Article Text

Download PDFPDF
Disease severity prediction for nursing home-acquired pneumonia in the emergency department
  1. S Y Man1,
  2. C A Graham1,
  3. S S W Chan1,
  4. P S K Mak1,
  5. A H Y Yu1,
  6. C S K Cheung1,
  7. P S Y Cheung1,
  8. G Lui2,
  9. N Lee2,
  10. M Chan3,
  11. M Ip4,
  12. T H Rainer1
  1. 1Accident and Emergency Medicine Academic Unit, The Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, NT, Hong Kong SAR
  2. 2Department of Medicine and Therapeutics, The Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, NT, Hong Kong SAR
  3. 3Department of Pathology, The Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, NT, Hong Kong SAR
  4. 4Department of Microbiology, The Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, NT, Hong Kong SAR
  1. Correspondence to Professor T H Rainer, Accident and Emergency Medicine Academic Unit, Chinese University of Hong Kong, Trauma & Emergency Centre, Prince of Wales Hospital, Shatin, NT, Hong Kong SAR; thrainer{at}cuhk.edu.hk

Abstract

Background Prediction rules exist for the assessment of community-acquired pneumonia but their use in nursing home-acquired pneumonia (NHAP) remains undefined. The objectives of this study were to compare the prognostic ability for severe NHAP of five prediction rules (PSI, CURB-65, M-ATS, R-ATS, España rule), and to evaluate their usefulness to identify patients with less severe disease in the emergency department for outpatient care.

Methods A prospective observational study of consecutive NHAP patients was conducted at a university teaching hospital emergency department in Hong Kong between January 2004 and June 2005. The primary outcome was severe pneumonia (defined as combined 30-day mortality and/or intensive care unit (ICU) admission).

Results 767 consecutive NHAP patients were included. Mean (SD) age was 83.4 (9.0) years; 350 (45.6%) were male and 644 (84.0%) had coexisting illness. 95 patients died within 30 days (12.4%), five patients were admitted to the ICU (0.7%) and 98 patients had severe pneumonia (12.8%). Sensitivity and specificity of each decision rule ranged from 37.8% to 95.9% and 15.1% to 87.6% respectively. The overall predictive performance of each rule was between 0.627 and 0.712. The negative likelihood ratios of PSI (0.27) and CURB-65 (0.23) were lower than M-ATS (0.71), R-ATS (0.45) and España (0.39). After excluding 204 patients with either poor functional status or those >90 years of age, sensitivities of M-ATS (96.0%) and R-ATS (100%) improved greatly with negative likelihood ratios of <0.1.

Conclusion PSI and CURB-65 are useful for identification of patients with less severe NHAP.

  • Infectious diseases
  • respiratory
  • pneumonia
  • pneumonia/infections

Statistics from Altmetric.com

Introduction

Nursing home-acquired pneumonia (NHAP) is a leading cause of death in nursing home patients and considered to be different from community-acquired pneumonia (CAP) in terms of its aetiology and prognosis.1–5 Nursing home residents in Hong Kong vary in levels of dependency, but they are generally older and have more coexisting illness than patients from the community. The management of this group of patients is often challenging as they usually have more serious disease but seldom receive intensive care because of their poor functional status and advanced age. In Hong Kong, as elsewhere, the ageing population is leading to an increasing proportion of nursing home patients and the authors' previous work6 showed that one-quarter of the admitted pneumonia cases in a university teaching hospital were from nursing homes. This reflects the hospital-based elderly healthcare system in Hong Kong, which results in a high reliance on public hospitals to provide geriatric care.

The huge burden of nursing home patients on healthcare systems in the future is foreseeable, and therefore an accurate disease severity assessment for NHAP is as important as for CAP in order to enhance patient care and resource utilisation. A decision tool that could safely identify less severely affected NHAP patients suitable for hospital outpatient management would be highly relevant to the practice of emergency departments (ED) in Hong Kong or elsewhere with similar demographics and healthcare systems.

There are multiple existing prediction rules that aim to stratify CAP into different risk categories based on combinations of different prognostic variables. The application of these rules is focused mainly on CAP, and their use in nursing home patients is not clearly defined. Only one study7 validated the pneumonia severity index (PSI)8 in a retrospective review of 158 nursing home patients.

Other important prediction rules, namely the British Thoracic Society's CURB-65 rule,9 the modified American Thoracic Society rule (M-ATS),10 11 the revised criteria (R-ATS) for intensive care recommended in the Consensus Guidelines on the management of CAP in adults1 and the newly developed eight-variables prediction rule for severe CAP by España et al,12 have not been validated in nursing home patients.

The objectives of this study were to compare the prognostic ability for severe pneumonia of five prediction rules (PSI, CURB-65, M-ATS, R-ATS and España rule) in nursing home patients and to evaluate their usefulness to identify patients with less severe disease in the ED for outpatient management.

Method

Design and study population

This study was conducted in the ED of the Prince of Wales Hospital, the main teaching hospital of the Chinese University of Hong Kong, which serves a population of 1.5 million people in the New Territories of Hong Kong. The ED provides a 24 h emergency physician covered service to the public and has an annual census of over 140 000. Ethical approval was obtained from the Institutional Review Boards of the Hospital Authority of Hong Kong and the Chinese University of Hong Kong to conduct a prospective study in patients with pneumonia.

Consecutive patients residing in nursing homes who presented to the ED between 1 January 2004 and 30 June 2005 with a diagnosis of ‘pneumonia’ made by an attending emergency physician were recruited for the study. Pneumonia was defined in this study as an acute infection of the pulmonary parenchyma that was associated with symptoms of acute infection, accompanied by a previously unknown infiltrate on a chest radiograph.1 All chest radiographs were interpreted by an attending emergency physician as in usual practice.

Patients who were aged <18 years; immunocompromised patients (HIV, neutropenia, receiving immunosuppressive drug therapy including steroids and chemotherapeutic agents for more than 3 months); or those who had been previously hospitalised within 14 days before the onset of symptoms were excluded. All patients were treated with empirical antibacterial agents according to local hospital guidelines including a β lactam antibiotic with or without a macrolide or a fluoroquinolone based on the disease severity. Haemodynamically unstable patients were further assessed by an on-call intensive care unit (ICU) physician in the ED for the decision of ICU admission. All medical and nursing staff members who were involved in patient management were not aware that a specific study on NHAP was in progress and they were also not provided with any clinical prediction rules for making disposal decisions.

Data collection and outcome measures

Data were recorded by research nurses on a pre-designed data collection sheet with all required variables amalgamated together, including baseline patient characteristics, coexisting illness, examination findings, laboratory and radiographic results, the length of stay (LOS) in hospital and outcome measures. A patient's mental status was assessed by the emergency physician clinically, and mental confusion was defined in the study as GCS <15 or a new onset of disorientation to time, place or person for practical purposes.

The primary outcome measure was severe pneumonia, which was defined in this study as the combined outcomes of 30-day mortality and/or ICU admission. Severity scores for the five predictive rules (PSI, CURB-65, M-ATS, R-ATS and España rule) shown in table 1 were calculated and their performance in predicting severe pneumonia in NHAP was compared.

Table 1

Calculation parameters of the PSI, CURB-65, modified-ATS, revised criteria for ATS and España rule

Statistical analysis

All data were analysed using SPSS v13.0. Descriptive statistics of demographic and clinical variables included frequencies, percentage, means and SDs, median and IQR (from the 25th centile to the 75th centile). χ2 Tests were used to compare categorical variables. Mann–Whitney U tests and Kruskal–Wallis tests were used to compare skewed continuous variables of two groups and more than two groups respectively. The sensitivity, specificity, positive predictive value, negative predictive value and likelihood ratios (LRs) of each rule were compared. The overall accuracy in predicting severe pneumonia was calculated as an area under the receiver operating characteristic curve (AUC). For all analyses, a two-tailed p value of <0.05 was considered statistically significant.

Results

A total of 767 nursing home patients with provisional diagnosis of pneumonia made in the ED were included in the study and all required 30-day data were collected. The mean (SD) age was 83.4 (9.0) years and 350 (45.6%) were male. Baseline characteristics are shown in table 2. Six-hundred and forty-four (84.0%) patients had coexisting illness, of which cerebrovascular disease (32.6%) was the most common, followed by congestive heart failure (15.5%) and chronic obstructive pulmonary disease (15.5%). Ninety-five patients (12.4%) died within 30 days. Two-hundred and four patients (26.6%) who had either very poor functional status secondary to coexisting illnesses or those who were aged >90 years, were identified as not suitable for potential intensive care regardless of disease severity (except for standard empirical antibiotic treatment). Fifteen of these patients required non-invasive mechanical ventilation and were admitted to the high dependency unit, and 75 patients died without ICU admission.

Table 2

Baseline characteristic of NHAP patients (n=767)

The mean (SD) LOS in hospital was 8.5 (6.4) days. Five patients (0.7%) requiring endotracheal intubation were admitted to the ICU, of whom two died. Eighty-nine patients (11.6%) requiring non-invasive mechanical ventilation were admitted to the high dependency unit of the medical ward. Ninety-eight patients (12.8%) had severe pneumonia. There were no severe acute respiratory syndrome13 cases identified in Hong Kong during the study.

Risk stratification by the five predictive rules

The risk stratification of each rule and the outcome measures are shown in tables 3 and 4. PSI and CURB-65 stratified the NHAP patients into risk groups showing significant differences in 30-day mortality and severe pneumonia, but no significant differences for ICU admission or hospital LOS. One-hundred and five patients (13.7%) were classified by PSI as low-risk group (classes I, II and III), with a 30-day mortality rate of 2.9% and severe pneumonia rate of 3.8%. Two-hundred and nineteen patients (28.6%) were classified by CURB-65 as low-risk group (score 0 and 1), with a 30-day mortality rate of 2.7% and severe pneumonia rate of 3.2%.

Table 3

Risk stratification by PSI and CURB-65 and outcome measures in NHAP

Table 4

Risk stratification by M-ATS, R-ATS and España rule and outcome measures in NHAP

M-ATS, R-ATS and the España rule stratified the NHAP patients into two risk groups with significant differences between the high-risk and low-risk groups for 30-day mortality and severe pneumonia rates. However, only the difference in ICU admission rate between the risk groups of M-ATS and R-ATS was significant. There was no significant difference in hospital LOS between the risk groups of any prediction rule for NHAP. More patients were identified as low risk by M-ATS (84.4%), R-ATS (71.7%) and España (45.5%), than by PSI and CURB-65, but with a higher 30-day mortality (M-ATS 9.3%; R-ATS 6.2%; España 5.4%) and higher severe pneumonia rates (M-ATS 9.4%; R-ATS 6.2%; España 5.4%) compared with the low-risk groups of PSI and CURB-65.

The data were re-analysed with the 204 patients who were considered not suitable for ICU admission excluded. PSI stratified 103 patients (18.3%) as low risk, with a mortality rate of 1% and a severe pneumonia rate of 1.9%; CURB-65 stratified 189 patients (33.6%) as low risk, with a mortality rate of 1.1% and a severe pneumonia rate of 1.6%; M-ATS stratified 466 patients (82.8%) as low risk, with a mortality rate of zero and a severe pneumonia rate of 0.2%; R-ATS stratified 425 patients (75.5%) as low risk, with the mortality and severe pneumonia rates equal to zero; España stratified 297 patients (52.8%) as low risk, with the mortality and severe pneumonia rates equal to 1.7%

Predictive characteristics of each predictive rule

The predictive characteristics (sensitivity, specificity, positive predictive value, negative predictive value, positive and negative LRs, and area under the curve (AUC)) of each decision rule in identifying severe pneumonia patients are shown in table 5. PSI class >3 and CURB-65 score >1 were the cut-offs chosen to represent intermediate-risk to high-risk patients.8 9

Table 5

Sensitivity, specificity, positive and negative predictive values, positive and negative likelihood ratios of severe pneumonia prediction by different predictive rules in NHAP (n=767: all patients included)

PSI and CURB-65 had sensitivities of more than 90% and the M-ATS had the highest specificity (87.6%) of all the decision rules. The positive LR of each decision rule was between 1.13 and 3.04, whereas the negative LR of each decision rule was between 0.23 and 0.71. The AUC of all decision rules varied from 0.627 to 0.712, with considerable overlapping in their 95% CIs.

With the exclusion of the 204 patients deemed unsuitable for ICU admission, the predictive characteristics of the five decision rules are shown in table 6. There were only minor changes in the predictive characteristics of each of the decision rules, except that the sensitivities of both the M-ATS and R-ATS were significantly enhanced and their negative LRs were close to zero. The AUCs of M-ATS and R-ATS were 0.909 and 0.894 respectively.

Table 6

Sensitivity, specificity, positive and negative predictive values, positive and negative likelihood ratios of severe pneumonia prediction by different predictive rules in NHAP (n=563: 204 patients identified not for aggressive treatment excluded)

Discussion

Although the pneumonia severity decision rules are primarily derived and used for CAP, they may be useful in the ED context as a triage tool for the immediate disposal of NHAP patients. The M-ATS and R-ATS rules can accurately exclude severe pneumonia when applied selectively to nursing home patients aged ≤90 years with a relatively normal functional status. PSI and CURB-65 are useful in excluding severe cases when applied to nursing home patients non-selectively.

The ability of a decision rule to discriminate is fundamental for clinical use.14 With knowledge of the pre-test probability, a LR can provide more information than sensitivity, specificity and AUC by indicating how likely it is that the patient would develop the problem based on a positive or a negative result in a specific locality. This information would allow emergency physicians to choose the right diagnostic tool for specific purposes under different circumstances.

LRs above 10 or below 0.1 are generally considered useful to ‘rule in’ or ‘rule out’ diagnoses respectively.15 This study aimed to characterise decision rules with low negative LRs, which would be useful for identifying potential patients for outpatient management (ie, not as a hospital inpatient), although it is acknowledged that disease severity is not the sole factor determining patient disposal.

PSI and CURB-65 had their negative LR close to 0.1, which made them useful in screening out severe cases. M-ATS and R-ATS had negative LRs of less than 0.1 in the selected NHAP patient group. These scores could complement PSI and CURB-65 in identifying less severe NHAP patients for potential outpatient care or for short-stay treatment in an emergency medicine ward setting.

Disease severity is not the sole factor to determine the decision for ICU care. It is accepted in Hong Kong that nursing home patients aged over 90 years or with poor functional status are unlikely to receive intensive care support in most hospitals despite severe disease due to resource constraints and the high likelihood of a poor outcome. This group has a high mortality rate but is often not classified as severe by M-ATS or R-ATS because mechanical ventilation (one of the two major criteria for disease classification as ‘severe’ in M-ATS or R-ATS1 10 11) is usually not considered for this group of patients even with serious disease severity. Therefore, the exclusion of the group with a low likelihood of benefitting from ventilation enhances the sensitivity and the overall severe disease prediction performance of M-ATS and R-ATS.

All the study decision rules have a relatively low positive predictive ability for severe disease, as this can be reflected by their low positive LR. Although the M-ATS rule had the best performance in identifying severe disease among all study decision rules, its positive LR was still well below 10. Therefore, the decision rules were more helpful in ‘ruling out’ rather than ‘ruling in’ the diagnosis of severe pneumonia.

There is no universal definition for severe pneumonia; its interpretation varies between different authorities.8–12 Different outcome measures like mortality, ICU admission, the development of shock and the requirement for mechanical ventilation have been used in previous studies to quantify disease severity. Using any single parameter to interpret disease severity can sometimes be misleading. The combined outcome of ICU admission and death within 30 days was therefore employed in the study to define severe pneumonia in order to incorporate most of the severe cases and to correlate clinical practice in local EDs.

The strength of this study relies on its large sample size, the completeness of data collection and the detailed analysis comparing the five decision rules. Several limitations in this study should be noted. First, this is only a single-centre study despite the wide catchment area covered by the university teaching hospital. Second, the focus was only on the clinical diagnosis made by emergency physicians, which may be different from the final or discharge diagnosis. However, such a weakness could also be considered as a merit of the study because it reflects real ED clinical practice as initial patient management and immediate disposal decisions are based on the provisional diagnosis rather than the formal discharge diagnosis.16 Third, there was an interobserver variability in radiographic interpretation between different emergency physicians. However, this is again a reflection of real ED practice as it is impossible to have all radiographs reviewed by another physician before making a disposal decision in a busy ED.

The implementation of using these decision rules in deciding patient disposal may help to identify a significant proportion of NHAP patients for outpatient care, which offers a considerable cost reduction. However, all predictive rules serve only as a guide to clinical management, and clinicians should exercise their judgement in interpreting the results of these rules in different clinical situations. Knowledge of the strengths, weaknesses and the predictive characteristics of each decision rule is crucial to the appropriate clinical application.

Conclusion

PSI and CURB-65 are useful tools in identification of less severe NHAP. The very low negative LRs of M-ATS and R-ATS in the selected group of nursing home residents could complement PSI and CURB-65 in identifying less severely affected patients for potential outpatient care.

References

Footnotes

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval This study was conducted with the approval of the Institutional Review Boards of the Hospital Authority of Hong Kong and the Chinese University of Hong Kong.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.