thus, it is essential for physicians working in the ED to distinguish between HCAP and CAP, in order to correctly assess and manage suspected cases of pneumonia. This approach to HCAP also applies to other health care—associated infections in which the pathogens are more similar to hospital-acquired organisms than to community-acquired ones [6, 11].
This supplement to Clinical Infectious Diseases represents the proceedings of a panel of investigators whose goal was to assess the quality of evidence in support of the clinical classification of HCAP as a distinct entity and the need for specific therapeutic interventions for HCAP. Ten clinical practice statements were drafted by the chair (M.H.K.) and the 2 workshop leaders (L.E.M. and R.P.B.) and were subsequently evaluated by the 11-member panel made up of leaders in infectious diseases, pulmonary and critical care medicine, and pharmacology (table 1). Before the summit was convened, each participant was assigned a statement and instructed to systematically review and summarize the evidence supporting or refuting that statement.
Health Care—associated Pneumonia (HCAP) Summit clinical practice statements.
Health Care—associated Pneumonia (HCAP) Summit clinical practice statements.
In the first phase of the live meeting, the simultaneously conducted workshops “Defining HCAP” and “Therapeutic Intervention” included a leader and 4–5 content experts and served as a forum for each individual to present the evidence for his or her statement. When the data were presented, primary attention was given to the study methodology, the number of patients enrolled, and the outcome events. After the presentation of data for each statement, workshop members discussed the evidence, graded the strength of the evidence, and assigned the statement a consensus numeric grade for the “Nature of the Evidence” and the “Grade of Recommendation” (table 2). In the second phase of the live meeting, all summit panelists reconvened, reviewed the workshop summaries, and discussed each statement further. After each discussion, all participants voted on their individual levels of support, using the grading scheme shown in table 2.
Workshop and Health Care—associated Pneumonia (HCAP) Summit panel voting schemes.
Workshop and Health Care—associated Pneumonia (HCAP) Summit panel voting schemes.
In addition to defining the level of evidence in support of each statement, the panel members also outlined additional data required to further refine the statements for future clinical use. The main intention of this meeting was to provide a framework for future discussion and research in the area of HCAP.
Before the summit meeting, clinical perspectives of practicing physicians were measured via a Web-based survey. E-mail polling was done to ascertain their level of support for the same 10 statements. The e-mail invitation to participate in the electronic survey was sent to 3200 members of the Infectious Diseases Society of America (IDSA) (all active e-mail addresses). of the IDSA members surveyed, 383 (11.9%) responded. The purpose of the electronic surveys was to provide information that would allow for the comparison of>Statement 1: The Patient in/from a Health Care—associated, Nonhospital Environment Who Develops a Clinical Presentation of Pneumonia Has HCAP
Rationale and Definition of Statement
In defining HCAP as a distinct clinical entity, the most recent ATS-IDSA nosocomial pneumonia guidelines defined a subset of patients at risk for harboring resistant organisms despite their residence in the community [3]. Criteria included hospitalization in an acute-care facility for 2 days within 90 days before the infection; residence in a nursing home or long-term-care facility; recent receipt of intravenous antibiotic therapy, chemotherapy, or wound care, within 30 days before the infection; or attending a hospital or hemodialysis clinic [3]. Although the ATS-IDSA guidelines were intended to apply only to hospitalized patients with HCAP, it is apparent that these concepts are being extrapolated to nonhospitalized patients with HCAP as well [12]. By definition, the ATS-IDSA guidelines apply to patients coming to an acute-care facility from a nonhospital environment, whether the patient is seen in an outpatient facility or ED or is admitted directly to the hospital. However, there is a question regarding whether the guidelines for such patients should also apply to patients who remain in a nonhospital environment, such as a nursing home or long-term-care facility, or who remain in another setting but who meet the other ATS-IDSA criteria for having HCAP [3].
Methods
A search of the OVID “1996-present” database to identify studies related to descriptions of HCAP was completed on 1 November 2006. The search of the combined term “health care associated or healthcare-associated” produced a total of 144 articles. Next, the text word search for the term “HCAP” yielded 66 articles, and the text word search for “healthcare-associated pneumonia” resulted in 7 articles. At this point, the 2 text word searches were combined with the first combined-term search, and the results were limited to the English language. This produced 82 articles. The same search strategy was then used in the OVID database “in process,” and an additional 10 articles were identified. By scanning the titles of these 92 articles, 14 relevant articles were noted, and a review of the references for these 14 articles added 4 articles to the total. Thus, 18 articles were considered to be relevant to this statement.
Evidence
Definitions of HCAP. Several definitions of HCAP are stated or implied in the medical literature. One prevalent use of the term, considered to be irrelevant to this discussion, is the use of “health care associated” as a replacement for or synonym of “nosocomial” or “hospital associated” [13]. A second use of the term is “hospitalized with community-acquired pneumonia,” which also fails to capture the concept presented in the ATS-IDSA guidelines [14]. The relevant concept is that pneumonia not acquired in an acute-care hospital (traditionally labeled as “community-acquired infection”) is more likely to have a spectrum of pathogens that resemble those associated with HAP or VAP than to have a distribution of microbes traditionally associated with CAP.
HCAP as a distinct entity. Differences in the likely prevalence of drug-resistant pathogens were highlighted in a study by Friedman et al. [6] of health care—associated bloodstream infections in adults, which defined patients with health care—associated bacteremia in a fashion similar to the ATS-IDSA guidelines. Patients with health care—associated bacteremia were similar to those with hospital-acquired bloodstream infections, with regard to frequency of comorbid conditions, pathogens and their susceptibility, and mortality rates. The authors concluded that “a separate category for [health care—associated] infections is justified, and this new category will have obvious implications for choices about empirical therapy and infection-control surveillance” [6, p. 791]. This theme was echoed and expanded to include pneumonia in a 2004 editorial by Craven, who stated, “compared with patients with community-acquired pneumonia (CAP), those with HCAP are often at greater risk for colonization and infection with a wider spectrum of multidrug-resistant organisms” [4, p. 153]. However, Grossman et al., in a review of practice guidelines for treatment of lower-respiratory-tract infections in hospitalized patients, concluded that HCAP is treated similarly to HAP “and may be considered with HAP” [15, p. 295].
Kollef et al. [2] reviewed a large database of 4543 patients with culture-positive pneumonia and identified 20% as having HCAP. The percentage of patients with a culture positive for Staphylococcus aureus was similar among those with HCAP (46.7%) and those with HAP (47.1%), and mean mortality rates (19.8% and 18.8%, respectively) were similar for these 2 groups of patients as well. The mean length of stay for patients with HCAP was intermediate between that for patients with CAP and that for patients with HAP. The authors concluded that this analysis “justified HCAP as a new category of pneumonia” [2, p. 3854].
Pop-Vicas et al. [16] noted an increasing prevalence of MDR, gram-negative bacilli recovered at admission to a tertiary-care hospital. Factors independently associated with the isolation of resistant organisms (age 65 years, prior antibiotic therapy for 2 weeks, and residence in a long-term-care facility) were similar to those used to define HCAP.
The editorial response to the ATS-IDSA definition of HCAP has been mixed. Hiramatsu and Niederman [1, p. 3786] note that “publication of these recommendations has been recognized by the Centers for Medicare and Medicaid Services in their application of core measures for the treatment of CAP.” This action has led to the exclusion of patients with HCAP from studies of adherence to antibiotic therapy recommendations for patients with CAP. Fujitani et al. [17, pp. 627 and 630] considered the classification of HCAP as a separate disease entity to be “a good idea” but noted “problems with its execution.” They noted that the definitions of HAP, CAP, and HCAP have varied among different large-scale studies and suggest that “classification schemes are inherently imprecise because patient groups overlap in the HCAP categories.” Wunderink [18, p. 2686] also noted that “a distinction between HCAP and CAP has never been totally clear.” He concluded, however, by stating that “despite these issues, defining the HCAP category has led to more appropriate antibiotic therapy for the majority of patients and clearly assisted decision making” [18].
Differences in application of the definition by setting. The previous studies dealt with patients seen in, or admitted to, acute-care hospitals. However, many patients with pneumonia acquired in a nursing home setting are not transferred to an acute-care hospital. For example, Loeb et al. [19] conducted a cluster-randomized controlled trial of different regimens for treatment of pneumonia in nursing home residents in Canada. Only 110 of 661 evaluated patients were hospitalized.
The degree to which the HCAP definition applies to patients with pneumonia who remain in nonhospital health care settings, such as nursing homes, is not clear. Mortality rates for NHAP are higher than those for CAP [20, 21], but controlling for different factors that affect this risk is difficult. For example, in a review by El Sohl et al. [22] of 88 patients with culture-confirmed cases of severe pneumonia, previous use of antibiotics (a component of the ATS-IDSA definition) was found to be predictive of the presence of drug-resistant bacteria. However, the other predictor of drug resistance in this study was a lower Activities of Daily Living (ADL) score, a feature not considered in the ATS-IDSA definition of HCAP. Likewise, at least 1 study suggests that the risk of MDR bacteria being present is no higher for NHAP than for CAP [20]. It was acknowledged that the literature on NHAP is dated and incomplete; this area needs further investigation.
Grading of Evidence
On the basis of a review of the studies cited above, the 5 members of this workshop agreed that the evidence available to support this statement was category III (a mixture of the 2 following votes) for the statement in general, category IV (primarily from definition) for the statement as it applies to hospitalized patients with HCAP, and category V (insufficient evidence) for the statement as it applies to nonhospitalized patients with HCAP (table 2).
Level of Support
When voting on the support for this statement in the group at large, 55% of the summit participants accepted the statement with some reservations, 27% accepted the statement with major reservations, and 9% rejected the statement completely. In comparison, of the 383 IDSA members who participated in the online survey, 48% accepted the statement completely, 42% accepted the statement with some reservations, 4% accepted the statement with major reservations, 4% rejected the statement with reservations, and 2% rejected the statement completely (figure 1). It was thought by the summit participants that IDSA members were considering the statement primarily for its intended purpose of applicability to patients with HCAP seen in the acute-care hospital setting.
Voting comparison for statement 1 (“The patient in/from a health care—associated, nonhospital environment who develops a clinical presentation of pneumonia has HCAP”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. HCAP, health care—associated pneumonia.
Voting comparison for statement 1 (“The patient in/from a health care—associated, nonhospital environment who develops a clinical presentation of pneumonia has HCAP”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. HCAP, health care—associated pneumonia.
Discussion
By definition, the ATS-IDSA guidelines apply to patients coming to an acute-care facility from a nonhospital environment, whether the patient is seen in an outpatient facility or ED or is admitted directly to the hospital, and the different panels supported this definition. The concept is presented graphically in figure 2, which represents CAP, HCAP, and HAP as separate entities for which, in general, the likelihood of the presence of MDR organisms increases [23]. Less well documented, but still likely, is that rates of mortality and morbidity also increase as one considers the entities on the right side of the figure. There is, however, overlap between the 3 defined entities; for example, patients with severe CAP might have higher mean mortality rates than do those with HCAP. Similarly, the prevalence of MDR organisms may be high for patients with CAP in some areas where selective pressure due to antimicrobial use is high. It is important to appreciate that these 2 features (severity and prevalence of MDR) are not necessarily linked—one may be high while the other is not.
Relationship of health care—associated pneumonia (HCAP) to community-acquired pneumonia (CAP), hospital-acquired pneumonia (HAP), and ventilator-associated pneumonia (VAP). Note also the increased risk for colonization and infection with multidrug-resistant (MDR) pathogens, morbidity, and mortality in these groups. Adapted in part from [23].
Relationship of health care—associated pneumonia (HCAP) to community-acquired pneumonia (CAP), hospital-acquired pneumonia (HAP), and ventilator-associated pneumonia (VAP). Note also the increased risk for colonization and infection with multidrug-resistant (MDR) pathogens, morbidity, and mortality in these groups. Adapted in part from [23].
There is, however, insufficient evidence to decide the validity of the HCAP category as it relates to patients remaining in nursing homes or other non—acute-care health settings, as is reflected in the summit participants' grading of the evidence. The value of extrapolating the ATS-IDSA definition to these settings requires further study. Likewise, for other non—acute-care health settings, including those of specific subsets of patients with HCAP (those receiving hemodialysis, home infusion, wound care, chemotherapy, or recent antibiotics or who have a relative with resistant pathogens), the available data are very limited.
Future Directions
The paucity of data highlighted in the previous section provides valuable opportunities to determine the relevance of HCAP for the special populations noted and to continue to explore the validity of specific criteria used to define the entity. of particular note is the observation by Fujitani et al. [17, p. 630], who suggested that a more precise classification to minimize such overlap would allow easier comparison among studies of this entity “so a rigorous database can be accumulated for future investigations.”
Statement 2: The Clinical and Microbiological Features of HCAP Are More Similar to HAP and VAP Than to CAP
Rationale and Definition of Statement
Historically, patients with HCAP have been treated with antibiotic regimens recommended by CAP guidelines. As the prevalence of antimicrobial resistance has increased, particularly in patients whose conditions meet HCAP criteria, many clinicians have questioned whether these antibiotic regimens are appropriate. This is of particular importance, given the association between inadequate empirical antibiotic therapy and increased mortality rates.
To address these concerns, the ATS-IDSA guidelines recommended broad empirical antibiotic therapy followed by culture-guided de-escalation for patients with HCAP. Although these concepts were intuitive and widely embraced by clinicians, critics expressed concerns regarding the data supporting such significant changes to the guidelines. Because the guidelines cite only 7 references relating to HCAP, this section aims to assess the strength of evidence supporting the assertion that clinical and microbiological features of HCAP are more similar to those of HAP than to those of CAP.
Methods
HCAP. A PubMed database search to identify studies related to the clinical and microbiological features of HCAP was completed on 24 October 2006. The search terms “health care associated pneumonia,” “healthcare associated pneumonia,” “health care-associated pneumonia,” and “healthcare-associated pneumonia” gave a total of 48,465 articles. The search terms “microbiology” and “pathogen” yielded 551,438 articles. Combining the HCAP search with the microbiology search by use of the “AND” function produced 138 articles. After limiting these to the English language, 129 articles were reviewed. Only 3 were deemed relevant to the statement. Because HCAP includes multiple entities previously referred to by use of other terminology, several additional searches were conducted.
NHAP. The search terms “pneumonia” and “lower respiratory tract infection” were combined with the “OR” function to identify 86,173 articles. The search terms “nursing home” and “long term care” were combined with the “OR” function to identify 44,206 articles. The nursing home and pneumonia searches were combined using the “AND” function to give 616 articles. When these were combined with the previous microbiology search and limited to the English language, 112 articles were reviewed. Only 5 were deemed relevant to the statement. One article was then added from these articles' references, and another recent abstract also was included.
Hemodialysis-associated pneumonia. The search terms “hemodialysis” and “dialysis” were combined with the “OR” function to identify 111,259 articles. When these were combined with the prior pneumonia and microbiology searches and limited to the English language, 54 articles were reviewed. One article was deemed relevant to the statement.
Home care— and wound care—associated pneumonia. The search terms “home care” and “wound care” were combined with the “OR” function for a total of 49,986 articles. These were combined with the prior pneumonia and microbiology searches and limited to the English language, yielding 134 articles for review. One article was deemed relevant to the statement.
Chemotherapy-associated pneumonia. The search term “cancer chemotherapy” yielded 311,108 articles. These were combined with the prior pneumonia and microbiology searches to give 348 articles. After limiting these to the English language, 268 articles were reviewed, and 4 were deemed relevant.
Queries using multiple search terms relating to pneumonia in patients who have a family member with resistant pathogens and/or patients who have received prior antibiotic therapy did not result in any articles being found.
Evidence
HCAP. Only 1 study was identified that specifically focused on the microbiology of HCAP [2]. In this retrospective cohort analysis of a multi-institutional database, 4543 cases of culture-positive pneumonia were identified by International Classification of Diseases, Ninth Revision codes. Patients were then stratified as having CAP (49%), HCAP (22%), HAP (18%), or VAP (11%). The most common pathogens in HCAP were methicillin-resistant S. aureus (MRSA) and Pseudomonas aeruginosa (26.5% and 25.3%, respectively), similar to HAP (22.9% and 18.4%; P<.01 for P. aeruginosa). Conversely, Streptococcus pneumoniae and Haemophilus species were seen more frequently in CAP (16.6% and 16.6%) than in HCAP (3.1% and 5.8%; P<.01 for both). The mortality rates among patients with HCAP and patients with HAP were similar (19.8% and 18.8%, respectively; P>.05 ) and were significantly higher than that among patients with CAP (10.0%; P<.0001 ). Limitations of this study included the use of data only for hospitalized patients, inclusion of only patients with early-onset pneumonia, misclassification bias, and exclusion of patients with negative culture results. Also unresolved are the very high rates of Pseudomonas species (17.1%) and methicillin-susceptible S. aureus (17.2%) as the causative pathogens in patients with CAP. The clinical features of HCAP were not assessed in this study.
Two additional studies assessed the risk factors for colonization with resistant organisms in hospitalized patients, many of whom did not have pneumonia. The first of these studies assessed variables associated with MDR gram-negative bacillus carriage [16]. In this prospective, case-control study, it was found that predictors of MDR gram-negative bacillus colonization included several subsets of patients also included in the new ATS-IDSA definition of HCAP: residents of long-term-care facilities, patients undergoing hemodialysis, and patients who have recently received antibiotic therapy. Similarly, a prospective surveillance study of MRSA showed that patients whose conditions met HCAP criteria (patients who have recently been hospitalized, residents of long-term-care facilities, patients undergoing dialysis, or patients receiving home nursing care) accounted for 99% of “community-acquired” MRSA cases [24].
NHAP. Seven studies of the clinical and microbiological features of NHAP were identified. A prospective case-control study comparing patients with NHAP and age-matched patients with CAP showed that the presentation of pneumonia was strikingly different between groups. Compared with patients with CAP, patients with NHAP were less likely to have a productive cough (61% vs. 35%; P<.05 ), chills (58% vs. 24%; P<.05 ), headache (32% vs. 5%; P<.05 ), sore throat (19% vs. 7%; P<.05 ), myalgia (33% vs. 7%; P<.05 ), or arthralgia (10% vs. 0%; P<.05 ). Although the difference was reported as being statistically nonsignificant, patients with NHAP were almost twice as likely to have confusion (70% vs. 37%). Patients with NHAP were also more likely to die in the hospital (32% vs. 14%; P<.05 ) [21].
A retrospective cohort analysis comparing the clinical presentations of NHAP and CAP found that patients with NHAP were more likely to have altered mental status (55.9% vs. 11.3%; P<.001 ), tachypnea (40.9% vs. 22.8%; P<.001 ), and hypotension (7.0% vs. 1.1%; P<.001 ) [25]. The presence of subjective variables, such as cough, dyspnea, and pleuritic chest pain, could not be assessed. Patients with NHAP also had a significantly higher mortality rate (18.6% vs. 8.0%; P<.001 ). In a prospective cohort of 437 consecutive patients with pneumonia, 40 (9%) of the patients resided in nursing homes [20]. These patients with NHAP were less likely than were those with other types of pneumonia to have a productive cough (OR, 0.4; P=.02 ) or pleuritic pain (OR, 0.1; P=.03 ) but were more likely to be confused (OR, 2.6; P<.001 ). Compared with age-matched control individuals living in the community, the patients with NHAP had a significantly higher mortality rate (53.0% vs. 13.4%; P<.001 ).
An article reviewing 18 primary studies published between 1978 and 1994 evaluated the etiology of pneumonia in residents of long-term-care facilities [26]. In this study, the most common pathogens were gram-negative bacilli (18%), S. pneumoniae (16%), Haemophilus influenzae (11%), and S. aureus (6%). Mycoplasma, Chlamydia, and Legionella species accounted for <1% of cases, and 29% of cases did not have an identifiable pathogen. The primary studies showed marked variability in the frequency of causative pathogens: gram-negative bacilli isolation varied from 0% to 55% across studies; that of S. pneumoniae ranged from 0% to 39%; that of S. aureus, from 0% to 33%; and that of Legionella species, from 0% to 6%. The primary studies' evaluations of the causative pathogens were widely discrepant: some had no microbiological criteria, others required a high-quality sputum specimen, and some allowed positive blood culture results to suffice if sputum results were negative. None of the studies rigorously pursued the isolation of atypical organisms.
A prospective cohort of 104 “very elderly” patients (mean age ± SD, 82.3±5.5 years) admitted to the intensive care unit (ICU) with severe pneumonia requiring mechanical ventilation identified a pathogen in 55 patients (53%) by use of an aggressive and invasive approach that included bronchoalveolar lavage (BAL) [27]. Although no formal statistical comparisons of community residents and long-term-care—facility residents were performed, patients with NHAP had higher rates of altered mental status (76% vs. 42%) and fever (65% vs. 44%) but a lower rate of chest pain (5% vs. 20%) at admission. Compared with patients with CAP, fewer patients with NHAP had S. pneumoniae isolated (8.5% vs. 14.0%), but more patients with NHAP had S. aureus isolated (29.7% vs. 7.0%). All Staphylococcus isolates from patients with CAP were methicillin susceptible, whereas 78.6% of Staphylococcus isolates (11 of 14) from patients with NHAP were methicillin resistant.
In a similar study by the same authors, patients with NHAP requiring mechanical ventilation underwent BAL in an attempt to identify those at risk for harboring resistant pathogens [22]. The most common pathogens included S. aureus (23.9%, of which 61.9% were MRSA) and S. pneumoniae (18.2%). Predictors of infection with resistant pathogens included functional dependence and prior antibiotic exposure.
Another recent study assessed the risk factors for colonization with MDR organisms in residents of long-term-care facilities [28]. In a point-prevalence study of 84 individuals, surveillance nasal and rectal cultures were assessed for organisms resistant to 3 frequently prescribed antibiotics. The prevalence of colonization with MDR organisms was 51%, with the most common organisms being Providencia stuartii (31% of isolates), Proteus mirabilis (21%), Escherichia coli (19%), and Morganella morganii (19%). Independent predictors of colonization by multivariate regression analysis included fecal incontinence (OR, 3.78; P=.038 ) and prior antibiotic exposure (OR, 2.5; P=.047 ). PFGE identified high rates of identical or related strain types, which suggested substantial horizontal transmission.
Dialysis-associated pneumonia. Only 1 study was identified as dealing specifically with pneumonia in patients undergoing long-term hemodialysis [29]. This retrospective cohort analysis linked the waves 1, 3, and 4 Dialysis Morbidity and Mortality Study data sets with Medicare claims to identify 3101 hospital admissions for pneumonia in patients undergoing long-term hemodialysis. Overall, the frequency of microbiological confirmation was very poor (18.2%). In patients with microbiologically confirmed pneumonia, the most common pathogens were S. pneumoniae (18.7%), P. aeruginosa (15.4%), Klebsiella species (8.8%), and H. influenzae (8.2%). Despite high rates of colonization with MRSA in the dialysis population, Staphylococcus species were infrequently the causative pathogen (2.2%).
Pneumonia in patients receiving home infusion therapy or wound care. One study was identified that specifically assessed pneumonia in patients receiving home nursing care. In this prospective case-control study of 175 patients with MRSA infection, 41 patients had pneumonia [30]. Multivariate regression analysis identified a highly significant association between MRSA infection and prior receipt of home nursing care (OR, 3.7; P<.001 ). Other independent risk factors included prior hospitalization and transfer from another institution, such as a nursing home.
Pneumonia in patients undergoing chemotherapy. Three studies were identified as relating specifically to pneumonia in patients undergoing chemotherapy. In a prospective observational cohort study of 52 consecutive pneumonia cases among patients with acute leukemia undergoing chemotherapy, the presentation was relatively subtle [31]. The signs and symptoms present in more than half of patients included fever (98%), dyspnea (79%), rales (77%), chills (73%), cough (65%), and sputum production (58%). A causative pathogen was found in 71% of cases, but only 52% were identified antemortem. The most common organisms were fungi (25.0%), Pseudomonas species (23.1%), and Klebsiella species (13.4%).
In another prospective cohort of 242 consecutive pneumonia cases among patients with malignancy undergoing antineoplastic chemotherapy, the clinical presentation was similarly subtle [32]. Clinical presentation in more than half of patients included fever (90%), a positive radiographic finding (83%), and rales (65%). Gram-negative bacilli accounted for 90% of cases, with the most common pathogens being Klebsiella species (31.8%) and Pseudomonas species (18.6%).
The final study was a prospective surveillance study of neutropenic patients with bacteremic pneumonia [33]. Although 408 cases of pneumonia were identified, clinical and microbiological data were reported only for the 40 patients with concurrent bacteremia. The only sign of pneumonia present in more than half of patients was fever (95%). Cough was the most common symptom of pneumonia and was present in only 47% of patients. The most common pathogens identified were P. aeruginosa (42.5%), S. pneumoniae (30.0%), and E. coli (12.5%).
Pneumonia in patients with a relative harboring MDR pathogens. No studies were identified as specifically assessing either the clinical presentation or microbiological features of pneumonia in patients who have a relative with known MDR pathogens.
Pneumonia in patients who have recently received antibiotics. No studies were identified as specifically assessing either the clinical presentation or microbiological features of pneumonia in patients who have recently received antibiotics. However, several of the previously cited studies identified prior antibiotic exposure as a risk factor for colonization or infection with resistant pathogens [16, 22, 28].
Grading of Evidence
On the basis of a review of the 15 studies cited above, the 5 members of this workshop agreed that the evidence available to support this statement was category II for the statement in general, category II for the statement as it applies to hospitalized patients with HCAP, and category V for the statement as it applies to nonhospitalized patients with HCAP (table 2).
Level of Support
When voting on the support for this statement, 9% of the summit participants voted to accept the statement completely, 64% voted to accept the statement with some reservations, 18% voted to accept the statement with major reservations, and 9% voted to reject the statement with reservations. In comparison, of the 383 IDSA members who participated in the online survey, 40% voted to accept the statement completely, 47% voted to accept the statement with some reservations, 7% voted to accept the statement with major reservations, 5% voted to reject the statement with reservations, and 1% voted to reject the statement completely (figure 3).
Voting comparison for statement 2 (“The clinical and microbiological features of HCAP are more similar to HAP and VAP than to CAP”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. CAP, community-acquired pneumonia; HAP, hospital-acquired pneumonia; HCAP, health care—associated pneumonia; VAP, ventilator-associated pneumonia.
Voting comparison for statement 2 (“The clinical and microbiological features of HCAP are more similar to HAP and VAP than to CAP”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. CAP, community-acquired pneumonia; HAP, hospital-acquired pneumonia; HCAP, health care—associated pneumonia; VAP, ventilator-associated pneumonia.
Discussion
This statement is of key importance, given that the relevance of the remaining statements discussed at the summit hinges on the assertion that HCAP constitutes a distinct clinical entity with unique microbiological features. At present, there is limited conclusive evidence supporting this statement, as is reflected in the summit participants' grading of the evidence.
All of the reviewed studies support the assertion that the clinical features of HCAP are different from those of CAP. Although patients with HCAP are generally less likely to have symptoms, the fever response is preserved, and the likelihood of altered mental status is increased. This combination of factors may lead to delayed recognition and initiation of therapy, thereby explaining the increased rates of hypoxemia, systemic hypotension, and death.
Presently, there is only 1 study of the microbiology of HCAP as it is specifically defined by the ATS-IDSA guidelines [2]. Although this study has several limitations, it corroborates the assertion of earlier, less rigorous studies concluding that HCAP is more likely to be caused by organisms that are particularly virulent and/or resistant to antibiotic therapy.
Of the individual subsets grouped together under the HCAP umbrella, NHAP provides the richest data in terms of the published literature. However, these studies were also the most likely to be confounded by variables such as comorbid illness or prior antibiotic therapy. These studies routinely had limited microbiological data that were of very poor quality. Another concern is the possible obsolescence of older studies, given the changes in pathogen susceptibilities over time. Although nursing home residents have very high rates of colonization with MDR pathogens—and these organisms are seen more frequently in patients with NHAP—CAP pathogens are also frequently isolated from these patients.
The present NHAP knowledge base also suffers from a hospitalization bias; the available studies have rigorously evaluated the causative organisms only in hospitalized patients. This is problematic, because prior therapy has frequently failed for these patients at nursing homes, thereby artificially increasing the frequency of virulent and/or resistant organisms. There is also an extensive literature base showing the good clinical responses of many patients with NHAP who receive in-facility CAP therapy. In all, NHAP appears to be a “hybrid” entity, blending conventional CAP organisms with increased rates of HAP pathogens.
For the remaining subsets of patients with HCAP (those receiving hemodialysis, home infusion, wound care, chemotherapy, or recent antibiotics or who have a relative with resistant pathogens), the available data are very limited and likely suffer from obsolescence. Although it is intuitive to group these patients together, given their increased contact with the health system, the data do not clearly justify doing so. The high rate of fungal infections in patients undergoing chemotherapy reinforces that HCAP subgroups—although clearly distinct from CAP—are also different from one another.
Future Directions
Future directions discussed by the summit members reflect many of the limitations of previously discussed studies. Appropriately designed epidemiologic studies with rigorous microbiological criteria clearly are needed to better delineate the causative pathogens of CAP, HAP, and HCAP. Although the use of standardized diagnostic and laboratory criteria would be essential, these criteria do not exist. These large, observational cohorts would need to be followed longitudinally to assess changes in pathogens and resistance patterns over time.
Statement 3: The Recommended Evaluation of HCAP with Treatment Failures Is the Same as That for HAP
Rationale and Definition of Statement
Nonresolving pneumonia is variably defined but represents a clinical syndrome in which focal infiltrates persist with signs and symptoms of acute pulmonary infection (e.g., fever, expectoration, malaise, or dyspnea). Despite receiving a minimum of 10 days of antibiotic therapy, patients either do not improve or worsen clinically, or radiographic opacities fail to resolve within 12 weeks after the onset of pneumonia [34–36].
Nonresolving pneumonia often represents treatment failure or a superinfection [37, 38]. In addition to being the result of initial therapy failure or a noninfectious etiology, nonresolving pneumonia may also be the result of an overwhelming immune response to a specific pathogen. It is critical to identify patients at risk for nonresponding pneumonia, to institute early appropriate therapy. Patients with severe HCAP, underlying comorbidities, and certain etiologic agents (viral, atypical) are at greater risk for nonresolving pneumonia and may benefit from alternative supportive approaches (e.g., early tracheostomy) as well as from immune modulation in specific circumstances (e.g., progression to acute lung injury or pneumococcal sepsis).
Methods
The original PubMed search was conducted in November 2006 and was augmented with a search performed in April 2007. By selection of articles published in English on the duration of nonresolving pneumonia, 50 references focusing on both CAP and HCAP were identified. These were reviewed along with their bibliographies for additional references.
Evidence
The use of the clinical pulmonary infection score (CPIS) to define resolution of nosocomial pneumonia was evaluated by Luna et al. [39]. These investigators prospectively evaluated 63 patients with microbiologically confirmed VAP. CPISs were followed serially and were noted to improve as early as the third day of therapy in the group of survivors. Patients who did not survive did not demonstrate any improvement in their CPIS, particularly in oxygenation. Patients without treatment response were also significantly more likely to receive inadequate initial treatment and had a higher mortality rate, compared with patients with treatment response. This study suggests that a clinical scoring system may be useful in the early identification of patients with pneumonia whose conditions are unlikely to respond to therapy or who have delayed resolution, in part related to inadequate initial therapy.
Resolution can also be defined microbiologically. On the basis of serial quantitative cultures from respiratory secretions, the eradication or persistence of an organism can be demonstrated. In the prospective series studied by Garrard and A'Court [40], nonbronchoscopic lung lavage specimens were obtained from 83 patients with nosocomial pneumonia. The investigators found that serial culture results correlated well with the scored clinical responses. Culture counts increased in the few days before the onset of therapy and decreased dramatically after its initiation. In most cases, the culture counts had decreased by 24 h, but they decreased no later than 72 h in all cases of resolving pneumonia. Nonresolving pneumonia was thus equally well defined by both clinical and microbiological criteria, although clinical nonresponse required more time to determine.
Menéndez et al. [41] identified factors associated with failure of empirical treatment and determined the incidence of both early (<72 h) and late treatment failure and related implications on the outcome. A prospective, multicenter cohort study was performed involving 1424 hospitalized patients from 15 hospitals. Treatment failure occurred in 215 patients (15.1%): 134 (62.3%) had early failures and 81 (37.7%) had late failures. The causes were infectious in 86 patients (40%), noninfectious in 34 patients (15.8%), and undetermined in 95 patients (44.2%). The independent risk factors associated with treatment failure in a stepwise logistic regression analysis were liver disease, pneumonia risk class, leukopenia, multilobar CAP, pleural effusion, and radiologic signs of cavitation. Independent factors associated with a lower risk of treatment failure were influenza vaccination, initial treatment with fluoroquinolones, and chronic obstructive pulmonary disease. Mortality was significantly higher in patients with treatment failure than in those without treatment failure (25% vs. 2%). Failure of empirical treatment increased the rate of mortality due to CAP 11-fold after adjustment for risk class.
Rosón et al. [42] performed an observational analysis of a prospective series of 1383 nonimmunosuppressed hospitalized adults with CAP to identify and categorize causes and factors associated with early failure. At 48–72 h, 238 patients (18%) remained febrile, but most of them responded without further changes in antibiotic therapy, and 81 patients (6%) had early failure. The main causes of early failure were progressive pneumonia (n=54 ), pleural empyema (n=18 ), lack of response (n=13 ), and uncontrolled sepsis (n=9 ). Independent factors associated with early failure were older age (>65 years) (OR, 0.35), multilobar pneumonia (OR, 1.81), Pneumonia Severity Index (PSI) score >90 (OR, 2.75), Legionella pneumonia (OR, 2.71), gram-negative bacillary pneumonia (OR, 4.34), and discordant antimicrobial therapy (OR, 2.51). Compared with patients with treatment response, patients with early treatment failure had significantly higher rates of complications (58% vs. 24%) and overall mortality (27% vs. 4%) (P<.001 for both).
An etiologic diagnosis was established in 598 patients with early treatment response (48%) and 55 patients with early treatment failure (68%) (P<.01 ). of patients with an etiologic diagnosis, 316 (53%) with early response and 48 (87%) with early failure were classified as definitive. The most frequently identified pathogens in the early-response and early-failure groups were S. pneumoniae (23% and 22%, respectively), Legionella species (6% and 21%), H. influenzae (6% and 5%), and organisms associated with aspiration pneumonia (6% and 6%). Legionella pneumophila and gram-negative bacilli were found more frequently in patients with early failure (P<.001 and P=.03 , respectively).
Most patients were initially treated with a single antimicrobial agent, mainly in the early-response group (77% of those with early responses and 65% of those with early failures). The antibiotics most frequently prescribed were the β-lactams (mainly ceftriaxone and amoxicillin-clavulanate). Overall, of the 81 patients for whom treatment failed, concordance of therapy could be determined in 52 patients. In general, patients with early failure received discordant antimicrobial therapy more frequently (16 [31%] of 52 patients) than did patients with early response (52 [9%] of 584 patients). Treatment failed in 1 patient owing to resistance to a recommended regimen; he experienced breakthrough levofloxacin-resistant S. pneumoniae bacteremia (MIC of levofloxacin, 64 µg/mL) after 48 h of intravenous levofloxacin therapy.
Because most studies of resolution have included only patients with CAP, the normal resolution of nosocomial pneumonia is more uncertain. Several investigators, however, have identified risk factors for poor outcomes, including death. By inference, the factors promoting a poor outcome may be linked to delayed resolution. Celis et al. [43] identified prognostic factors for nosocomial pneumonia. They reported that respiratory failure, the presence of a fatal underlying condition, age >60 years, and the presence of bilateral radiographic involvement were associated with a significantly increased risk of mortality. Additionally, infection with “high-risk” organisms, such as P. aeruginosa, S. aureus, other gram-negative bacilli, Candida species, or Aspergillus species, was associated with worse outcomes.
In addition to observing longer resolution times with gram-negative bacterial infections, Graybill et al. [44] noted the significance of certain host factors, such as cardiovascular disease, a variety of malignant neoplasms, prior cerebrovascular accident, alcoholism, chronic obstructive pulmonary disease, renal impairment requiring hemodialysis, and diabetes. In his 1973 study of 82 patients with nosocomial pneumonia, prolonged resolution was defined as radiographic abnormalities extending beyond 4 weeks. Mortality was higher among patients who developed the pneumonia postoperatively, while undergoing ventilation, after aspiration, or while receiving antibiotic therapy. Other studies of nosocomial pneumonia have identified similar risk factors [45, 46].
Montravers was one of the first investigators to compare microbiological data with clinical outcomes [45]. Serial bronchoscopy and protected specimen brushings were performed in 76 patients requiring mechanical ventilation. The level of eradication of a pathogen was shown to correlate with subsequent clinical improvement: high-level growth (>103) found by follow-up protected specimen brushing at 72 h was more common among patients with persistent symptoms of pneumonia. Despite sterilization of the lung in 94% of patients, clinical resolution occurred in only 20% of cases.
To further evaluate the relationship between clinical and microbiological data, Dennesen et al. [47] performed a prospective study of 27 patients with VAP. Temperature, WBC count, and oxygenation (as measured by PaO2/FiO2 ratios) were scored daily. Semiquantitative cultures of endotracheal aspirates were also obtained at the day of admission and twice weekly thereafter. Unlike in the Montravers et al. [45] study, persistent bacterial growth was more common with certain pathogens, including Pseudomonas and Enterobacteriaceae species, despite significant clinical improvement. These findings suggest that bacterial eradication is an imperfect marker of clinical response in VAP.
Chastre et al. [48] evaluated patients with microbiologically treated VAP randomized to receive 8 days versus 15 days of appropriate antimicrobial therapy. This study found similar rates of survival and secondary outcomes between the 2 treatment groups. However, patients with VAP secondary to P. aeruginosa pneumonia who were treated for 8 days were more likely to require reevaluation and retreatment in some cases. Similar findings in nosocomial pneumonia due to P. aeruginosa and other high-risk pathogens have been demonstrated by other investigators [49, 50].
The ATS-IDSA guidelines have focused attention on patients with nonresolving pneumonia, particularly patients showing no response to the initial antimicrobial regimen [3]. Patients without response to initial therapy or who develop a pattern of nonresolving pneumonia should be carefully evaluated to ensure that antimicrobial treatment is covering the offending pathogen, that a noninfectious diagnosis has not been missed, or that some complicating factor has not occurred.
Grading of Evidence
On the basis of a review of the studies cited previously, the 5 members of this workshop agreed that the evidence available to support this statement was category V for the statement in general, category IV for the statement as it applies to hospitalized patients with HCAP, and category V for the statement as it applies to nonhospitalized patients with HCAP (table 2).
Level of Support
When voting on the support for this statement, 9% of the summit participants voted to accept the statement with some reservations, 82% voted to accept the statement with major reservations, and 9% voted to reject the statement completely. In comparison, of the 383 IDSA members who participated in the online survey, 51% voted to accept the statement completely, 41% voted to accept the statement with some reservations, 4% voted to accept the statement with major reservations, 3% voted to reject the statement with reservations, and 1% voted to reject the statement completely (figure 4).
Voting comparison for statement 3 (“The recommended evaluation of HCAP with treatment failures is the same as that for HAP”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. HAP, hospital-acquired pneumonia; HCAP, health care—associated pneumonia.
Voting comparison for statement 3 (“The recommended evaluation of HCAP with treatment failures is the same as that for HAP”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. HAP, hospital-acquired pneumonia; HCAP, health care—associated pneumonia.
Discussion
The optimal evaluation of nonresolving pneumonia is not clearly established. The existing data regarding treatment failure in HAP are predominately from VAP studies. In these reports, the identified causes are very diverse, highlighting the need for a rigorous stepwise approach incorporating repeated microbiological evaluation, evaluation of infectious complications, and assessment of noninfectious causes. At present, there are no existing data evaluating treatment failure in HCAP. The summit participants agreed, on the basis of professional opinion alone, that the evaluation of nonresolving HCAP should be the same as that for other classes of pneumonia until definitive data specific to HCAP are available.
Future Directions
Summit participants agreed that, given our current lack of knowledge regarding the evaluation of HCAP treatment failures, retrospective analyses of large data sets are needed to guide future prospective studies of this complex topic. Data are specifically needed regarding whether the assessment of HCAP treatment failure should be universal or should vary by subset (e.g., patients undergoing dialysis, nursing home residents, and patients with prior hospitalization).
Statement 4: The Definitions Are the Same for HCAP and HAP Treatment Failures
Rationale and Definition of Statement
Nonresolving pneumonia is a common clinical problem among patients with HCAP and patients with HAP. Although quantifying the frequency of this problem is difficult, previous studies reported that ∼15% of pulmonary consultations and 8% of bronchoscopies were done to evaluate “nonresolving pneumonia” [51]. Treatment failures constitute a subset of all nonresolving pneumonias. For patients with HCAP and HAP, identifying treatment failure is important because the presence of treatment failure implies that, without some type of intervention, the patient is more likely to have an adverse outcome.
The most recent ATS-IDSA nosocomial pneumonia guidelines identified a subset of patients at risk for harboring resistant organisms despite their residence in the community [3]. The new guidelines tried to define HCAP as a distinct clinical entity on the basis of this recognition.
However, the guidelines provide a limited discussion of the definition of treatment failure and nonresolving pneumonia. The guidelines do recommend serial assessment of clinical parameters to define response to initial empirical therapy and review some of those parameters that might be useful in assessing response: chest x-ray (CXR film) findings, WBC count, oxygenation, temperature, CPIS, and culture results. However, there are few specifics and limited discussion as to whether the definition of treatment failure should be the same for HCAP and HAP and whether the causes of treatment failure are the same in these 2 groups. On the basis of this, the present review aimed to assess the strength of the evidence supporting the assertion that the definitions are the same for HCAP and HAP treatment failures.
Methods
The literature search, conducted in the PubMed database, was completed in November 2006. Because the current definition of HCAP would include some patients previously classified as having CAP, the search was intentionally broad. The search term “Pneumonia treatment failure” yielded 344 articles. “Health care associated pneumonia treatment failure” yielded 7 articles, “Health care associated pneumonia failure” yielded 94 articles, “Hospital Acquired Pneumonia treatment failure” yielded 94 articles, and “Hospital Acquired Pneumonia failure” yielded 33 articles. “Pneumonia Prediction Outcome” yielded 54 articles. “Health care associated pneumonia failure” in a search for guidelines, meta-analysis, or randomized, controlled trials yielded 34 articles, and “Hospital Acquired Pneumonia failure” in a search for guidelines, meta-analysis, or randomized, controlled trials yielded 104 articles. After elimination of duplicates, there were 736 articles. After the search was limited to articles in English and the titles and abstracts were reviewed, only 65 manuscripts were relevant. These were reviewed, and 12 studies were selected. References from these articles and review articles identified 1 additional report, for a total of 13 studies that were included. Subsequently, after assessment of what would constitute a “good” functional definition of treatment failure, 2 of these were eliminated, leaving 11 studies.
Evidence
As part of the assessment process, the populations of interest were carefully defined on the basis of the most recent guidelines. Articles were included that reported information on treatment failures, provided that at least some of the patients studied met the current definitions for having either HAP or HCAP. Most studies fell into 1 of 2 categories: for HAP, most studies involved patients with VAP or pneumonia in the ICU; for HCAP, most studies involved patients with CAP, in which at least a subset would have met the ATS-IDSA definition for having HCAP.
An attempt was made to clarify which definitions of treatment failure were being compared. However, the ATS-IDSA guidelines are not that specific, and a wide range of definitions have been used in different studies. Given this ambiguity, it was important to identify what would constitute a “good” clinical definition of treatment failure and then to ask whether there was a “good” definition applicable to both HCAP and HAP.
Conceptual framework for assessing definitions of treatment failure. When trying to assess whether a particular definition of treatment failure is “good,” it is important to recognize the clinical context in which the term “treatment failure” will be used. Often, there is a significant amount of diagnostic uncertainty as to whether a given patient even has pneumonia, because a specific pathogen often cannot be identified. As a result, when a presumed pneumonia fails to respond to treatment, one of the first questions is whether the diagnosis of pneumonia is correct, because many conditions can mimic the disease. Therefore, it is more precise to use the term “nonresolving pneumonia syndrome,” because, in many instances, infection may not be present. Some authors might consider treatment failure to include only those patients with true pneumonia. For the purposes of this review, the more clinically relevant idea of a nonresolving pneumonia syndrome will be used, because physicians in practice may not always be certain whether their patients have true pneumonia.
Given the context of high diagnostic uncertainty, another way to formulate the question is to ask whether there is a set of criteria that effectively discriminates between those patients who are likely to do well with their current treatment regimen and those who are likely to have adverse outcomes. These criteria would need to identify patients with a significant variance from the normal pattern of resolution for both patients with HCAP and patients with HAP, to help guide treatment and management decisions.
Note that, because the definition involves treatment failure, for any definition to be clinically useful, it must use at least 1 intermediate outcome measure assessed after treatment is started. Baseline predictors of outcome that are measured only before initiation of treatment, such as the CPIS, CURB (confusion, urea >7 mmol/L, respiratory rate of 30 breaths/min, and blood pressure <90 mm Hg systolic or <60 mm Hg diastolic), or Acute Physiology and Chronic Health Evaluation (APACHE) scores at admission, would not qualify, because they cannot show whether the treatment is actually working. These would be prognostic measures but would not, by themselves, constitute markers of treatment failure. Note that results of serial testing with these same instruments, such as the development of a worsening CPIS, would constitute a potential definition of treatment failure because it measures how well the patient responds to treatment. Therefore, a “good” definition of treatment failure would need to integrate at least some intermediate outcome data obtained after initiation of treatment.
This idea of intermediate outcomes is also relevant when looking at mortality. Death is frequently used as a definition of treatment failure. Note that death, while being the ultimate epidemiologic measure of treatment failure, does not constitute a clinically useful definition of treatment failure, because it precludes any corrective action being taken. Indeed, clinical definitions of treatment failure are really aimed at predicting the likelihood of death or permanent disability, because the hidden premise in constructing “good” clinical definitions of treatment failure is that they must allow clinicians the opportunity to intervene to potentially alter the course of disease. Therefore, clinically useful definitions of treatment failure will require measures that occur after the initiation of treatment but before death, so that intervention is possible. Finally, good definitions would ideally need to be readily available in everyday clinical practice, affordable, and highly reproducible between centers.
Studies of HAP. Given this conceptual framework, studies were categorized on the basis of sample size, patient population, the definition of pneumonia used for entry, the definition of treatment failure used, the treatment failure rate reported, and the causes of treatment failure reported (table 3). Most studies could not meet all of the criteria of a good definition of treatment failure, but any study that provided insight into any of these categories was eligible. The definitions of treatment failure were analyzed, and only definitions that might have at least 1 intermediate outcome were included.
Studies of pneumonia treatment failure that included patients meeting the definition of having hospital-acquired pneumonia (HAP).
Studies of pneumonia treatment failure that included patients meeting the definition of having hospital-acquired pneumonia (HAP).
A prospective, observational, single-center trial of patients with VAP by Dennessen et al. [47] described the clinical and microbiological response to treatment, as well as relapses in patients with microbiologically confirmed VAP. All patients had adequate initial antibiotic therapy. The investigators measured time to resolution of colony-forming units, WBC count, temperature, and oxygenation, as measured by the PaO2/FiO2 ratio. Most patients who showed improvement in their clinical and microbiological measures did so within the first 6 days. The mean (median) time to complete resolution of all parameters was 9 (8) days. When microbiological parameters were excluded, the mean (median) time to complete resolution of clinical and laboratory parameters was 6 (6) days. The only covariate of resolution that reached statistical significance was oxygenation, as measured by the PaO2/FiO2ratio (P<.01 ). Six patients developed a second episode of VAP; 3 had relapses of illness caused by the original pathogen, whereas the other 3 had superinfection with new pathogens, all 3 of which were Pseudomonas species. The study was too small to demonstrate a difference based on pathogens or between survivors and nonsurvivors.
Another study of 63 patients with microbiologically confirmed VAP evaluated the utility of the CPIS as a predictor of response [39]. CPIS was evaluated at days −3, 0, 3, 5, and 7. The mortality rate was 51% (32 of 63 patients). A decrease in CPIS after initiation of treatment was associated with significantly lower mortality rate (P=.018 for patients with a CPIS <6 at 3 or 5 days after onset of VAP, vs. those with a CPIS >6). of the components of the CPIS, only oxygenation, as measured by the PaO2/FiO2 ratio, was able to distinguish survivors from nonsurvivors. This was consistent with the finding that, among patients subsequently found to have received adequate initial antibiotic therapy (defined as the identified pathogen being susceptible to the initial antibiotics prescribed), PaO2 improved from day 0 to 3, whereas those receiving inadequate initial antibiotic therapy had worsening oxygenation. Thus, in this study, serial measures of CPIS and oxygenation were able to identify patients more likely to have an adverse outcome.
A similar study of 63 patients with VAP evaluated the ability of the APACHE II, Sepsis-related Organ Failure Assessment (SOFA), and CPIS to predict mortality [56]. The mean APACHE II, SOFA, and CPIS scores all were higher at the time of VAP diagnosis in survivors than in nonsurvivors (P=.001 , .002, and .025, respectively). However, there was no measurement of APACHE II, SOFA, or CPIS subsequent to the initiation of treatment for VAP. Therefore, although these scores were good prognostic markers, they were not considered to provide clinically useful definitions of treatment failure.
A prospective, multicenter, observational study of ICU-acquired pneumonia evaluated the causes of nonresponse to treatment in 71 patients [52]. Pneumonia was defined as the presence of a new or progressive infiltrate with at least 2 of the following 3 criteria: leukocytosis, fever, and purulent sputum. Not all cases of pneumonia were microbiologically confirmed. Nonresponse was defined as 1 of the following occurring 72 h after treatment: (1) failure to improve oxygenation or intubation 24 h after initiation of antibiotic therapy plus purulent respiratory secretions; (2) persistent fever or hypothermia; (3) worsening radiographic infiltrates; or (4) occurrence of septic shock or multiple organ system failure not present on day 1. Nonresponse occurred in 44 (62%) of patients. This definition of nonresponse was associated with a significantly increased risk of hospital mortality (50% vs. 7%; P<.001 ). Possible causes of nonresponse could be identified in 28 of the 44 patients without response, with 8 of these 28 patients having >1 cause. Causes of nonresponse included noninfectious causes (7), superinfection (6), inadequate initial treatment (10), and concomitant foci of infection (13).
A prospective observational study of VAP evaluated factors associated with the recurrence of VAP among 103 patients who survived VAP for 8 days [53]. VAP and recurrent VAP were defined by CXR film findings, purulent secretions, and BAL findings. Predictors of recurrence were measured at the time of the initial bronchoscopy and on day 8 of VAP. Recurrence was identified in 28 patients (27%) at a mean of 21 days after the initial episode of VAP. Causes of recurrent VAP included relapse (same organism) in 11 patients and superinfections in 17 patients. Multivariable analysis identified a day 1 radiology score >7, a day 8 temperature >38°C, and acute respiratory distress syndrome (ARDS) on day 8 as risk factors for subsequent recurrence of VAP. However, no data were available to determine whether the presence of these factors on day 8 differentiated survivors from nonsurvivors; this was not considered to be evidence that these markers could be used clinically as definitions of treatment failure.
A prospective, randomized, controlled trial [48] comparing 8 versus 15 days of therapy for VAP provided the data necessary for a nested observational cohort trial evaluating predictors of infection recurrence and death in patients with VAP [54]. of 401 patients with microbiologically confirmed VAP in the study, 110 (27%) developed recurrent infections, some of which were polymicrobial. Relapse with the same pathogen occurred in 56 patients (14%), and superinfection occurred in 77 patients (19%). Predictors of VAP recurrence measured on day 8 after VAP onset included SAPS II admission score; radiology score; temperature; gram-negative, nonfermenting pathogens; or MRSA. VAP recurrence was not associated with 28-day mortality (17% mortality rate for those with recurrence, vs. 18% for those without recurrence; P=.88 ). Only sex, age, day 8 SOFA score, and gram-negative nonfermenting pathogens were predictive of 28-day mortality. Therefore, in this study, the only intermediate outcome measure able to identify those more likely to have an adverse outcome was the organ failure score on day 8.
A prospective, randomized, controlled trial comparing 2 different drug regimens for empirical treatment of pneumonia in 400 patients in the ICU [55] provided data on treatment failures as well. Clinical failures in this trial were defined as either persistence or progression of symptoms leading to a change of antibiotics, a worsening CXR film finding leading to a change of antibiotics, superinfection, or death. A nonbacterial origin of the pneumonia was identified in 3.8% of the patients. In the clinical efficacy analysis, 73 patients (18%) had persistent pneumonia. Superinfection occurred in 27 patients, resulting in the death of 6 patients. There were 277 patients evaluable for bacteriologic efficacy analysis; of these, 223 had HAP. Death occurred in 113 (28%) of 399 patients receiving the study drug. There was no clear linkage between any intermediate outcome and mortality reported.
Studies of populations that included patients with HCAP. By use of the same conceptual framework, studies were categorized that included at least a significant proportion of patients with HCAP on the basis of the same criteria of sample size, patient population studied, definition of pneumonia used for entry, definition of treatment failure used, treatment failure rate reported, and causes of treatment failure reported (table 4). Most studies could not meet all of the criteria for a good definition of treatment failure, and many studies were primarily of CAP. However, any study that provided insight into any of these categories was eligible. The definitions of treatment failure were analyzed, and only definitions that might have at least 1 intermediate outcome were included.
Studies of pneumonia treatment failure that included some patients meeting the definition of having health care—associated pneumonia (HCAP).
Studies of pneumonia treatment failure that included some patients meeting the definition of having health care—associated pneumonia (HCAP).
A prospective observational study of patients hospitalized for CAP included a significant number of patients with neoplasia and immunosuppression and, therefore, was included [57]. CAP was defined as an acute illness with a new or progressive infiltrate on chest radiograph associated with 1 of the following respiratory signs or symptoms: fever (temperature, >38°C), a new cough, purulent sputum production, new dyspnea and/or tachypnea (>20 breaths/min), pleuritic pain and leukocytosis (>10×109 cells/L) or leukopenia (<4×109 cells/L), or >10% band forms if the leukocyte count was between 4×109 cells/L and 10×109 cells/L. A total of 224 of 228 patients were evaluable and completed follow-up. Eight patients developed antibiotic adverse effects, but these were not considered to be treatment failures; 54 patients (24%) had treatment failure. The study defined treatment failure as at least 1 of the following: fever for >3 days (or for >6 days if bacteremic), clinical deterioration necessitating a change in antibiotics, or death after at least 48 h of antibiotic therapy. Fourteen patients (26%) died. The most common causes of treatment failure were host factors in 34 patients (63%), unusual pathogens in 10 patients (19%), superinfection in 4 patients (7%), incorrect drug dosing or compliance in 3 patients (6%), and incorrect diagnosis of pneumonia in 3 patients (6%).
A prospective, observational, single-center study of treatment failure in CAP was included because of a high percentage of cases with malignancy, renal disease, or prior outpatient antibiotic therapy [58]. Treatment failures were defined as either nonresponding pneumonias (defined as persistent fever with a temperature >38°C and/or clinical symptoms [malaise, cough, expectoration, dyspnea] after at least 72 h of antimicrobial treatment) or progressive pneumonias (defined as clinical deterioration in terms of the development of acute respiratory failure requiring ventilatory support and/or septic shock after at least 72 h of antimicrobial therapy). During the study, there were 444 patients with CAP, of whom 49 were identified as experiencing treatment failure (11%). of these 49 patients, 30 (61%) were patients without treatment response, and 19 (39%) had progressive pneumonia. A definite etiology of treatment failure could be established in 32 (65%) of the 49 patients. Etiologies comprised primary infections (n=8 ), definite persistent infection (n=4 ), nosocomial infection (n=8 ), noninfectious causes (n=8 ), and combinations of the above (n=4 ). Those classified as having nonresponding pneumonia were more likely to have persistent infections (14 of 30) than nosocomial infections (2 of 30). Those classified as having progressive pneumonia were less likely to have persistent infections (4 of 19) and more likely to have nosocomial infections (6 of 19). Among those with treatment failure, only nosocomial pneumonia was associated with mortality in multivariate analysis. However, there was no comparison reported between those who developed treatment failures and those who did not, in terms of their mortality risk.
A prospective observational study of severe pneumonia in very elderly individuals was included because 47 of the 104 patients enrolled were nursing home residents [27]. The primary objective was to evaluate the prevalence of pathogens in this population and its impact on morbidity and functional status. No specific definition of treatment failure was used. In multivariate analysis, only 4 variables were predictive of mortality: multilobar involvement, septic shock at presentation, inadequate antimicrobial therapy, and 24-h urine output.
A prospective, single-center, observational study of early treatment failures in patients hospitalized for CAP had a small percentage of patients with either malignancy, renal failure, or nursing home residence [42]. Early failures were defined as lack of response or worsening of clinical or radiographic status at 48–72 h, requiring changes in antibiotic therapy or invasive procedures. Early failure occurred in 81 (6%) of 1383 patients. The most common causes were progression of pneumonia (67%) and empyema (22%). Superinfection occurred in only 3 cases (4%). Among patients with an identified pathogen, initial antibiotics that did not cover the pathogen were more common. Early failure was associated with higher complication rates (P<.001 ), increased hospital length of stay, and increased mortality (27% vs. 4%; P<.001 ).
A meta-analysis of 16 studies evaluating the causes of treatment failure in clinical trials of CAP was included [59]. There was no stratification with regard to the patient populations involved, although analysis of the individual studies demonstrated that some patients with HCAP would have been included on the basis of prior antibiotic therapy and comorbidities. In this meta-analysis, 6 different definitions of treatment failure were used, one of which was death; another was discontinuation for personal reasons. Persistent fever >72 h after antibiotic therapy, clinical deterioration requiring admission to an ICU or requiring vasopressors, change of antibiotic therapy for any cause, and resistant pathogens with worsening symptoms constituted the other 4 definitions of treatment failure used. There was significant heterogeneity in terms of failure rates reported (range, 0%–34%; P=.008 ). When adverse effects of medications were excluded, treatment failures occurred in 333 patients (16%). In the majority of cases (82%), no cause could be identified. Resistant pathogens were the most common identified cause (n=28 ; 8%). Superinfection occurred in only 7 patients (2%). There was no stratification based on HCAP risk factors, and no evaluation of the relationship between treatment failure and mortality risk could be made.
Grading of Evidence
On the basis of a review of these 11 studies, the 5 members of this workshop agreed that the evidence available to support this statement was category IV for the statement in general, category V for the statement as it applies to hospitalized patients with HCAP, and category V for the statement as it applies to nonhospitalized patients with HCAP (table 2).
Level of Support
When voting on the support for this statement, 9% of the summit participants voted to accept the statement completely, 36% voted to accept the statement with some reservations, 46% voted to accept the statement with major reservations, and 9% voted to reject the statement with reservations. In comparison, of the 383 IDSA members who participated in the online survey, 47% voted to accept the statement completely, 42% voted to accept the statement with some reservations, 7% voted to accept the statement with major reservations, 3% voted to reject the statement with reservations, and 1% rejected the statement completely (figure 5).
Voting comparison for statement 4 (“The definitions are the same for HCAP and HAP treatment failures”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. HAP, hospital-acquired pneumonia; HCAP, health care—associated pneumonia.
Voting comparison for statement 4 (“The definitions are the same for HCAP and HAP treatment failures”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. HAP, hospital-acquired pneumonia; HCAP, health care—associated pneumonia.
Discussion
At present, there is limited conclusive evidence supporting this statement, as is reflected in the summit participants' grading of the evidence. Because the recognition and classification of HCAP is new, none of the studies cited specifically evaluated patients with HCAP exclusively. Additionally, the lack of a standardized definition of treatment failure makes conclusive statements regarding treatment failure problematic. Furthermore, the existing literature did not report stratified analyses for the HCAP and HAP populations; therefore, any conclusions should be viewed as preliminary.
Given these limitations, this systematic assessment of treatment failure definitions and their relationship to mortality in HAP and HCAP raises interesting questions and preliminary observations. First, because of differences in study design, inclusion criteria, populations studied, and definitions used, there is significant heterogeneity in terms of treatment failure rates both within and between groups. Although it appears that treatment failure rates are higher among patients with HAP than among patients with HCAP, because of the lack of a stratified analysis, this is not necessarily the case. HCAP may fall somewhere between HAP and CAP in terms of treatment failure rates. However, a properly stratified analysis needs to be performed before more-specific conclusions can be drawn, and clinically useful definitions need to be standardized.
One consistent finding was that superinfection appeared to be more of a problem in patients with HAP than in the CAP/HCAP population reviewed. Unusual or resistant pathogens were more of a problem in the latter group. This is probably a result of differences in infection control and exposure, antibiotic use, and host factors. However, this finding should be viewed with caution when HAP and HCAP are compared, owing to the lack of properly stratified data analysis.
Finally, this review demonstrates the importance of constructing sound and clinically useful definitions before approaching complex problems. Whether the same or different definitions of treatment failure are used for HAP and HCAP does not matter if the definitions are not valid and clinically useful. At a minimum, a “good” definition of treatment failure should do the following:
Provide good discriminatory power and capture variance from the norm
Utilize information that is intermediate in time, prior to the outcome of interest
Identify a group/subset in whom an intervention is needed
Be widely available
Be highly reproducible
Encompass the clinical scenario of treatment failure as a syndrome of unknown etiology (not necessarily infectious), rather than being limited to those patients in whom a primary infection is certain, because in clinical practice this is often not the case
Include proof that definitions of treatment failure are predictive of future adverse outcomes of interest
Future Directions
Future directions discussed by the summit members focused on the limitations of the previously discussed studies. Appropriately designed epidemiologic studies are clearly needed to better delineate the causes of treatment failure in CAP, HAP, and HCAP. Careful consideration and construction of clinically useful definitions of treatment failure should be done before these studies are performed, so that clinically useful recommendations can be made. Once standardized definitions can be validated, trends in treatment failure can be followed longitudinally to assess changes in treatment failure rates and causes over time.
Statement 5: Severe CAP Is Not HCAP
Rationale and Definition of Statement
Over the past decade, there has been an increase in infections due to MDR pathogens in individuals referred to acute-care hospitals who have been previously hospitalized, have received broad-spectrum antibiotics, or reside in nursing homes or other long-term-care facilities [2, 3, 13, 16, 23, 60]. These individuals may need different empirical antibiotics for pneumonia, to avoid delays in receiving appropriate antibiotic therapy. Such delays have been shown to result in poorer outcomes from serious infections [2, 3, 61–63].
HCAP, as defined in the 2005 ATS-IDSA guidelines, includes pneumonia in patients referred to hospitals for evaluation and treatment who were more likely to be colonized or infected with MDR pathogens [3]. Risk factors for different MDR pathogens are variable and complex; these factors may include previous hospitalization, prior antibiotic therapy, chemotherapy, hemodialysis, wound therapy, and residence in nursing homes and long-term-care facilities, either alone or in combination. Fewer data are available on these patients managed in non—acute-care settings where presentations, clinical data, and therapy are more limited [64, 65].
Principles for the treatment of patients with HCAP referred to acute-care hospitals and clinics share more similarities in etiology and management with those for HAP and VAP than with those for CAP (figure 2) [3, 23]. Thus, broader-spectrum, empirical antibiotic therapy has been suggested to be aimed at MDR pathogens, such as P. aeruginosa, Klebsiella pneumoniae producing extended-spectrum β-lactamases (ESBLs), or MRSA. The principles of the 2005 ATS-IDSA guidelines were aimed at appropriate, initial antibiotic therapy regardless of disease severity. In addition, de-escalation of antibiotics was recommended for patients with treatment response, on the basis of the availability of microbiological cultures, clinical response, and reducing duration of antibiotic therapy to 7–8 days.
There are many definitions of severe CAP, and several scoring systems have been used for assessing CAP severity and the need for hospital admission or intensive care [66–69]. Patients with severe CAP or severe HCAP may need intensive care or mechanical ventilation or may have pneumonia complicated by sepsis, shock, bacteremia, or multiple-organ failure. These assessments, often performed in the ED or clinic for patients with CAP, should also apply to patients with HCAP.
This section examines the hypothesis that “severe CAP is not HCAP” in the acute-care setting. Clearly, definitions are key, and applications to all types of patients are difficult. For example, HCAP management in the acute-care setting is better defined than management in nursing homes and other types of long-term-care facilities.
Methods
Severe CAP. A search of PubMed was performed for severe CAP on 14 November 2006. The search term “community acquired infections” (4552) was combined with the search term “pneumonia” (54,650) and then was combined with the “AND” function for a total of 2562 articles. A search for articles with the text words “severe CAP” or “SCAP” yielded 248 articles. A search for articles with the text words “severe community acquired pneumonia” yielded 225 articles, and that combined with the “OR” function yielded a total of 420 articles. The 2 searches above combined with the “AND” function yielded 194 articles. A search for articles with the text words “severe community acquired pneumonia” yielded 200 articles, and, combined with the “OR” function, the above result yielded 194 articles, for a total of 250 articles. No abstracts were included.
HCAP. A literature search of PubMed was performed on 14 November 2006. The search term “cross infection” yielded 31,350 articles. The term “pneumonia” generated 54,650 articles. When the term “pneumonia” was combined with the “AND” function, there were 24,096 articles noted. The text words “healthcare associated” or “health care associated” yielded 331 articles. The search term “pneumonia” yielded 54,650 articles and, when combined with the term “HCAP,” yielded 23 articles. The text words “healthcare associated pneumonia” or “health care associated pneumonia” or “healthcare” yielded 24 articles. When the 3 searches above were combined with the “OR” function, there were a total of 2509 articles. Fewer articles were identified using English as the only language. Eighteen articles were deemed pertinent to this review.
Evidence: What is Severe CAP?
CAP, like HCAP, has varying degrees of severity and may be caused by a wide spectrum of bacterial, atypical, or viral pathogens [70]. Patients with severe CAP are often evaluated in EDs before hospital admission, and patients with severe CAP often require admission to the ICU as a result of shock, ARDS, or multiple-organ failure requiring mechanical ventilation. These patients represent a subset of patients with higher morbidity, mortality, and length of stay in the hospital and ICU.
Microbiology of severe CAP. Severe CAP may be caused by several pathogens, including S. pneumoniae, H. influenzae, and anaerobic bacteria that typically are not MDR strains [70]. Even in CAP caused by S. aureus or gram-negative bacteria, such as K. pneumoniae, the pathogen is usually not MDR if the patient has not had prior antibiotic therapy or close contact with the health care system. Atypical pathogens, such as Chlamydophila pneumoniae, Mycoplasma pneumoniae, and L. pneumophila, are common in the United States. Coinfection with these bacteria occurs, and, in contrast to HCAP, the pathogens are not MDR or necessarily associated with prior antibiotic use or contact with the health care system. In addition, although CAP is more likely to be caused by bacteria, influenza viruses, respiratory syncytial viruses, and adenoviruses are also important. Patient risk factors for CAP include underlying medical diseases, such as chronic lung disease; exposure to animals; risk of aspiration; exposure to other infected persons; or seasonal epidemics.
In a review of severe CAP by Ewig and Torres [71], microbial patterns in Barcelona, Spain; Lille, France; and South Africa were examined. Rates of isolation of S. pneumoniae were 15%, 27%, and 29%, respectively; those of K. pneumoniae were 2%, 2%, and 19%, respectively; and those of S. aureus were 0%, 19%, and 3%, respectively. A review of etiologic agents identified in 16 studies of severe CAP found that rates of S. pneumoniae ranged from 12% to 38%, rates of H. influenzae from 0% to 13%, rates of enteric gram-negative bacilli from 0% to 34%, rates of S. aureus and other Staphylococcus species from 0% to 15%, rates of P. aeruginosa from 0% to 5%, and rates of L. pneumophila from 0% to 30%. Unfortunately, these differences represent the diversity of patient populations studied and different time periods but do not address MDR pathogens. The study includes some patients who would now be classified as having HCAP because of their risk for infection with MDR pathogens. A later study by Rello et al. [72] in Barcelona, Spain, compared the microbiological assessments of 106 patients with severe CAP requiring mechanical ventilation with 98 patients with CAP who were not receiving ventilation. A microbiological diagnosis was made in 57.3% of patients, and the most common bacterial pathogens were S. pneumoniae, L. pneumophila, and H. influenzae. P. aeruginosa (6.6% vs. 1.0%; P<.05 ) and L. pneumoniae (15.1% vs. 7.1%; P<.05 ) infections were more common in intubated patients than in nonintubated patients. Overall mortality was 44.3% in intubated patients, versus 23.5% overall. of note, bacteriologic investigation led to changes in antibiotic therapy in 41.6% of patients, including 11 patients (5%) in whom initial treatment was ineffective.
Definitions of severe CAP. Several methods have been designed to assess CAP severity to determine the need for hospital admission or triage to intensive care and to identify those with a higher risk of death [67–69, 73, 74]. Severe CAP is independent of the source of the pathogen or risk for infection with an MDR pathogen and may occur in patients with CAP or HCAP.
Because severe CAP has many definitions, a universally accepted one does not exist. Several investigations have suggested methods of identifying patients with severe CAP who should be admitted to the hospital or ICU, those at risk for respiratory or multiple-organ failure, and those with a greater risk of mortality [68–70]. Ewig et al. [68] defined severe CAP as CAP requiring admission to the ICU. of the 64 (16%) admitted patients with severe CAP who were followed prospectively, the mortality rate was 30%, versus 5% for those not admitted. The 10 criteria for severe CAP, initially defined by the ATS, include respiratory rate >30 breaths/min, PaO2/FiO2 <250, bilateral involvement on chest radiograph, multilobar involvement, systolic blood pressure <90 mm Hg, diastolic blood pressure <60 mm Hg, mechanical ventilation, progressive infiltrates, septic shock, and renal failure. Although the ATS criteria demonstrated good sensitivity (98%) but low specificity (32%), the positive predictive value was low (24%). Therefore, a modified ATS score (mATS) was suggested, in which 2 or 3 “minor” baseline criteria (systolic blood pressure <90 mm Hg, multilobar involvement, and PaO2/FiO2 <250) and 2 “major” criteria (mechanical ventilation and presence of septic shock) demonstrated a sensitivity of 78%, a specificity of 94%, a positive predictive value of 75%, and a negative predictive value for mortality of 95% [68].
Fine et al. [69] developed the PSI derived from a large database that used points assigned for age, underlying disease, physical findings, and laboratory data to identify patients who required admission to the hospital and those at risk for death. of the risk groups (I–V), patients with the highest PSI scores (>90) in risk groups IV and V were considered to have “severe CAP” and had the highest mortality (8% and 29%, respectively). It is important to note that these studies included patients evaluated in the ED who had CAP as well as HCAP, and, thus, these assessments should also apply to those with HCAP.
The PSI is widely used as a benchmark for assessing the need for hospital admission and risk of mortality. Because the PSI requires 20 variables, it is labor intensive, difficult for clinical assessment in the ED, not a good predictor for ICU admission, and heavily weighted by age. Therefore, it may underestimate severe CAP in younger patients and is limited for identifying patients eligible for activated protein C therapy. As a result, other scoring systems have been evaluated.
A more recent approach using a modified PSI score was suggested by Espana et al. [73], who evaluated 1057 patients for CAP in an ED; 11.5% of patients were admitted to the hospital, 3% were admitted to the ICU, 2.3% had shock, and 1.5% underwent mechanical ventilation. Overall mortality was 9.1%. Severe CAP was defined as a score of >10 points for 2 major criteria, pH <7.3 (13 points) and systolic blood pressure <90 mm Hg (11 points), and for 6 minor criteria: respiratory rate >30 breaths/min (9 points), blood urea nitrogen (BUN) level >30 mg/dL (5 points), change in mental status (5 points), PaO2/FiO2 (5 points), age >80 years (5 points), and multiple bilateral infiltrates (5 points). of note, the scores >20 points were better predictors of severe CAP.
Another scoring system for stratifying the severity of CAP is the CURB-65, which awards 1 point each for confusion, urea concentration >7 nmol/L, respiratory rate 30 breaths/min, low blood pressure, and age 65 years [74]. Those with 3 points have been found to have a mortality rate of 21%, those with 4 points a mortality rate of 42%, and those with 5 points a mortality rate of 60% [74]. This scoring system also correlated with the need for mechanical ventilation, length of stay, and PSI score. Because the blood urea concentration is often not readily available, it was omitted, and the CRB-65 was suggested; it was easier to apply in the ED. Mortality rates for the CRB-65 correlated well with those for the CURB-65 and were 19% for patients with 2 points, 44% for those with 3 points, and 55% for those with 4 points. Advantages of these methods of scoring include simplified calculation, clear admission criteria for those with 2 points, good correlation with both mortality and the need for ICU admission for those with 2 points, a good predictor of CAP mortality due to bacteremia, and a good predictor of those in need of activated protein C therapy [67].
Data on the use of these scoring systems (PSI-V; PSI-IV,V; CURB-65 3; and the mATS score for assessing ICU admission and mortality) are shown in table 5 [66]. Note that, for assessing ICU admission, sensitivity ranged from 48% to 92%, specificity ranged from 45% to 87%, the positive predictive value ranged from 10% to 33%, and the negative predictive value ranged from 95% to 99%. Better results were identified for assessing mortality.
Outcomes of severe community-acquired pneumonia in terms of mortality and intensive care unit (ICU) admission.
Outcomes of severe community-acquired pneumonia in terms of mortality and intensive care unit (ICU) admission.
There is great heterogeneity of patient populations, definitions, etiologic agents, and clinical assessments for severe CAP. Is severe CAP best defined as admission to the ICU, the need for mechanical ventilation, or a poor score as defined by PSI, CURB-65, CRB-65, or modified PSI? Admission to the ICU varies between hospitals, as does the need for mechanical ventilation, methods of diagnosis, and etiologic agents. In addition, there is no assessment for MDR pathogens in patients with severe CAP similar to HCAP, and there is no assessment for patients in nursing homes or patients receiving long-term care or who have the use of prior antibiotics as a risk factor.
Many studies defining severe CAP lack validation and have small study populations and variable study definitions, as well as variable populations and risks for infection and mortality. In addition, data for severe CAP may be confounded by the presence of sepsis.
Evidence: What is HCAP?
According to the ATS-IDSA guidelines, patients with HCAP are a subset of patients who present to the hospital with pneumonia and are at greater risk of colonization and infection with MDR pathogens (figure 2) [3]. Previously, many of these patients who presented to acute-care facilities for evaluation of lower-respiratory-tract infections were considered to have CAP. Risk factors for infection with MDR pathogens identified in the ATS-IDSA guidelines are summarized in table 6. MDR and non-MDR pathogens of concern include MRSA, P. aeruginosa, Acinetobacter baumannii, and ESBL-producing gram-negative bacilli, such as E. coli, K. pneumoniae, and Enterobacter species (table 7). Patients with HCAP managed in nursing homes or residential facilities, which may have limited resources for evaluation and treatment of pneumonia, may need a referral to another facility. These nonhospital settings also could employ an approach that includes assessment, empirical antibiotic therapy based on a clinical suspicion of HCAP, and subsequent monitoring of response to therapy [64, 65, 75].
Risk factors for infection with multidrug-resistant (MDR) pathogens.
Risk factors for infection with multidrug-resistant (MDR) pathogens.
Initial antibiotic therapy for patients at risk of infection with multidrug-resistant (MDR) health care—associated pneumonia (HCAP) pathogens versus those without risk factors for infection with MDR pathogens, as outlined in the American Thoracic Society—Infectious Diseases Society of America guidelines.
Initial antibiotic therapy for patients at risk of infection with multidrug-resistant (MDR) health care—associated pneumonia (HCAP) pathogens versus those without risk factors for infection with MDR pathogens, as outlined in the American Thoracic Society—Infectious Diseases Society of America guidelines.
The reason for identifying patients with HCAP versus CAP was based on increasing rates of exposure to and colonization with MDR pathogens in patients with a greater risk of underlying disease and prior exposure to antibiotics that could increase the risk of receiving inappropriate empirical antibiotic therapy and having poorer outcomes (table 7) [3]. The use of broader initial antibiotic coverage for MDR pathogens was coupled with an emphasis on de-escalating and reducing the duration of antibiotic therapy for HCAP, HAP, and VAP (figure 6). If, indeed, broader initial empirical antibiotic therapy was appropriate for MDR pathogens, then de-escalation of the initial therapy was based on the response of the patients and the availability of microbiological data within 24–48 h.
Antibiotic options for patients with suspected health care—associated pneumonia (HCAP) who are referred to the emergency department or a clinic. MDR, multidrug resistant. Adapted from the American Thoracic Society/Infectious Diseases Society of America guidelines [3], with permission from the American Thoracic Society.
Antibiotic options for patients with suspected health care—associated pneumonia (HCAP) who are referred to the emergency department or a clinic. MDR, multidrug resistant. Adapted from the American Thoracic Society/Infectious Diseases Society of America guidelines [3], with permission from the American Thoracic Society.
If there were no risk factors for infection with MDR pathogens, patients with HCAP would be managed like patients with CAP, which would cover the usual CAP pathogens, versus the broader, empirical coverage recommended for MDR pathogens.
No recommendations were made in the ATS-IDSA guidelines for altering empirical antibiotic coverage for HCAP on the basis of disease severity [3]. Furthermore, the focus of these recommendations was on patients referred to acute-care hospitals for evaluation. The resources for the evaluation and management of severe CAP may be limited in some nursing homes or long-term-care facilities. Thus, the question arises as to whether and how the ATS/IDSA guidelines for HCAP apply to management in nursing homes and long-term-care facilities. Options for management would include collecting data for documenting the presence of HCAP, initiating empirical antibiotic therapy, and assessing the clinical response to therapy on site or making a referral if HCAP becomes severe. If there is no clinical improvement or progression while receiving the initial antibiotic regimen in the long-term-care setting, the patient would then be eligible for referral to an acute-care hospital or clinic for evaluation (figure 7).
Management strategies for health care—associated pneumonia (HCAP) for residents in nursing homes or other long-term-care facilities. Possible management options include “on-site” management versus referral to acute-care hospitals for evaluation. IV, intravenous; PO, by mouth.
Management strategies for health care—associated pneumonia (HCAP) for residents in nursing homes or other long-term-care facilities. Possible management options include “on-site” management versus referral to acute-care hospitals for evaluation. IV, intravenous; PO, by mouth.
Community-acquired MRSA was first seen in the 1990s in children and more recently has occurred in adults [76]. This strain is distinct from the hospital MRSA associated with HAP, VAP, and HCAP, because it has a mec IV gene and the Panton-Valentine leukocidin gene, which may account, in part, for its increased virulence and predisposition to abscess formation and severe pneumonia. Outbreaks have occurred in nursing homes and long-term-care facilities and have now been identified in hospitals. In comparison with hospital-acquired MRSA, community-acquired MRSA is more sensitive to antibiotics such as ciprofloxacin and clindamycin.
HCAP questions of concern. Several questions need attention. Can and should severe HCAP be managed in nursing homes or long-term-care facilities? Do the HCAP time lines for prior antibiotic use and prior hospitalization apply to management, or do they need to be altered? Are the HCAP definitions accurate? Should the severity of disease alter initial antibiotic management for patients with HCAP? How does the rapid evolution of community-acquired MRSA in nursing homes, long-term-care facilities, and hospitals alter the HCAP recommendations?
Grading of Evidence
On the basis of a review of the studies cited above, the 5 members of this workshop agreed that the evidence available to support this statement was category III for the statement in general, category III for the statement as it applies to hospitalized patients with HCAP, and category V for the statement as it applies to nonhospitalized patients with HCAP (table 2).
Level of Support
When voting on the support for this statement, 91% of the summit members and 83% of the IDSA members who responded to the survey accepted it completely; it was accepted with some reservations by 9% of the summit members and 13% of the IDSA members. One percent of IDSA members accepted the statement with major reservations, 2% rejected the statement with reservations, and 1% rejected it completely (figure 8).
Voting comparison for statement 5 (“Severe CAP is not HCAP”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. CAP, community-acquired pneumonia; HCAP, health care—associated pneumonia.
Voting comparison for statement 5 (“Severe CAP is not HCAP”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. CAP, community-acquired pneumonia; HCAP, health care—associated pneumonia.
Discussion
The management of pneumonia is dynamic, and the evolution of MDR pathogens in community and health care settings will require frequent refining and careful monitoring. The statement that severe CAP is not HCAP is based on multiple factors, including the lack of a consensus on the definitions of severe CAP, the clinical heterogeneity of patients, the pathogens, and the lack of validation of scoring systems. Other factors include the lack of assessment for MDR pathogens, patient residence in nursing homes, prior antibiotic therapy, outcomes related to appropriate antibiotic therapy, and confounding by other conditions, such as sepsis syndrome.
The definitions and data presented support the concept that severe CAP is not HCAP. Although the approaches to diagnosis and the management principles are similar, the definition of HCAP is focused on the risk factors for infection with MDR pathogens that may alter initial therapy for pneumonia; the definition of severe CAP is based on severity of disease that may be caused by a wide variety of non-MDR pathogens. Several scoring systems have been used to identify patients who may need more clinical attention or intensive care and have a greater risk of mortality. Some patients with severe pneumonia who present to a clinic or hospital ED may have originally received diagnoses of CAP but, with new definitions, are categorized as having HCAP because of their risk for infection with MDR pathogens. HCAP may share similar management principles with CAP, including antibiotic de-escalation and duration of therapy. However, the spectrum of HCAP outside of the ATS-IDSA guidelines and the rapid evolution of community-acquired MRSA in the community may have had an impact on the spread of MDR isolates in the community outside of the current concepts of HCAP.
Future Directions
Future directions discussed by the summit members reflected many needs, including better-designed epidemiologic studies, more-rigorous definition of terms, improved epidemiologic and microbiological criteria, and better-standardized diagnostic and laboratory criteria. Better data for HCAP are also needed in the acute-care setting versus various long-term-care settings in terms of epidemiology, diagnosis, management, and short- and long-term outcomes. There is a great deal to learn, and work is needed to improve our databases and the current guidelines for both prevention and therapy for specific at-risk patient populations. Ideally, observational, multicenter cohort studies using clear definitions and optimal data collection and analysis are needed. Also, patients should be followed longitudinally to assess changes in colonization with MDR pathogens over time.
Statement 6: Initial Empirical Therapy for HCAP Is the Same as That for HAP
Rationale and Definition of Statement
Many issues in addition to antibiotic choice enter into the decision making regarding initial empirical therapy for HCAP. The current definition of HCAP includes a heterogeneous group of patients, with variability in features such as site of care (hospital or nonhospital), route of therapy (oral or intravenous), and risk factors for infection with MDR pathogens. Some patients are also at risk for infection with other organisms, such as Legionella and viruses, which can be seen in CAP more than in HAP. These organisms can be epidemic in certain nursing homes. Because of these varying patient characteristics, the initial empirical therapy needs to account for differences in treatment approaches to HCAP and HAP, allowing for the possibility that some subpopulations should be managed like patients with HAP, some like patients with CAP, and some with a hybrid approach. HCAP includes many patient populations, some of which have been extensively studied, such as those with NHAP; other populations, such as those undergoing hemodialysis and those recently hospitalized, are less well described.
When the ATS-IDSA guidelines suggested that HCAP be treated like HAP, with a focus on MDR pathogens, it was implied that the patients evaluated were those in the hospital who were treated with intravenous antibiotics. However, as discussed in this review, HCAP also includes patients who are not ill enough to require hospital admission, those who are not at risk for infection with MDR pathogens but are at risk for infection with CAP-associated pathogens, those who are treated orally, and those who prefer to be treated at home or in a nursing home, regardless of illness severity.
Methods
Studies of therapy were evaluated by searching PubMed. When the term “healthcare associated pneumonia” (399 articles) was combined with the term “antibiotic therapy” (201,780 articles) by use of the “AND” function, a total of 19 articles remained. The term “nursing home pneumonia” (452 articles) combined with “antibiotic therapy” yielded 47 articles. These searches were limited to adults, clinical trials, reviews, meta-analyses, or practice guidelines.
To broaden the search, the term “hemodialysis” (35,454 articles) was combined with the term “pneumonia” to yield 107 articles. Finally, the term “prior hospitalization” (5084 articles) was combined with the term “pneumonia” to yield 205 articles. Fourteen articles were deemed relevant to the statement. This database of articles was reviewed and cross-referenced to evaluate original studies of therapy for patient populations included within the definition of HCAP.
Evidence
Differences in the approach to therapy between HCAP and HAP. There are a number of differences between HCAP and HAP, making it likely that the initial empirical therapy for both illnesses will not always be the same. By definition, HAP occurs in the hospital and is treated in the hospital. However, HCAP can arise outside of the hospital or in patients from health care environments after they are admitted to the hospital, and it can be treated both out of and in the hospital. If patients are managed out of the hospital, therapy can be oral, as in the case of the quinolones, which have been highly effective as therapy for patients with NHAP managed both in the nursing home and in the hospital [19, 77, 78].
HCAP arising in patients in nursing homes has been effectively treated with oral quinolone therapy, and, in many instances, this approach has averted hospital admission. In one cluster-randomized trial of 680 patients >65 years of age with radiographically diagnosed pneumonia at 20 nursing homes, patients were randomized to receive either usual care or a clinical pathway [19]. The pathway allowed for oral therapy with 500 mg of levofloxacin daily in the nursing home as long as the patient was able to eat and drink, had an oxygen saturation of 92%, and had vital signs with a pulse of <100 beats/min, a respiratory rate of <30 breaths/min, and a systolic blood pressure of 90 mm Hg. When this pathway was used, only 10% of patients were hospitalized, compared with 22% who received usual care (P=.001 ), and there were fewer total hospital days and a cost savings of at least $1000 per patient. Mortality and functional status were similar in both groups.
Out-of-hospital care is also used for some patients with HCAP, because of individual preferences for avoiding admission to the hospital. Thus, the “hospital at home” can include intravenous medications and oxygen [79]. Many patients prefer to remain in the nursing home, receiving oral therapy for acute illness [80]. Patients undergoing hemodialysis can receive intravenous antibiotics at each dialysis appointment.
Other factors to consider in the approach to therapy are the associated mortality rates and related pathogens implicated in pneumonia. In a retrospective cohort study of a large US database of 59 acute-care hospitals involving 4543 patients, ∼22% (998) of patients received diagnoses of HCAP. The mean mortality rates were comparable for patients with diagnoses of HCAP (19.8%) and HAP (18.8%) (P>.05 ) and were statistically significantly higher than those among patients with diagnoses of CAP (P<.0001 ). The distribution pattern of pathogens varied among the 4 pneumonia subtypes; however, S. aureus was the primary organism identified. The incidence of MRSA infections (56.8%) was significantly higher among patients with diagnoses of HCAP compared with all other pneumonia subcategories, including patients with diagnoses of HAP (48.6%; P<.05 ). S. aureus was the only pathogen associated with significantly higher mortality rates (P<.0001 ).
Should empirical antibiotic choices be the same as for HAP for all patients with HCAP? The recommendation that empirical antibiotic choice be the same for HCAP as for HAP is based on the idea that both patient populations are at risk for infection with the same MDR pathogens. However, many studies of patients with NHAP have shown a high frequency of S. pneumoniae, atypical pathogens, and Legionella species, as well as viruses mandating a different approach to therapy than is common in HAP [81–83]. In some nursing-home epidemics, colonization of the drinking water has led to Legionella infection, whereas, in other nursing homes, viruses such as rhinovirus have affected up to 40% of hospitalized patients with pneumonia [81, 83]. In addition, although enteric gram-negative bacteria can be found in patients with NHAP, the risk of infection with these organisms is only present in some individuals. In one study of severe NHAP, only 17 of 88 patients had drug-resistant pathogens, and they were individuals who, in addition to severe illness, had a history of antibiotic therapy in the past 6 months, a poor functional status (defined by ADL scores), or both [22].
Perhaps the best data to address initial empirical therapy for HCAP are the findings from previous studies showing that therapies not recommended for HAP and MDR pathogens have been highly effective in patients with HCAP [27, 84]. Some data are older, but, even in current studies, not all patients with HCAP have required multiple antibiotics directed at MDR gram-negative bacteria and MRSA to achieve high rates of clinical success. In one study of 40 patients with mild-moderate NHAP treated in the nursing home, both oral ciprofloxacin and intramuscular cefamandole were effective and were associated with low mortality rates (6.5%), even though some patients had gram-negative bacteria in sputum samples [77]. In another study of 45 hospitalized patients, many with NHAP, intravenous ciprofloxacin was more effective than intravenous ceftazidime [78]. The experience with oral levofloxacin for therapy within the nursing home (nonsevere illness) was mentioned above [19].
In a prospective, double-blind, randomized study of 51 patients with HCAP, 23 received intravenous monotherapy with ertapenem, while 28 received intravenous therapy with cefepime; however, patients at risk for pseudomonal infection or severe illness were excluded. Even though nearly 80% of patients with HCAP had gram-negative bacteria, the favorable responses with both therapies were high (90% with cefepime and 75% with ertapenem) [84]. In a retrospective study of 104 patients with severe pneumonia, including 47 from nursing homes and the rest with CAP, the mortality rate was 57% for NHAP and 55% for CAP [27]. Although mortality was higher for inadequate therapy (OR, 2.6; P=.03 ), 47% of patients received monotherapy; the mortality was the same as with combination therapy. Common therapies included second- and third-generation cephalosporins, β-lactam/β-lactamase inhibitor combinations, and quinolones.
One study evaluated 63 patients with CAP who were hospitalized after outpatient antibiotic therapy failed. This group of patients might be categorized as having HCAP and might be suspected to be infected with MDR pathogens [85]. Patients were randomized to receive monotherapy with moxifloxacin or standard therapy, and both clinical failure rates (6% vs. 30%) and 28-day failure rates (6% vs. 21%) were lower for the quinolone monotherapy than for standard therapy. These positive results occurred even though some patients were infected with S. aureus (5 patients) and enteric gram-negative bacteria (3 patients).
Thus, in many studies, therapy for patients with HCAP has been heterogeneous, but therapies not recommended for HAP due to MDR pathogens, including monotherapy with quinolones, ertapenem, and cephalosporins, have been highly effective.
Grading of Evidence
In evaluating the nature of evidence to support the statement for all patients, the panel of 6 graded the evidence as category III. In evaluating the nature of the evidence for patients admitted to the hospital, the panel graded it as category III. In applying the statement to patients never admitted to the hospital, the panel graded the nature of the evidence as category I (2 votes), category III (1 vote), category IV (2 votes), and category V (1 vote) (table 2).
Level of Support
When all members of the summit voted on the initial statement, 9% accepted it with some reservations, 46% accepted it with major reservations, 36% rejected it with reservations, and 9% rejected it completely. of the 383 IDSA members who responded to the survey, 29% voted to accept it completely, 52% accepted it with some reservations, 7% accepted it with major reservations, 9% rejected it with reservations, and 3% rejected it completely (figure 9).
Voting comparison for statement 6 (“Initial empirical therapy for HCAP is the same as that for HAP”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. HAP, hospital-acquired pneumonia; HCAP, health care—associated pneumonia.
Voting comparison for statement 6 (“Initial empirical therapy for HCAP is the same as that for HAP”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. HAP, hospital-acquired pneumonia; HCAP, health care—associated pneumonia.
After presentation of the evidence, the members of the summit panel developed a new statement (“statement 6.5”): empirical MDR antibiotics should be given to patients with HCAP who have at least 2 of the 3 mentioned risk factors (severe illness [requiring mechanical ventilation or ICU care], prior antibiotic therapy [for >3 days in the preceding 6 months], and poor functional status [ADL score, >12.5]). Among the summit members, 73% accepted this statement with some reservations, 18% accepted it with major reservations, and 9% rejected it with reservations (figure 10).
Voting comparison for statement 6.5 (“Empirical MDR antibiotics should be given to patients with HCAP who have at least 2 of the 3 mentioned risk factors”). “Summit members” refers to the 11-member summit panel. HCAP, health care—associated pneumonia; MDR, multidrug resistant.
Voting comparison for statement 6.5 (“Empirical MDR antibiotics should be given to patients with HCAP who have at least 2 of the 3 mentioned risk factors”). “Summit members” refers to the 11-member summit panel. HCAP, health care—associated pneumonia; MDR, multidrug resistant.
Discussion
The evidence presented highlights the complexity of HCAP and its empirical therapy. For a number of reasons, HCAP should not always be treated the same as HAP. If similar therapy were routinely administered, it could needlessly overtreat some patients with an unnecessarily broad spectrum of antibiotics. Many of the studies demonstrated the efficacy of monotherapy regimens that would not be recommended for patients with HAP at risk for infection with MDR pathogens (ciprofloxacin, levofloxacin, cefepime, and ertapenem), and some studies showed efficacy of therapies that did not cover the presumed MDR pathogens in complex patients with HAP (efficacy was present in all studies without specific MRSA coverage). In addition, HAP therapy would not adequately treat some patients with HCAP who might be infected with community pathogens, such as Legionella species. Finally, the inclusion of outpatients in the HCAP definition requires that some patients receive oral therapy instead of intravenous therapy, unlike those already in the hospital with HAP. The members of the summit panel recognized these issues when only 50% of members accepted the initial statement with major reservations and 50% rejected it.
Although not all members of the panel could accept the original statement, there are certainly patients with HCAP who are likely to be infected with MDR pathogens and, thus, who would require therapy identical to therapy for HAP. These are patients who generally have multiple risk factors, such as severe pneumonia, prior antibiotic therapy, and poor functional status [22]. Thus, the panel voted on a second statement, that empirical therapy for MDR pathogens should be administered to patients with HCAP who have at least 2 of the 3 risk factors (severe illness, prior antibiotic therapy, and poor functional status); 70% of the summit panel accepted the statement with some reservations, and 20% accepted it with major reservations.
On the basis of the discussion of the panel and the available evidence, HCAP therapy should probably be divided into 2 categories: limited-spectrum therapy and broad-spectrum therapy. This categorization is dictated by whether the patient with HCAP has 2 of the 3 identified risk factors for MDR pathogens in this population. Limited-spectrum therapy can be given to patients in or out of the hospital who do not have 2 of these 3 risk factors. Therapy can be a respiratory quinolone alone (moxifloxacin or levofloxacin) or, alternatively, combined with a selected β-lactam (ceftriaxone, cefepime, piperacillin-tazobactam, or ertapenem) with good activity against drug-resistant S. pneumoniae, with consideration of adding a macrolide (especially if Legionella, Chlamydophila, or Mycoplasma species have been present in patients from the same environment). Patients who receive limited-spectrum therapy can be treated in the nursing home with oral therapy if a quinolone is used. Broad-spectrum therapy should be given to patients in or out of the hospital who do have 2 of the 3 identified risk factors, and these patients should receive therapy active against drug-resistant S. pneumoniae, P. aeruginosa (ideally with 2 agents), and MRSA, as well as consideration of Legionella in appropriate settings. This could be achieved with a β-lactam (cefepime, imipenem, meropenem, or piperacillin/tazobactam) combined with an antipseudomonal quinolone (ciprofloxacin or high-dose levofloxacin), with either linezolid or vancomycin. If a quinolone cannot be used, because of allergy, intolerance, or recent therapy in the past 3 months, an aminoglycoside should be added in its place while considering the addition of a macrolide to the β-lactam and the MRSA therapy. The antimicrobials used for broad-spectrum therapy are primarily available for parenteral and not oral administration.
Future Directions
These therapy recommendations are based on the best available data, which are, unfortunately, quite limited. Future validation of this approach is required. In addition, more data are needed for populations of patients with HCAP other than those with NHAP. As new therapeutic options become available, they should be tested in patients with HCAP specifically, so that data from patients with HAP do not need to be extrapolated to this population.
Statement 7: Patients with HCAP Who Are at Risk for Gram-Negative Infections Should Receive Dual Empirical Antibiotic Coverage
Rationale and Definition of Statement
The issue of monotherapy versus combination antibiotic therapy for serious infections is not new [86, 87]. Many clinicians utilize antibiotic combinations when serious gram-negative bacterial infections are suspected, whereas others criticize such an approach as lacking sound evidence [86, 87]. A number of issues need to be considered if empirical combination antibiotic therapy is considered. First, what are the consequences of inadequate empirical antibiotic therapy for the bacillary infection? If data exist showing that mortality or morbidity is seriously compromised by inadequate empirical therapy, it behooves clinicians to consider how they may improve empirical antibiotic regimens. For example, inadequate antibiotic therapy may have few consequences in a patient with an uncomplicated urinary tract infection but substantial consequences in a critically ill patient in an ICU who develops a bloodstream infection [88].
Second, is the adequacy of an empirical antibiotic regimen more likely to be improved by the use of a combination regimen than by the use of an improved single agent? Incorporated into this consideration are issues of synergy, development of resistance, and toxicity. Potentially, an advantage may arise when synergy is observed in vitro between 2 antimicrobial agents (although, for most infections, this has not been borne out in clinical studies). For some organisms (e.g., Mycobacterium tuberculosis or HIV), resistance may be less likely to develop when combinations of antimicrobials are used. Although there are some in vitro data suggesting that this may also occur with respect to treatment of gram-negative bacilli, there are also conflicting data showing no benefit to this approach. Finally, there is the potential that the use of 2 drugs (e.g., addition of an aminoglycoside to a β-lactam) may increase the risk of toxicity and, therefore, detract from the benefit of using combination therapy.
In this review, these issues are addressed with respect to HCAP. This is first approached from a review of randomized trials of therapy for patients with HCAP. Second, a review of epidemiologic studies was performed to determine whether patients with HCAP may be at increased risk for infection with MDR gram-negative bacilli, thereby necessitating the use of empirical combination therapy.
Methods
A search of PubMed was performed in November 2006, cross-linking articles with the key words “randomized trial” and “healthcare-associated pneumonia,” “nursing home pneumonia,” or “hospital-acquired pneumonia.” A second search was performed with the key words “microbiology” or “pathogen” and “healthcare-associated pneumonia,” “nursing home pneumonia,” or “hospital-acquired pneumonia.” The references of articles discovered in this search were reviewed for further articles pertinent to the topic. With regard to studies of patients with HAP, those with >50% patients receiving ventilation were excluded (unless there was a subgroup analysis of patients not receiving ventilation). Eleven articles were reviewed for this statement.
Evidence
Randomized trials. There are no randomized trials evaluating empirical monotherapy versus combination therapy for HCAP. However, there are several randomized trials evaluating 2 different monotherapy regimens or evaluating 1 combination therapy regimen versus another combination therapy regimen. Randomized trials evaluating 2 different monotherapy regimens have included those of ciprofloxacin versus ceftriaxone, ciprofloxacin versus ceftazidime, and ertapenem versus cefepime. The trial evaluating combination therapy was of tobramycin plus piperacillin/tazobactam versus tobramycin plus ceftazidime [89].
Ciprofloxacin monotherapy has been compared with ceftriaxone monotherapy in a small randomized trial of nursing home—acquired lower-respiratory-tract infection requiring hospitalization [90]. Fifty patients were enrolled, all of whom were 60 years of age. A successful outcome was defined as resolution or marked improvement in clinical signs and symptoms of lower-respiratory-tract infection and was achieved in 50% (12/24) of patients treated with ciprofloxacin and 54% (14/26) of patients treated with ceftriaxone. In a second small study evaluating ciprofloxacin monotherapy, 44 hospitalized patients with HAP or NHAP were randomized to receive ciprofloxacin or ceftazidime [78]. All (23/23) ciprofloxacin-treated patients had a favorable response, compared with only 71% (15/21) of the ceftazidime-treated patients.
The largest study of monotherapy was a 303-patient, prospective, double-blind, randomized, international, multicenter study of ertapenem versus cefepime [84]. Patients enrolled in the study had pneumonia acquired in a hospital or a skilled-care facility, such as a nursing home. Patients with conditions believed to increase the risk of P. aeruginosa or Acinetobacter infection were excluded. These conditions include pneumonia acquired in an ICU or while receiving mechanical ventilation, an immunocompromising illness or therapy, and cystic fibrosis. Fifty-four percent of the population had a pathogen isolated—this comprised Enterobacteriaceae, such as Klebsiella species or E. coli (in 20% of enrolled patients); S. pneumoniae (in 13%); S. aureus (in 12%); and P. aeruginosa (in 4%). A successful clinical response was observed in 92% of patients treated with ertapenem and 88% of patients treated with cefepime. A successful microbiological response was observed in 84% of patients treated with ertapenem and in 83% of patients treated with cefepime.
The trial evaluating combination therapy was a 300-patient, open-label, randomized, comparative, multicenter study of tobramycin plus piperacillin/tazobactam versus tobramycin plus ceftazidime [89]. The majority of patients had pneumonia, but 21% had bronchitis. Thirteen percent of patients had nursing home acquisition of their lower-respiratory-tract infection, with the remainder having hospital-acquired infections. The most commonly isolated pathogens in evaluable patients were H. influenzae (32 patients), S. aureus (31 patients), P. aeruginosa (22 patients), S. pneumoniae (21 patients), E. coli (16 patients), and K. pneumoniae (14 patients). Clinical success was observed in 74.2% of patients treated with piperacillin/tazobactam plus tobramycin, versus 57.9% of patients treated with ceftazidime plus tobramycin. Bacteriologic response was observed in 65% of patients treated with the piperacillin/tazobactam plus tobramycin regimen, versus 38% of patients treated with the ceftazidime plus tobramycin regimen (P=.03 ).
Observational studies of the microbiology of HCAP. A number of studies have evaluated the etiology of HCAP and, in some circumstances, compared the etiology with that of CAP. An 18-month prospective cohort study from the United Kingdom compared the etiology of NHAP and CAP [20]. S. pneumoniae was the predominant pathogen identified as being responsible for 55% of NHAP cases and 43% of CAP cases. Gram-negative bacilli were rarely isolated, and, in general, the etiologies of NHAP and CAP were similar. In contrast to these results, a database of 4543 patients with HCAP and a concomitant positive respiratory bacterial culture showed that S. aureus was responsible for 47% of cases, P. aeruginosa for 25%, Klebsiella species for 8%, and S. pneumoniae for 6% [2]. Etiologies of HCAP were more similar to those of HAP or even VAP than to those of CAP.
In a large review of 10,635 hemodialysis recipients with 3101 episodes of pneumonia, no organism was identified in 81.8% of patients [29]. Gram-positive organisms (predominantly S. pneumoniae) were found in 4.8% of patients, and gram-negative bacilli were found in 11.1% of patients. Almost 3% of the patients had P. aeruginosa.
Similarly, a study of NHAP requiring ICU admission showed that S. pneumoniae was a common pathogen [22]. However, MRSA and gram-negative bacilli were isolated quite frequently in this particular study. Enterobacteriaceae (E. coli; 9/135 patients), K. pneumoniae (6/135 patients), Serratia marcescens (5/135 patients), Enterobacter cloacae (5/135 patients), and Proteus mirabilis (3/135 patients) were found in 21% of patients, and P. aeruginosa was found in 7% of patients. Patients were excluded from this study if they were immunocompromised or if they had been hospitalized for >48 h in the 6 months before ICU admission.
Grading of Evidence
On the basis of a review of the studies cited above, the 5 members of this workshop agreed that the evidence available to support this statement was category V for the statement in general, a tie between categories III and IV for the statement as it applies to hospitalized patients with HCAP, and category V for the statement as it applies to nonhospitalized patients with HCAP (table 2).
Level of Support
When voting on the support for this statement in the group at large, 9% of the summit participants voted to accept it completely, 9% accepted the statement with some reservations, 64% accepted the statement with major reservations, and 18% rejected the statement with reservations. In comparison, of the 383 IDSA members who participated in the online survey, 25% accepted the statement completely, 33% accepted the statement with some reservations, 13% accepted the statement with major reservations, 23% rejected the statement with reservations, and 5% rejected the statement completely (figure 11).
Voting comparison for statement 7 (“Patients with HCAP who are at risk for gram-negative infections should receive dual empirical antibiotic coverage”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. HCAP, health care—associated pneumonia.
Voting comparison for statement 7 (“Patients with HCAP who are at risk for gram-negative infections should receive dual empirical antibiotic coverage”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. HCAP, health care—associated pneumonia.
Discussion
The question raised in this review is whether patients with HCAP should receive dual empirical antibiotic coverage aimed against gram-negative pathogens. There are no randomized trials that provide a simple answer to this question. Several studies of monotherapy antibiotic regimens have been performed. Monotherapy regimens (comprising ciprofloxacin, ceftriaxone, ceftazidime, cefepime, or ertapenem) were generally successful, both clinically and microbiologically. However, inclusion criteria, exclusion criteria, and outcome measures were variable. In some circumstances, success rates were only 50%, but it is unclear whether this was a marker of particularly stringent outcome definitions in these small studies or truly represented a deficiency in the monotherapy regimen. The optimal study design to answer the question of superiority of an empirical dual antibiotic regimen would be a randomized trial of monotherapy versus combination therapy (with the agent used in the monotherapy arm plus an additional agent). No such study exists or is currently under way.
The second way to approach this question is to review whether patients with HCAP are at high risk for infection with MDR gram-negative bacilli. Organisms such as P. aeruginosa or A. baumannii are often MDR; in some scenarios (e.g., in ICUs), no single antibiotic will cover >80%–85% of strains. Therefore, empirical use of combination therapy may be important to ensure that empirical therapy is likely to be microbiologically adequate. Studies of pneumonia in nursing homes in the United Kingdom and among hemodialysis recipients in the United States have not revealed a high frequency of organisms such as P. aeruginosa or A. baumannii as etiologic agents. The frequency of gram-negative bacilli, including MDR strains, appears to be higher among patients in nursing homes who require ICU admission. The study by Kollef et al. [2] suggests that patients with HCAP have etiologic agents (such as P. aeruginosa) more like those in hospitalized patients. What is not clear, however, is whether P. aeruginosa and other gram-negative isolates from patients with HCAP have resistance profiles of a magnitude similar to those in patients in the ICU.
It is likely that HCAP is actually quite a heterogeneous condition. Clearly, patients in nursing homes differ in their functional status, and this may, in turn, influence their likelihood of being colonized (and subsequently infected) with MDR gram-negative bacilli. Patients with recent hospitalization may also vary in the likelihood that they are colonized with antibiotic-resistant gram-negative bacilli. In turn, hemodialysis recipients may be at different risk, compared with patients in nursing homes or previously hospitalized patients.
Future Directions
Like other forms of pneumonia, HCAP varies in severity. The consequences of inadequate antibiotic therapy for patients with severe HCAP requiring ICU admission are likely to be greater than those for patients admitted to a general ward or even managed in a nursing home or some other setting outside of the hospital. It is unlikely that a randomized trial will directly answer the question of whether empirical combination therapy is optimal for patients with HCAP. Attention is needed to determine which patients are at risk for infection with MDR gram-negative bacilli, and whether inadequate empirical therapy influences outcomes for all, or subsets of, patients with HCAP. Only then will the question of optimization of gram-negative coverage for patients with HCAP be answerable.
Statement 8: Patients Should Receive Initial Empirical Therapy That Covers MRSA at the Time of HCAP Diagnosis
Rationale and Definition of Statement
MRSA infections, a common problem encountered by clinicians, result in significant morbidity, mortality, and economic consequences. The annual incidence of MRSA has profoundly increased throughout ICUs in the United States, rising 3% on average since 1992 to an estimated 65% of all S. aureus infections in 2004 [91]. Unfortunately, the dissemination of MRSA is not limited to the ICU. Data reported by the National Nosocomial Infections Surveillance Systems indicate that, on average, 31% of outpatient and 46% of non-ICU inpatient isolates of S. aureus are MRSA [92]. As a pulmonary pathogen, S. aureus accounts for 20%–30% of HAPs and VAPs, with MRSA accounting for >50% of these infections, especially in patients with a high severity of illness and those with prior antibiotic exposure [93]. MRSA pneumonia has been associated with longer hospital stays and higher costs, compared with methicillin-susceptible S. aureus (MSSA) pneumonia, regardless of severity of illness [94–96]. The issue of attributable mortality due to MRSA pneumonia compared with MSSA pneumonia is difficult to resolve, because of the high rates of inappropriate initial antimicrobial therapy in many cases of MRSA pneumonia. However, MRSA doubles the attributable mortality compared with MSSA in patients with bacteremia [97]. Given the extraordinary burden of illness caused by MRSA pneumonia in hospitalized patients, this section reviews the level of evidence supporting empirical coverage of MRSA in patients with diagnoses of HCAP.
Methods
A PubMed database search to identify microbiological features and clinical outcomes in patients with HCAP was conducted. The search terms “methicillin-resistant Staphylococcus aureus” and “health care associated” were combined, using the “AND” function, to yield 141 articles. This search was combined with the term “pneumonia” by use of the “AND” function, producing 123 articles that were reviewed and selected. Eleven articles were deemed relevant to the statement.
Evidence
The microbiological etiology of HCAP in the era of increasing MRSA incidence has been an understudied subject and has generally been relegated to the subgroup of patients who require hospital admission. A retrospective analysis of the Cardinal Health-Atlas Research Database characterized the microbiology and outcomes of 4534 cases of pneumonia identified by International Classification of Diseases, Ninth Revision codes. HCAP was distinguished in this study as a positive respiratory culture result within 48 h after hospital admission in patients who were transferred from another health care facility, were receiving long-term hemodialysis, or had been hospitalized in the previous 30 days and did not require mechanical ventilation [2]. Although this definition is imprecise and likely does not capture the true HCAP population, 21.7% (n=988 ) of the sample population was designated with this classification of pneumonia. Within this stratum, the pathogen most frequently isolated was MRSA (26.5%), followed by P. aeruginosa (25.3%) and MSSA (21.1%). Interestingly, S. aureus isolation in all pneumonias was associated with increased in-hospital mortality (OR, 1.58; 95% CI, 1.32–1.89; P<.0001 ). The authors hypothesized that this finding might reflect clinicians' lack of precision in differentiating HCAP from CAP, resulting in omitted coverage for MRSA and inappropriate empirical coverage. A prospective, randomized, multicenter comparative trial of 2 antibiotics lacking activity against MRSA sought to determine the microbiological etiologies in patients not receiving ventilation who were admitted to the hospital with HCAP or HAP [84]. In this heterogeneous sample, only 12% of all isolates were positive for S. aureus, 40% of which were MRSA.
NHAP includes a select group of patients who fall under the umbrella of HCAP. The isolation of MRSA in this patient population, compared with patients with CAP, has been variable. A prospective study comparing the clinical and microbiological characteristics of NHAP with those of CAP requiring hospitalization in the United Kingdom from 1998–1999 revealed only 1 case of pneumonia caused by S. aureus (not differentiated between MRSA and MSSA) [20]. A prospective study comparing patients with NHAP and patients with CAP at a single center in the United States from 1996 to 1999 found that patients with NHAP were more likely to be infected with S. aureus than were patients with CAP (29% vs. 7%). However, the number of patients with MRSA was small (3 vs. 0 cases, respectively) [27]. In contrast, results of BAL cultures in a group of patients with severe NHAP who were admitted to the ICU at a single US hospital between 1998 and 2003 showed S. aureus to be the most common pathogen identified (23.9%), of which 61.9% were MRSA [22]. Clearly, microbiological data from patients with NHAP are limited by the period during which they were obtained. In fact, it could be argued that differences exist in the frequency of MRSA isolation in these studies compared with today. However, outside of the single retrospective study mentioned above, evidence is lacking.
Given the inconsistencies in the frequency with which MRSA is thought to cause HCAP, attention must be focused on identifying patients at risk for MRSA infection at admission to the hospital or in the nursing home. A detailed history of the patient's recent contact with health care systems is the most important component in the categorization of MRSA infection risk. A prospective surveillance study of MRSA infections found that, of 123 positive culture samples obtained within the first 48 h of hospitalization, only 1 was from a patient who did not have recent health care contact, including hospitalization, transfer from another health care facility, residence in a long-term-care facility, dialysis, home nursing care, or day surgery [24]. Similarly, a case-control study comparing patients with blood cultures positive for MRSA and patients with positive blood cultures without MRSA in the first 24 h of hospitalization found that all patients with MRSA had recent health care exposure [98]. Risk factors for community-acquired infection with health care—associated MRSA were elucidated in a prospective, case (MRSA)—control (MSSA) study conducted at a French teaching hospital [30]. Among patients with respiratory tract (27.3%), urinary tract (17.2%), primary bloodstream (9.8%), and skin/soft tissue (38.5%) infection, risk factors associated with MRSA infection at hospital admission included home nursing care (adjusted OR [AOR], 3.7; 95% CI, 2.0–6.7), prior hospitalization (AOR, 3.8; 95% CI, 1.8–7.9), transfer from another hospital or nursing home (AOR, 2.3; 95% CI, 1.2–74.3), age 65 years (AOR, 1.8; 95% CI, 1.1–2.5), and home nursing care or inpatient surgery in the past 3 years (AOR, 3.1; 95% CI, 1.2–8.0). Additionally, previous use of antibiotics has been linked to infection with drug-resistant bacteria, including MRSA in patients with severe NHAP [22]. These findings suggest that patients presenting from the community with pneumonia who have recent health care exposure or antibiotic use should be considered at risk for MRSA HCAP.
The ATS-IDSA guidelines for the empirical treatment of HCAP indicate that coverage for MRSA with either vancomycin or linezolid should be instituted [3]. However, the impact of this recommendation on clinical outcomes, including mortality, is limited. In general, omission of antibiotic coverage with MRSA activity has not been associated with poor outcomes in patients with NHAP [19, 84, 85]. This finding may be a result of the low incidence of MRSA in these specific study populations. A retrospective study of MRSA sterile-site infections at a single institution found that appropriate empirical therapy was significantly more likely to be prescribed to patients who had MRSA isolated 48 h after admission (hospital acquired, 39%), versus those with MRSA isolated within the first 48 h after admission (health care associated, 22%) (P<.001 ) [99]. This was despite the finding that 86% of the patients with positive culture results within 48 h of hospitalization had recent health care exposure. Subsequent multivariate regression analysis found inappropriate initial empirical therapy (omission of MRSA coverage) to be an independent predictor of hospital mortality in this cohort (AOR, 1.92; 95% CI, 1.48–2.50). This suggests that a thorough assessment of health care exposure was not undertaken in this population and may have had an impact on patient outcomes.
Grading of Evidence
On the basis of this literature review, 6 members of the HCAP Therapeutic Intervention workshop voted that the nature of the evidence for the statement ranged from category II to V for all patients, from category II to IV for patients admitted to the hospital, and from category II to V for patients never admitted to the hospital. Therefore, the workshop voted that there was poor evidence to support the statement for all patients and fair evidence to support the statement for hospitalized patients; there was a range of votes from poor evidence to support to good evidence to reject the statement for nonhospitalized patients (table 2).
Level of Support
Overall, 9% of the summit participants voted to accept the statement with some reservations, 55% voted to accept the statement with major reservations, and 36% voted to reject the statement with reservations. In comparison, of the 383 IDSA members who participated in the online survey, 25% accepted the statement completely, 37% accepted the statement with some reservations, 13% accepted the statement with major reservations, 2% rejected the statement with reservations, and 5% rejected the statement completely (figure 12).
Voting comparison for statement 8 (“Patients should receive initial empirical therapy that covers MRSA at the time of HCAP diagnosis”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. HCAP, health care—associated pneumonia; MRSA, methicillin-resistant Staphylococcus aureus.
Voting comparison for statement 8 (“Patients should receive initial empirical therapy that covers MRSA at the time of HCAP diagnosis”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. HCAP, health care—associated pneumonia; MRSA, methicillin-resistant Staphylococcus aureus.
Discussion
This statement is of great significance, given the increasing incidence of MRSA in hospitals and the community. Currently, the microbiological evidence supporting MRSA as a major pathogen in HCAP is limited. Similarly, outcome data supporting this statement are nonexistent. Evidence to the contrary has been published related to NHAP, such that omitting MRSA coverage in the therapeutic regimen did not appear to have adverse consequences. One study evaluating patients with health care—associated MRSA sterile-site infection in comparison with patients with hospital-acquired MRSA infection found the former group to be significantly less likely to have MRSA coverage initiated in the first 24 h, which was subsequently associated with increased odds of in-hospital mortality.
Future Directions
Until prospective controlled trials are published that specifically study microbiology and outcomes in patients with HCAP treated with and without antibiotics that have MRSA activity, support of this statement is based on opinion.
Statement 9: When Microbiological Data Are Unavailable, De-Escalation in Patients with HCAP Should Not Occur
Rationale and Definition of Statement
“De-escalation” can mean discontinuation of therapy with some antibiotics, changing from a broad-spectrum to a narrow-spectrum antibiotic, or discontinuation of all antibiotic therapy. The concept of de-escalation arose from the treatment of patients with HAP, especially those with VAP, and was proposed as a method to minimize exposure to broad-spectrum antibiotics [61, 100, 101]. Shorter courses of antibiotic therapy are associated with lower rates of superinfection with resistant bacteria [48, 101]. However, some pathogens seem to require longer treatment to avoid clinical relapse [102]. Because of this discrepancy, the decision to de-escalate may require microbiological data. This concept is included in the ATS-IDSA guidelines for treating HAP [3].
Clinicians often de-escalate treatment only on the basis of clinical response to therapy. The safety and relapse rates associated with this approach are not known.
Methods
A PubMed search was performed in November 2006. By looking at “duration of antibiotic therapy pneumonia” and limiting the search to clinical trials published in English, 180 references were identified. By looking specifically at “health care associated pneumonia duration of therapy,” 14 references were identified; all were regarding VAP. Searching with the term “nursing home pneumonia duration of therapy” identified 23 references.
References from the last 2 categories were reviewed. The majority of references simply provided the duration of therapy and made no attempt to analyze why one duration was used versus another. There were some exceptions, and 13 studies were reviewed.
Evidence
The most direct evidence regarding duration of antibiotic therapy has been for cases of VAP. Although this is a subset of nosocomial pneumonia, the information was believed to be relevant for several reasons. First, it provides a worst-case scenario, since the rate of mortality due to VAP is higher than that reported for HCAP. However, the mortality rate for HCAP is closer to that for VAP and HAP than to that for CAP [2]. In addition, the microbiology information from invasive procedures done for VAP were felt to be more revealing than those done for most cases of HCAP. In particular, semiquantitative cultures from bronchoscopic and nonbronchoscopic BAL samples were often used in VAP studies. Current opinion is that such semiquantitative information is more accurate than information from nonquantitated samples, such as sputum [103].
One study directly compared 2 treatment durations in patients with VAP. All patients had undergone diagnostic BAL and had received initial, appropriate, adequate antibiotics to remain in the study. In a randomized trial of 8 versus 15 days of therapy for VAP, those patients who had nonfermenting gram-negative rods (e.g., P. aeruginosa) were more likely to relapse if treated for only 8 days [48]. Although the rate of relapse was not statistically significantly different, the 60% higher relapse rate has led some physicians to require microbiological data before stopping antibiotic therapy at 1 week.
In another study examining the clinical outcome of S. aureus versus other pathogens in serial nonbronchoscopic BAL studies of VAP, it was determined that patients with S. aureus and drug-resistant gram-negative rods were likely to have persistent bacteria in the BAL sample 2–5 days into therapy [102]. The authors found that patients with >1000 cfu of bacteria/mL of BAL sample in the follow-up BAL had a significantly higher 28-day mortality rates than did those who cleared the bacteria [102].
There is no similar microbiological information for HCAP. Guidelines have been established for treating patients in nursing homes who have suspected pneumonia [104], and these led to a reduction in the number of hospitalizations and overuse of antibiotics but did not change mortality rates [105].
A cluster-randomized controlled trial of 680 nursing-home residents was performed over a 16-month period in 22 nursing homes in Canada. Patients who met a predefined criterion of pneumonia based on clinical grounds were treated with either standard care or a clinical pathway, which included use of oral antimicrobials, portable chest radiographs, and oxygen-saturation monitoring. Although 76 (22%) of 353 residents receiving standard care for their pneumonia required hospitalization, only 34 (10%) of 327 of those in the clinical pathway were hospitalized. There also was a reduction in the number of hospital days for those in the clinical pathway group who were admitted, compared with the standard care group. Overall, health care costs were significantly reduced, with no difference in mortality [19]. The clinical pathway did not have a specific de-escalation procedure.
Another study used the clinical status of the patient to separate 170 episodes of pneumonia in nursing home patients into 4 broad categories: pneumonia, aspiration pneumonitis with infiltrates that resolve within 24 h, aspiration pneumonitis with infiltrates that persist for >24 h, and aspiration without pneumonitis [106]. Patients were categorized prospectively on the basis of their initial presentation and follow-up evaluation at day 3–5. The results are summarized in table 8. The authors found that patients believed to have aspiration pneumonitis who did not have infiltrates after day 3 were treated for a shorter time and were less likely to receive antibiotics at discharge [106]. A major problem with this study was that treatment was not directed by any protocol. Many of the patients without evidence of pneumonia were still treated with antibiotics in this study.
Comparison of pneumonia and aspiration pneumonitis.
Comparison of pneumonia and aspiration pneumonitis.
The ATS-IDSA HAP guidelines have focused attention on the reevaluation of the patient at day 3. In addition to the microbiological data, clinical criteria are useful. These include variations of the CPIS. Patients whose conditions respond to therapy for VAP will have a decrease in their CPIS by day 3, whereas those who die have no change or have an increase in the score [39]. The major reason for the decrease in the CPIS was improvement in the PaO2/FiO2 ratio [39]. Other physicians have discontinued antibiotic therapy as soon as day 3 if the CPIS has remained stable or has decreased [101].
These observations have led to the suggestion that the use of broad-spectrum antibiotics for possible MDR pathogens can be modified or even discontinued at days 3–7 on the basis of clinical criteria alone. This is especially true when a CXR film shows resolution of the infiltrate and the clinical status of the patient is improving by day 3.
Grading of Evidence
On the basis of a review of the studies cited above, the 6 members of this workshop agreed that the evidence available to support this statement was category V for the statement in general, category V for the statement as it applies to hospitalized patients with HCAP, and category IV for the statement as it applies to nonhospitalized patients with HCAP (table 2).
Level of Support
As shown in figure 13, 9% of the summit members accepted the statement with major reservations, 82% rejected it with reservations, and 9% rejected it completely. In contrast, 54% of the 383 IDSA members who completed the online survey believed there was some degree of support for the statement. However, this may have been based on the assumption that what was true for VAP would apply to HCAP. Thirty-five percent rejected the statement with reservations, and 1% rejected it completely.
Voting comparison for statement 9 (“When microbiological data are unavailable, de-escalation in patients with HCAP should not occur”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. HCAP, health care—associated pneumonia.
Voting comparison for statement 9 (“When microbiological data are unavailable, de-escalation in patients with HCAP should not occur”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. HCAP, health care—associated pneumonia.
Discussion
The summit panel recommended that broad-spectrum antibiotic therapy for possible MDR pathogens could be modified or even discontinued at days 3–7 on the basis of clinical criteria alone. This is especially true when a CXR film shows improvement. Although this recommendation was not supported by strong evidence, the current information would support this conclusion.
The major limitation of this analysis was the infrequency with which an adequate lower respiratory culture was obtained from patients with HCAP. Because of this, the clinician was left with the dilemma of making a decision on the basis of the clinical presentation of the patient. Although this appears to be common practice, there is little scientific evidence to support it.
The studies to date do suggest that de-escalation can be done in HCAP when the patient meets certain clinical criteria, such as a clear CXR film and return to baseline respiratory status.
One caveat is that most of this information was based on patients with probable susceptible microbacteria. The VAP data would suggest that de-escalation may not be possible in cases caused by MDR bacteria. This would be especially true if the patient does not receive adequate initial antibiotic therapy.
Future Directions
There is a need for specific studies of de-escalation in HCAP cases. It would be useful to obtain microbiological data in these cases. However, studies without microbiological information could still be informative if the reasons for de-escalation were clearly defined before the study. Outcomes could then be tested against those obtained with standard therapy.
Identifying risk factors for truly resistant bacteria, which may require longer antibiotic therapy, is also an issue. These bacteria could include MRSA and P. aeruginosa. Clearly, not all cases of HCAP are the same, and the approach for de-escalation may need differ among groups.
Statement 10: The Duration of Antibiotic Therapy for Patients with HCAP with a Clinical Response Should Be 7 Days
Rationale and Definition of Statement
It is widely accepted that appropriate antimicrobial stewardship includes optimal selection, dose, and duration of treatment, as well as control of antibiotic use. It is anticipated that appropriate antimicrobial stewardship will prevent or slow the emergence of resistance among microorganisms [107]. However, it is distressing that there are few data on the optimal duration of antibiotic therapy for many infectious diseases, including HCAP.
The recent ATS-IDSA nosocomial pneumonia guidelines define HCAP as a distinct clinical entity occurring in a subset of patients at risk for harboring resistant organisms despite their residence in the community [3]. These patients have historically been treated with antibiotic regimens recommended in CAP guidelines. As the prevalence of antimicrobial resistance has increased in patients meeting HCAP criteria, many clinicians questioned whether these antibiotic regimens were appropriate. The present review aims to ascertain whether evidence exists and to assess the strength of that evidence supporting the assertion that the duration of antibiotic therapy for patients with HCAP who have a clinical response should be 7 days.
Methods
A PubMed database search to identify studies related to the duration of treatment of HCAP was completed on 31 October 2006. The search terms “pneumonia” and “health-care associated” resulted in 79,547 and 46,168 articles, respectively. Combining the 2 terms yielded 632 articles. The text word “antibiotic” produced 410,820 articles. Combining the above searches gave a total of 157 articles. When these searches were combined with the term “duration,” 23 articles remained. When the search was limited to English, 20 articles were reviewed, of which 5 were deemed relevant.
Evidence
Although no study was identified that specifically focused on the optimal duration of therapy for HCAP, these 5 studies provided insight into the optimal duration of therapy for hospitalized adult patients with VAP.
Dennison et al. [47] performed a prospective cohort study of 27 adult patients with VAP receiving appropriate antibiotic therapy. The primary objective was to define the time to resolution of VAP symptoms after initiation of antibiotics. They observed that response to antimicrobial therapy for VAP occurs within the first 6 days of therapy. However, colonization with resistant pathogens occurs after 6 days, and colonization with resistant gram-negative bacteria will persist in many patients.
Singh et al. [101] conducted a prospective and randomized but unblinded study of 81 patients with VAP. The goal was to devise an operational approach for patients with possible nosocomial pneumonia. Only patients with a CPIS of 6 were included in the study and were randomized to receive standard therapy, as determined by the clinician, or a 3-day course of ciprofloxacin. Mortality, extrapulmonary infections, and the number of patients who developed a CPIS of >6 at 3 days did not differ. However, antimicrobial resistance and/or superinfections were encountered significantly less frequently in the patients treated for 3 days.
Ibrahim et al. [108] investigated the impact of a guideline that incorporated de-escalation, by use of a cohort design including 50 patients treated before implementation of the guideline and 52 patients treated after implementation. The guideline called for a respiratory culture, followed by empirical therapy using the combination of vancomycin, imipenem, and ciprofloxacin, and reassessment after 24–48 h. Patients treated under the guideline had significantly better rates of adequate therapy, a shorter duration of therapy, and a lower probability of secondary infections. Length of stay and mortality were numerically but not significantly lower.
Micek et al. [100] also investigated the value of short-course therapy in a randomized trial involving 290 patients. of these patients, 140 received conventional therapy at the discretion of the treating physician. For the other 150 patients, the investigators followed the patients and made recommendations to the treating physician. Discontinuation was recommended if the patient had a noninfectious etiology or if the patient's signs and symptoms resolved. Treating physicians usually discontinued antibiotic therapy within 48 h after the recommendation. Recommendations for discontinuation produced a significant reduction in the duration of therapy, with no significant differences in mortality, length of stay, or occurrence of secondary VAP.
Finally, Chastre et al. [48] completed a prospective, randomized, double-blind study of 401 patients with VAP diagnosed by use of BAL and quantitative cultures. Only patients receiving effective antibiotic therapy, as determined by their respiratory culture findings, were enrolled in the study. The objective was to determine whether 8 days is as effective as 15 days of antibiotic therapy. Patients who received a short course had neither excess mortality nor excess pulmonary infection recurrence. There were no significant differences regarding the number of days alive without mechanical ventilation or without organ failure, new antibiotic therapy received during the study period, the duration of ICU stay, or the mortality rate at day 60. Patients infected with nonfermenting gram-negative rods had a trend toward a higher chance of relapse when treated for 8 days. Resistant pathogens emerged more frequently in patients with recurrent pulmonary infection who had received antibiotics for 15 days.
Grading of Evidence
On the basis of a review of the 5 studies cited above, the members of the workshop agreed that the evidence available to support this statement was category V for the statement in general, category IV for the statement as it applies to hospitalized patients with HCAP, and category IV for the statement as it applies to nonhospitalized patients with HCAP (table 2).
Level of Support
When voting on the support for this statement, 9% of the summit participants voted to accept the statement completely, 64% voted to accept the statement with some reservations, and 27% voted to accept the statement with major reservations. In comparison, of the 383 IDSA members who participated in the online survey, 13% voted to accept the statement completely, 55% voted to accept the statement with some reservations, 15% voted to accept the statement with major reservations, 14% voted to reject the statement with reservations, and 3% rejected the statement completely (figure 14).
Voting comparison for statement 10 (“The duration of antibiotic therapy for patients with HCAP with a clinical response should be 7 days”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. HCAP, health care—associated pneumonia.
Voting comparison for statement 10 (“The duration of antibiotic therapy for patients with HCAP with a clinical response should be 7 days”). “Summit members” refers to the 11-member summit panel; “IDSA members” refers to the members of the Infectious Diseases Society of America who responded to a Web-based survey. HCAP, health care—associated pneumonia.
Discussion
There are currently no data on whether the duration of antibiotic therapy for patients with HCAP with a clinical response should be 7 days. However, hospitalized patient data demonstrate that patients with VAP can have a shorter therapy duration, and data on nonhospitalized patients with CAP indicate that we can treat with shorter duration [109]. The benefits of a shorter course of antibiotic therapy include lower cost, fewer potential adverse drug events, and, most importantly, a lower likelihood of selecting resistant bacteria. Until adequate data are collected, it seems prudent to recommend that the duration of antibiotic therapy for patients with HCAP with a clinical response should be 7 days.
Future Directions
Appropriately designed epidemiologic studies with rigorous microbiological criteria are clearly needed to better delineate the optimal duration of therapy for HCAP.
Conclusions
The recent definition of HCAP as a distinct subset of pneumonia was intended to identify those patients with an increased risk of infection caused by MDR pathogens. Identification of patients at risk for infection with MDR pathogens increases the likelihood of adequate empirical therapy while minimizing overuse of broad-spectrum antibiotics. Because initially inappropriate antibiotic therapy is associated with increased mortality and overuse of antibiotics leads to increased antibiotic resistance, this strategy is intended to improve short-term outcomes for individual patients and long-term outcomes for the general population.
The goal of the HCAP Summit was to critically appraise the existing literature to assess the relative strengths and limitations of our current knowledge in this area. The review was particularly challenging, given the historic use of the term “HCAP” to describe many diverse entities, including HAP, VAP, and NHAP. Overall, it is very clear that much is still unknown regarding every aspect of HCAP examined during the summit.
A recurring theme, regardless of which practice statement was being discussed, was the paucity of HCAP-specific data and the frequent extrapolation of data from other nosocomial infections. In the Defining HCAP workshop, the constraints of the current definition of HCAP were frequently identified as problematic. Because this is a relatively new definition, there is room for debate regarding which patient subsets should be included, the prevalence of various MDR pathogens in these patients, the lack of distinction between hospitalized patients and those treated in non—acute-care settings, overlap of the outcomes of HCAP with both CAP and HAP, and the lack of consideration regarding severity of illness. However, there was nearly unanimous agreement that severe CAP and HCAP constitute distinct entities. Although the treatment of these entities shares similar principles, they are distinguished by the risk for infection with MDR pathogens, with the former also being defined by severity-of-illness measures.
In the Therapeutic Intervention workshop, these complex theoretical concerns translated into discrepant opinions regarding empirical therapy, de-escalation of antibiotic therapy, and the duration of treatment. These discussions focused on the need to balance empirically covered MDR pathogens through the use of broad-spectrum therapy while minimizing the generation of more resistance through unnecessary antibiotic use. Highlighting the importance—and lack of consensus—regarding this issue, summit participants felt compelled to revise one of the statements to specify that the choice of empirical HCAP therapy must consider treatment location, severity of illness, prior antibiotic therapy, and functional status. Although this revised statement was more widely accepted, it was explicitly stated that this strategy was based more on intuition and opinion than on clinical data.
An examination of the disparities between summit participants' opinions regarding the clinical practice statements and those of a large pool of practicing infectious disease clinicians is interesting. Generally speaking, practicing specialists were much more likely to accept the statements than were the summit participants. These disparities are likely attributable to differences in the interpretation of the statements. Although we attempted to write concise statements that avoided vague terminology, summit participants had clear reservations regarding the application of the statements to nonhospitalized patients with HCAP. The authors have been careful to highlight where such discrepancies led to limitations in their evidence-based analyses.
At the summit's conclusion, participants identified several areas of research that merit priority to refine our care of patients with HCAP. Large-scale, multicenter, observational, cohort studies with rigorous microbiological data are needed to better define the precise subsets of patients at risk for infection with MDR pathogens, as well as to better delineate risk factors for specific pathogens. Similar studies are needed regarding the implications of severity of illness for outcomes. In addition, a clear need exists for specific studies on antibiotic therapy de-escalation, specifically by pathogen species, and on the optimal duration of therapy. Investigators should be actively encouraged to pursue these lines of investigation in the future.
Note. Medical literature is continuously being updated. After our summit was held, a new guideline by the IDSA and the Society for Healthcare Epidemiology of America addressing the issue of antimicrobial stewardship was published [110]. We include this reference for completeness, recognizing that this document was not used in the summit discussions.
Acknowledgments
Supplement sponsorship. This article was published as part of a supplement entitled “Health Care—associated Pneumonia (HCAP): A Critical Appraisal to Improve Identification, Management, and Outcomes—Proceedings of the HCAP Summit,” sponsored by Medical Education Resources and Consensus Medical Communications and supported by an unrestricted educational grant from Ortho-McNeil administered by Ortho-McNeil Janssen Scientific Affairs, LLC.
Potential conflicts of interest. M.H.K. has received grants/research support from Merck, Pfizer, Elan, Bard, Wyeth, and Johnson & Johnson. L.E.M. has been a speakers' bureau participant for Pfizer, Ortho-McNeil, and Schering Plough. D.E.C. has received grants/research support from Bard and Nomir; has been a speakers' bureau participant for Merck, Elan, Pfizer, Wyeth, and Sanofi Pasteur; and has received financial support from the Data and Safety Monitoring Board of Johnson & Johnson. J.E.M. has received grants/research support from AstraZeneca, Elan, Johnson & Johnson, PRD, Pfizer, and 3M, and has been a consultant for Merck, Elan, Replidyne, and Wyeth. S.T.M. has received grants/research support from Johnson & Johnson. M.S.N. has been a consultant, shareholder, and speakers' bureau participant for Pfizer, Schering Plough, Ortho-McNeil, Aventis, Merck, Elan, AstraZeneca, and Wyeth. D.L.P. has received grants/research support from AstraZeneca, Elan, and Pfizer; has been a consultant for Merck, Cubist Pharmaceuticals, Elan, Genzyne, KeyBay, Acureon, Wyeth, and Johnson & Johnson; and has been a speakers' bureau participant for Merck, Elan, and Cubist Pharmaceuticals. All other authors: no conflicts.
References
K
, MS
. Health-care-associated pneumonia: a new therapeutic paradigm
, , 2005
, vol. (pg. 3784
-)MH
, A
, YP
, et al. Epidemiology and outcomes of health-care-associated pneumonia: results from a large US database of culture-positive pneumonia
, , 2005
, vol. (pg. 3854
-)Guidelines for the management of adults with hospital-acquired, ventilator-associated, and healthcare-associated pneumonia
, , 2005
, vol. (pg. 388
-)DE
. What is healthcare-associated pneumonia, and how should it be treated?
, , 2006
, vol. (pg. 153
-)LF
, EW
. , National hospital ambulatory medical care survey: 2004 emergency department summary
Available at: http://www.cdc.gov/nchs/data/ad/ad372.pdf. Accessed 2 April 2007
ND
, KS
, JE
, et al. Health care-associated bloodstream infections in adults: a reason to change the accepted definition of community-acquired infections
, , 2002
, vol. (pg. 791
-)RR
, RV
, MB
, JA
, M
. Nursing home-acquired pneumonia: an emergency department treatment algorithm
, , 2004
, vol. (pg. 1309
-)PM
, DW
, W
, A
, JG
. Timing of antibiotic administration and outcomes for Medicare patients hospitalized with community-acquired pneumonia
, , 2004
, vol. (pg. 637
-)M
, S
, G
, VJ
, MH
. Clinical importance of delays in the initiation of appropriate antibiotic treatment for ventilator-associated pneumonia
, , 2002
, vol. (pg. 262
-)CM
, P
, MS
, et al. Impact of BAL data on the therapy and outcome of ventilator-associated pneumonia
, , 1997
, vol. (pg. 676
-)AF
, YP
, AD
, et al. Healthcare-associated bloodstream infection: a distinct entity? Insights from a large U.S. database
, , 2006
, vol. (pg. 2588
-)DR
. Guidelines for the management of adults with health care-associated pneumonia: implications for nursing facility residents
, , 2006
, vol. (pg. 719
-)OC
, LJ
, R
, C
, R
. Guidelines for preventing health-care-associated pneumonia, 2003: recommendations of CDC and the Healthcare Infection Control Practices Advisory Committee
, , 2004
, vol. RR-3
(pg. -36
)O
, J
, UB
, et al. Penicillin as empirical therapy for patients hospitalised with community acquired pneumonia at a Danish hospital
, , 2001
, vol. (pg. 84
-)RF
, JC
, JS
. Antimicrobial treatment of lower respiratory tract infections in the hospital setting
, , 2005
, vol. Suppl 7A
(pg. -38
)AE
, EM
. The rising influx of multidrug-resistant gram-negative bacilli into a tertiary care hospital
, , 2005
, vol. (pg. 1792
-)S
, VL
. A new category—healthcare-associated pneumonia: a good idea, but problems with its execution
, , 2006
, vol. (pg. 627
-)RG
. Healthcare-associated bacteremia: stirring the mud
, , 2006
, vol. (pg. 2685
-)M
, SC
, R
, et al. Effect of a clinical pathway to reduce hospitalizations in nursing home residents with pneumonia: a randomized controlled trial
, , 2006
, vol. (pg. 2503
-)WS
, JT
. A prospective comparison of nursing home acquired pneumonia with community acquired pneumonia
, , 2001
, vol. (pg. 362
-)TJ
, W
. A comparison of nursing home-acquired pneumonia patients with patients with community-acquired pneumonia and nursing home patients without pneumonia
, , 1997
, vol. (pg. 50
-)AA
, C
, A
, M
, E
. Indicators of potentially drug-resistant bacteria in severe nursing home-acquired pneumonia
, , 2004
, vol. (pg. 474
-)DE
, R
, DP
. Healthcare-associated pneumonia in adults: management principles to improve outcomes
, , 2004
, vol. (pg. 939
-)PA
, AG
, TM
, H
, G
. Community-acquired methicillin-resistant Staphylococcus aureus infection in Singapore is usually “healthcare associated”
, , 2003
, vol. (pg. 436
-)TP
, JM
, J
, et al. Process of care performance, patient characteristics, and outcomes in elderly patients hospitalized with community-acquired or nursing home-acquired pneumonia
, , 2000
, vol. (pg. 1378
-)RR
. Pneumonia in residents of long-term care facilities: epidemiology, etiology, management, and prevention
, , 1998
, vol. (pg. 319
-)AA
, P
, F
, J
. , Am J Respir Crit Care Med
, , vol. 163
(pg. -51
)A
. Debilitated elderly in long-term care facility (LTCF): a major reservoir of colonization with multidrug-resistant gram-negative pathogens [abstract 1067]
, Program and abstracts of the 44th Annual Meeting of the Infectious Diseases Society of America (Toronto)
, Arlington, VA
Y
, RN
, AJ
. Clinical epidemiology of pneumonia in hemodialysis patients: the USRDS waves 1, 3, and 4 study
, , 2006
, vol. (pg. 1135
-)FX
, G
, M
, et al. Community-acquired infection with healthcare-associated methicillin-resistant Staphylococcus aureus: the role of home nursing care
, , 2006
, vol. (pg. 1213
-)EA
, VM
, WH
, PH
. , Ann Intern Med
, , vol. 79
(pg. -34
)M
, B
, J
, V
, GP
. Gram-negative bacillary pneumonia in the compromised host
, , 1977
, vol. (pg. 241
-)J
, B
, A
, F
, F
. Bacteremic pneumonia in neutropenic patients with cancer: causes, empirical antibiotic therapy, and outcome
, , 1998
, vol. (pg. 868
-)AA
, AT
, H
, F
. Radiographic resolution of community-acquired bacterial pneumonia in the elderly
, , 2004
, vol. (pg. 224
-)AM
, SH
. The approach to nonresolving pneumonia in the elderly
, , 1993
, vol. (pg. 59
-)T
, JI
. M
, G
, J
. Nonresolving, slowly resolving, and recurrent pneumonia: infectious etiologies and noninfectious mimics
, , 2001
Lippincott Williams and Wilkins
(pg. -31
)DE
, T
, T
. , Curr Opin Pulm Med
, , vol. 11
(pg. -52
)CM
, KV
. , Clin Chest Med
, , vol. 26
(pg. -58
)CM
, D
, MS
, et al. Resolution of ventilator-associated pneumonia: prospective evaluation of the clinical pulmonary infection score as an early clinical predictor of outcome
, , 2003
, vol. (pg. 676
-)CS
, CD
. , Chest
, , vol. 108
(pg. 17
-)R
, A
, R
, et al. Risk factors of treatment failure in community acquired pneumonia: implications for disease outcome
, , 2004
, vol. (pg. 960
-)B
, J
, N
, et al. Causes and factors associated with early failure in hospitalized patients with community-acquired pneumonia
, , 2004
, vol. (pg. 502
-)R
, A
, JM
, et al. Nosocomial pneumonia: a multivariate analysis of risk and prognosis
, , 1988
, vol. (pg. 318
-)JR
, LW
, P
, CK
, VB
. , Am Rev Respir Dis
, , vol. 108
(pg. -40
)P
, JY
, J
, et al. Follow-up protected specimen brushes to assess treatment in nosocomial pneumonia
, , 1993
, vol. (pg. 38
-)J
, E
, V
, et al. Incidence, etiology, and outcome of nosocomial pneumonia in mechanically ventilated patients
, , 1991
, vol. (pg. 439
-)PJ
, AJ
, AG
, G
, MJ
. Resolution of infectious parameters after antimicrobial therapy in patients with ventilator-associated pneumonia
, , 2001
, vol. (pg. 1371
-)J
, M
, JY
, et al. Comparison of 8 vs 15 days of antibiotic therapy for ventilator-associated pneumonia in adults: a randomized trial
, , 2003
, vol. (pg. 2588
-)MA
, TC
, EW
, et al. The appropriate diagnostic threshold for ventilator-associated pneumonia using quantitative cultures
, , 2004
, vol. (pg. 931
-)JY
, J
, AJ
, et al. Nosocomial pneumonia in ventilated patients: a cohort study evaluating attributable mortality and hospital stay
, , 1993
, vol. (pg. 281
-)SH
, AM
, MS
, DE
, DH
. Utility of fiberoptic bronchoscopy in nonresolving pneumonia
, , 1990
, vol. (pg. 1322
-)M
, M
, M
, et al. Causes and predictors of nonresponse to treatment of intensive care unit-acquired pneumonia
, , 2004
, vol. (pg. 938
-)A
, C
, JL
, et al. Factors predicting ventilator-associated pneumonia recurrence
, , 2003
, vol. (pg. 1102
-)A
, CE
, JY
, et al. Early predictors for infection recurrence and death in patients with ventilator-associated pneumonia
, , 2007
, vol. (pg. 146
-)M
. Comparison of strategies using cefpirome and ceftazidime for empiric treatment of pneumonia in intensive care patients. The Cefpirome Pneumonia Study Group
, , 1998
, vol. (pg. 28
-)G
, S
. Value of APACHE II, SOFA and CPIS scores in predicting prognosis in patients with ventilator-associated pneumonia
, , 2006
, vol. (pg. 503
-)D
, R
, L
, et al. Analysis of factors that contribute to treatment failure in patients with community-acquired pneumonia
, , 2006
, vol. (pg. 159
-)F
, S
, JA
, et al. Antimicrobial treatment failures in patients with community-acquired pneumonia: causes and prognostic implications
, , 2000
, vol. (pg. 154
-)D
, L
, TN
, D
. Community-acquired pneumonia: causes of treatment failure in patients enrolled in clinical trials
, , 2003
, vol. (pg. 949
-)AA
, AT
, RS
, et al. Impact of invasive strategy on management of antimicrobial treatment failure in institutionalized older people with severe pneumonia
, , 2002
, vol. (pg. 1038
-)EH
, S
, G
, et al. Experience with a clinical guideline for the treatment of ventilator-associated pneumonia
, , 2001
, vol. (pg. 1109
-)MN
, W
, JA
, et al. Predictors of mortality for methicillin-resistant Staphylococcus aureus health-care-associated pneumonia: specific evaluation of vancomycin pharmacokinetic indices
, , 2006
, vol. (pg. 947
-)MH
, LE
, MS
, et al. Clinical characteristics and treatment patterns among patients with ventilator-associated pneumonia
, , 2006
, vol. (pg. 1210
-)DW
, S
, K
, et al. Practice guideline for evaluation of fever and infection in long-term care facilities
, , 2000
, vol. (pg. 640
-)TJ
. , Infect Control Hosp Epidemiol
, , vol. 23
(pg. -64
)KL
, KA
, JF
, et al. A prospective comparison of severity scores for identifying patients with severe community acquired pneumonia: reconsidering what is meant by severe pneumonia
, , 2006
, vol. (pg. 419
-)A
, PP
, JM
, et al. Validation of a predictive rule for the management of community-acquired pneumonia
, , 2006
, vol. (pg. 151
-)S
, M
, J
, et al. Severe community-acquired pneumonia: assessment of severity criteria
, , 1998
, vol. (pg. 1102
-)MJ
, TE
, DM
, et al. A prediction rule to identify low-risk patients with community-acquired pneumonia
, , 1997
, vol. (pg. 243
-)LA
, RG
, A
, et al. Infectious Diseases Society of America/American Thoracic Society consensus guidelines on the management of community-acquired pneumonia in adults
, , 2007
, vol. Suppl 2
(pg. -72
)S
, A
. , Clin Chest Med
, , vol. 20
(pg. -87
)J
, M
, D
, et al. Microbiological testing and outcome of patients with severe community-acquired pneumonia
, , 2003
, vol. (pg. 174
-)PP
, A
, I
, et al. Development and validation of a clinical prediction rule for severe community-acquired pneumonia
, , 2006
, vol. (pg. 1249
-)WS
, MM
, R
, et al. Defining community acquired pneumonia severity on presentation to hospital: an international derivation and validation study
, , 2003
, vol. (pg. 377
-)PE
, D
. , Chest
, , vol. 124
(pg. -36
)TS
, KH
, K
, et al. Comparison of community- and health care-associated methicillin-resistant Staphylococcus aureus infection
, , 2003
, vol. (pg. 2976
-)PK
, D
, DR
, et al. Prospective study of lower respiratory tract infections in an extended-care nursing home program: potential role of oral ciprofloxacin
, , 1988
, vol. (pg. 164
-)GM
, BA
, J
, LC
, S
. Randomized study of intravenous/oral ciprofloxacin versus ceftazidime in the treatment of hospital and nursing home patients with lower respiratory tract infections
, , 1989
, vol. 5A
(pg. -8
)B
, L
, S
, et al. , J Am Geriatr Soc
, , vol. 54
(pg. -63
)SC
, M
, L
. Pneumonia care and the nursing home: a qualitative descriptive study of resident and family member perspectives
, , 2006
, vol. pg. 2
LA
, CW
, PH
, et al. Two outbreaks of severe respiratory disease in nursing homes associated with rhinovirus
, , 2006
, vol. (pg. 284
-)E
, AM
. Evidence-based guidelines for management of nursing home-acquired pneumonia
, , 2002
, vol. (pg. 709
-)MH
, VL
, RR
. Legionnaires' disease in long-term care facilities: overview and proposed solutions
, , 2005
, vol. (pg. 875
-)SV
, LS
, GL
, et al. Ertapenem versus cefepime for initial empirical treatment of pneumonia acquired in skilled-care facilities or in hospitals outside the intensive care unit
, , 2006
, vol. (pg. 633
-)C
, R
, M
, H
. Moxifloxacin versus standard therapy in patients with pneumonia hospitalized after failure of preclinical anti-infective treatment
, , 2006
, vol. (pg. 190
-)M
, I
, K
, L
. β lactam monotherapy versus β lactam-aminoglycoside combination therapy for sepsis in immunocompetent patients: systematic review and meta-analysis of randomised trials
, , 2004
, vol. pg. 668
M
, K
, L
. β lactam monotherapy versus β lactam-aminoglycoside combination therapy for fever with neutropenia: systematic review and meta-analysis
, , 2003
, vol. pg. 1111
EH
, G
, S
, VJ
, MH
. The influence of inadequate antimicrobial treatment of bloodstream infections on patient outcomes in the ICU setting
, , 2000
, vol. (pg. 146
-)M
, J
, J
, BA
, O
. Piperacillin/tazobactam plus tobramycin versus ceftazidime plus tobramycin for the treatment of patients with nosocomial lower respiratory tract infection. Piperacillin/tazobactam Nosocomial Pneumonia Study Group
, , 1999
, vol. (pg. 389
-)CA
, DJ
, DR
, RP
, PK
. A randomized study of ciprofloxacin versus ceftriaxone in the treatment of nursing home-acquired lower respiratory tract infections
, , 1991
, vol. (pg. 979
-)RM
, JR
, FC
, et al. Changes in the epidemiology of methicillin-resistant Staphylococcus aureus in intensive care units in US hospitals, 1992–2003
, , 2006
, vol. (pg. 389
-)National Nosocomial Infections Surveillance (NNIS) System Report, data summary from January 1992 through June 2004, issued October 2004
, , 2004
, vol. (pg. 470
-)JP
3rd. Hospital-acquired pneumonia: risk factors, microbiology, and treatment
, , 2001
, vol. Suppl 2
(pg. -84
)A
, CE
, JY
, et al. Impact of methicillin resistance on outcome of Staphylococcus aureus ventilator-associated pneumonia
, , 2004
, vol. (pg. 786
-)J
, A
, M
, et al. Ventilator-associated pneumonia by Staphylococcus aureus: comparison of methicillin-resistant and methicillin-sensitive episodes
, , 1994
, vol. (pg. 1545
-)AF
, A
, MH
, J
. Methicillin-resistant Staphylococcus aureus prolongs intensive care unit stay in ventilator-associated pneumonia, despite initially appropriate antibiotic therapy
, , 2006
, vol. (pg. 700
-)Trc0603c
SE
, G
, EN
, et al. Comparison of mortality associated with methicillin-resistant and methicillin-susceptible Staphylococcus aureus bacteremia: a meta-analysis
, , 2003
, vol. (pg. 53
-)E
, L
, PC
, EMC
. Methicillin-resistant Staphylococcus aureus bacteraemia diagnosed at hospital admission: distinguishing between community-acquired versus healthcare-associated strains
, , 2004
, vol. (pg. 474
-)GE
, JA
, JA
, ST
, MH
. Methicillin-resistant Staphylococcus aureus sterile-site infection: the importance of appropriate initial antimicrobial treatment
, , 2006
, vol. (pg. 2069
-)ST
, S
, VJ
, MH
. A randomized controlled trial of an antibiotic discontinuation policy for clinically suspected ventilator-associated pneumonia
, , 2004
, vol. (pg. 1791
-)N
, P
, CW
, MM
, VL
. Short-course empiric antibiotic therapy for patients with pulmonary infiltrates in the intensive care unit: a proposed solution for indiscriminate antibiotic prescription
, , 2000
, vol. (pg. 505
-)RP
, MA
. Ventilator-associated pneumonia patients who do not reduce bacteria from the lungs have a worse prognosis
, , 2003
, vol. (pg. 269
-)RF
, A
. Evidence-based assessment of diagnostic tests for ventilator-associated pneumonia: executive summary
, , 2000
, vol. Suppl 2
(pg. -81
)BJ
, JM
. Treatment guideline for nursing home-acquired pneumonia based on community practice
, , 2000
, vol. (pg. 82
-)BJ
, JM
, F
, J
, RL
. Antibiotic use, hospital admissions, and mortality before and after implementing guidelines for nursing home-acquired pneumonia
, , 2001
, vol. (pg. 1020
-)JM
, S
, M
. Pneumonia versus aspiration pneumonitis in nursing home residents: prospective application of a clinical algorithm
, , 2005
, vol. (pg. 755
-)DM
, DN
, JF
Jr, et al. Society for Healthcare Epidemiology of America and Infectious Diseases Society of America Joint Committee on the Prevention of Antimicrobial Resistance: guidelines for the prevention of antimicrobial resistance in hospitals
, , 1997
, vol. (pg. 275
-)EH
, L
, C
, VJ
, MH
. The occurrence of ventilator-associated pneumonia in a community hospital: risk factors and clinical outcomes
, , 2001
, vol. (pg. 555
-)J
, HR
, RE
. Principles of antibiotic treatment of community-acquired pneumonia in the outpatient setting
, , 2005
, vol. Suppl 7A
(pg. -8
)TH
, RC
, JE
Jr, et al. Infectious Diseases Society of America and the Society for Healthcare Epidemiology of America guidelines for developing an institutional program to enhance antimicrobial stewardship
, , 2007
, vol. (pg. 159
-)© 2008 by the Infectious Diseases Society of America
0 Comments
You have entered an invalid code
Thank you for submitting a comment on this article. Your comment will be reviewed and published at the journal's discretion. Please check for further notifications by email.