INTRODUCTION
Community-acquired pneumonia (CAP) requiring hospitalization is a common and serious illness that mostly affects those 65 years of age and older25,32. Indeed, over the past 15 years the rates of hospitalization for CAP among this age-group have increased14. The short-term mortality of those hospitalized with CAP has been well documented, and consistently seems to range from 10% to 11%21,26.
In contrast, few studies have assessed the long-term morbidity or mortality associated with an episode of CAP requiring hospitalization. Early studies suggested that patients with CAP may have high long-term mortality; however these studies were limited by small size and did not clearly differentiate between the shorter-term mortality associated with the acute CAP event and the longer-term mortality that was unlikely to be directly related to pneumonia4,16,20,34,42. More recently, 2 large studies7,19 using administrative databases reported 1-year CAP-associated mortality ranging from 25% to 40%; however, these studies lacked important clinical and comorbidity data, thereby limiting their interpretation. The few studies that have attempted to address risk factors associated with long-term mortality have been limited by relatively small sample sizes4,16,20,40-42.
Better knowledge of long-term outcomes may help guide patient care by improving the physician's ability to prognosticate, and could help identify a group of patients at high risk of adverse events. Furthermore, we hypothesized that the Pneumonia Severity Index (PSI), often calculated for patients at the time of hospital admission as a validated prognostic index and risk adjustment tool designed to predict 30-day pneumonia-related mortality, might also predict longer term morbidity and mortality11. We considered this hypothesis because the scoring of the PSI is heavily weighted by age, demographics, and comorbidities that predated pneumonia-that is, it might be better conceptualized as a measure of chronic ill-health and susceptibility to morbidity and mortality, rather than just a measure of "pneumonia severity." To both carefully describe the long-term morbidity and mortality associated with an episode of pneumonia and to test the aforementioned PSI hypothesis, we linked a prospective population-based clinical registry of more than 3000 patients hospitalized with CAP with multiple administrative databases.
METHODS
Patients and Setting
Between 2000 and 2002 all patients with CAP admitted to all 6 hospitals within Capital Health (Edmonton, Alberta, Canada) were enrolled in a clinical registry and treated according to a previously validated clinical pathway for the management of CAP15,27,28. Capital Health is one of the largest integrated health systems in Canada, serving over 1 million people within Edmonton, Alberta, Canada6. The population-based clinical cohort and data collection methods have been previously described in detail3,18,24,29. Briefly, all adult patients >17 years of age with CAP were enrolled. CAP was defined as 2 or more symptoms or signs of CAP (cough, productive or nonproductive; pleuritic chest pain; shortness of breath; temperature >38°C; and crackles, or bronchial breathing on auscultation) plus radiographic evidence of pneumonia as interpreted by the treating physician. The only patients excluded from the population-based cohort were those who had tuberculosis or cystic fibrosis; were immunocompromised; or were pregnant.
After initial triage in the emergency department, an inpatient physician was consulted for patients with a PSI score of 90 or greater or those with lower scores if the emergency department physician felt inpatient care was needed29. The PSI was intended to predict the risk of 30-day pneumonia-related mortality, and it is based on weighted sums of 3 demographic variables (age, sex, nursing-home resident), 5 comorbidities (cancer, liver disease, congestive heart failure, cerebrovascular disease, and renal disease), 5 physical findings (altered mental status, respiratory rate, systolic blood pressure, temperature, and pulse), and 7 laboratory tests (arterial blood gas, blood urea nitrogen, sodium, glucose, hematocrit, oxygen saturation, and presence of pleural effusion)11. The score is then tallied for each patient and stratified into 1 of 5 risk classes: class I (no predictors), class II (score ≤70), class III (score 71-90), class IV (score 91-130), and class V (score >130)11. Trained research nurses prospectively collected data using standardized abstraction forms, and patients were followed until discharge. Clinical characteristics (that is, premorbid functional status, need for advance directive, smoking status, comorbidities) as well as laboratory (chemistries, microbiology) and radiographic data were collected. The Health Research Ethics Board of the University of Alberta approved the study.
Linkage to Administrative Data
The province of Alberta provides universal health coverage for a population of approximately 3.3 million people8. All beneficiaries who are registered in the Alberta Health Care Insurance Plan are assigned a unique lifetime personal health number (PHN). The government's ministry of health maintains and continually updates demographic data on all registered residents. The ministry also merges Alberta Vital Statistics death data with the registry data by using deterministic linkage (over 99% linked). Data on all hospitalizations are routinely collected. There is also the potential to link the PHN with other health services such as physician office visits and emergency department visits.
Patient hospitalization data are collected by Alberta Health and Wellness (AHW). Between 2000 and 2002, AHW collected and classified up to 16 International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) discharge diagnoses. After 2002, ICD, Tenth Revision, Canadian version (ICD-10-CA) was used, and up to 25 discharge diagnoses were collected. Diagnostic coding is conducted by trained health records administrators at each hospital, and routine validation of coding accuracy is conducted both provincially and through a central agency for Canada10.
Using PHNs, AHW linked our population-based cohort of patients to the various provincial health services and vital statistics databases. The merged data were released to us for analysis in an anonymous and de-identified manner. As a result, all postdischarge individual health service data from the time of study entry to March 31, 2006, were extracted for each patient who was a registered beneficiary in Alberta.
Outcome Measures
The primary outcome measure was all-cause mortality by study end. We also evaluated 30-day (short-term) and 1-year (intermediate-term) mortality in an effort to differentiate mortality associated with the acute CAP event and that associated with the longer time frames where the episode of CAP itself could no longer directly contribute to mortality.
Our main secondary outcome was re-hospitalization for pneumonia. Similar to mortality, in patients surviving the initial CAP hospitalization, we assessed CAP-related hospitalizations at 30-days, 1-year, and at the end of follow-up. CAP-related hospitalization was defined as a hospital admission with a most responsible discharge diagnosis with ICD-9-CM values of 480.0-487.7 or ICD-10-CA values of J10-J18. Re-admission was defined as any inpatient hospitalization occurring ≥48 hours postdischarge, so as to exclude patient transfers between hospitals. These ICD diagnostic codes have been shown to be highly accurate (98% sensitivity and 97% specificity) for identification of pneumonia compared to medical records2,12,38. Further, these codes have been used in numerous studies evaluating the effectiveness of therapies and outcomes of patients with pneumonia in Canada and the United States7,19,20. A similar evaluation of all-cause re-hospitalization was also completed.
Analysis
For all subjects successfully linked to the administrative data, mortality rates from the time of initial CAP presentation until the time point of interest (that is, 30-day, 1-year, and study end) were described. Time-to-event curves were used to display the natural history of long-term mortality for pneumonia patients. Censoring occurred at the time of death, dis-enrollment (for example, moved out of province), or end of study (March 31, 2006). Kaplan-Meier product limits methods with log-rank tests and hazard ratios from Cox proportional hazards models were used to assess the relationship between the PSI (according to 5 risk classes) and mortality at study end.
We specifically chose to use the previously validated PSI as our "multivariable" adjusted model rather than undertake the creation of a new risk adjustment model for 3 reasons. First, this was our main hypothesis. Second, the PSI is already often calculated in many other jurisdictions, and thus would permit direct risk-adjusted comparisons of outcomes across many different populations, permitting future validation and replication of our work. Third, any modeling strategy that we developed may not be easily validated, raising concerns that our models might be over-fit to our study population. Assessment of the proportional hazards assumptions during the entire follow-up indicated no violations. All first-order interaction terms were assessed, and no significant interactions were observed (p value ≥ 0.1 for all comparisons). Similar analyses were conducted from the time of hospital discharge for patients surviving the initial CAP hospitalizations to all-cause or CAP-related hospitalization. All analyses were conducted using Statistical Package for Social Sciences (SPSS) version 15 (SPSS Inc., Chicago, IL).
RESULTS
Patient Characteristics
Of the 3415 patients with pneumonia enrolled in the initial cohort, we were able to successfully link 3284 (96.2%) patients to provincial databases. The primary reason for linkage failure (n = 131; 3.8%) was out-of-province residency (n = 67; 49%). The patients who were not linked were nearly identical to the patients successfully linked according to all variables (p > 0.1 for all examined differences) except that they were on average 4 years younger (69 yr vs. 65 yr; p = 0.02) and were less likely to be admitted from a nursing home (626 [19%] vs. 11 [8%]; p = 0.002).
Most patients were ≥65 years of age (n = 2178; 66%), 1726 (53%) were male, and 626 (19%) were from a nursing home (Table 1). Most (n = 2054, 63%) patients had a greater than 18% predicted 30-day mortality according their PSI score at presentation (that is, class IV or V). The median follow-up was 1377 days (3.8 yr), with maximal follow-up of 5.4 years.
;)
TABLE 1: Characteristics of 3284 Patients Admitted for Community-Acquired Pneumonia, Stratified According to Long-Term Mortality
Mortality
With respect to all-cause mortality, 378 (12%) patients had died within 30 days, 911 (28%) within 1 year, and 1738 (53%) by study end (Figure 1). In general, those who died during follow-up were more likely to be older and male, and had higher initial PSI scores (Table 1). Overall, 82 (19%) patients aged <45 years died (reference group), compared with 200 (30%) patients aged 45-64 years (hazard ratio [HR], 1.62; 95% confidence interval [CI], 1.26-2.10), and 1456 (67%) patients aged ≥65 years (HR, 5.07; 95% CI, 4.06-6.34) by study end. Male sex was also associated with higher mortality (971 [56%] men died vs. 767 [49%] women; HR, 1.20; 95% CI, 1.13-1.37).
FIGURE 1: Kaplan-Meier plot showing survival from time of presentation until study end.
Particularly noteworthy was the direct and graded relationship between the initial PSI class and subsequent events (Figure 2). By study end, 92 (15%) patients with the lowest-risk pneumonia (PSI risk class I-II, reference group) had died compared with 223 (36%) deaths in patients with PSI risk class III (HR, 2.75; 95% CI, 2.15-3.49), 807 (62%) deaths in those with PSI risk class IV (HR, 5.84; 95% CI, 4.70-7.24), and 616 (82%) deaths in those with the highest-risk pneumonia (PSI risk class V; HR, 11.80; 95% CI, 4.70-14.70). Furthermore, there was a consistent gradient effect of mortality according to PSI risk class for all 3 time periods assessed (Table 2). Major causes of death among the 1738 patients who died by study end included cardiovascular disease (31%), respiratory disease (including CAP) (26%), and cancer (18%).
FIGURE 2: Kaplan-Meier plot showing "adjusted" survival according to PSI.
TABLE 2: Short and Long-Term Mortality According to PSI Risk Class of 3284 Patients Admitted for Community-Acquired Pneumonia
Morbidity
Of the 2950 (90%) patients who survived the initial CAP hospitalization, a total of 7716 re-hospitalizations occurred with 2117 (72%) patients being re-hospitalized at least once during the follow-up period. The median number of hospitalizations per patient during follow-up was 2, ranging from 0 to 34 admissions. Four hundred seventy-six (16%) patients were readmitted for a repeat episode of pneumonia during the follow-up period; 66 (2%) patients were readmitted within 30 days of hospital discharge, and 271 (9%) patients within 1 year. Excluding pneumonia admissions, 2042 patients had at least 1 hospitalization during the follow-up period. The most common reasons for nonpneumonia-related admissions to hospital were other respiratory diseases (28%), circulatory disorders (16%), digestive system disorders (9%), fall-related injuries (7%), and cancer (5%). Remaining cause-specific categories occurred in less than 2% of total hospitalizations and are not presented.
DISCUSSION
In this large, population-based cohort of patients with pneumonia and up to 5.4 years of follow-up, we have shown that long-term morbidity and mortality after an episode of hospitalization for pneumonia is high. In fact, fewer than half of patients were still alive after a median 3.8 years of follow-up. Older age, male sex, and higher calculated PSI scores were associated with increased long-term mortality. Morbidity was also high; almost three-quarters of patients were re-hospitalized at some point during the follow-up period, and 16% were re-hospitalized for another episode of pneumonia.
To our knowledge, this is the largest outcomes study of patients hospitalized with pneumonia to extend follow-up beyond 1 year. Most prior studies were small, with enrollment ranging from 100 to 400 patients4,16,19,20,40,41. One larger study by Mortensen et al33 followed 1555 patients enrolled in the Pneumonia Patient Outcomes Research Team (PORT) cohort study for a mean of 5.9 years, and reported a long-term mortality of 39%. The PORT cohort followed patients who were treated for pneumonia on an inpatient and outpatient basis. When outpatients were excluded, the long-term mortality among hospitalized patients with pneumonia was 48%33. Of note, in contrast to our population-based "real world" cohort of patients, the PORT investigators screened over 12,500 potential cases of CAP to identify their final sample of 1555, and patients from the Canadian sites were excluded13. Regardless, the intermediate-term and long-term mortality associated with pneumonia is striking and is comparable to other conditions needing incident hospitalization such as chronic obstructive pulmonary disease (COPD), stroke, heart failure, or hip fracture1,9,17,23,41.
In the current study, older age, male sex, and higher PSI score were associated with greater mortality. Older age has previously been shown to be associated with higher long-term mortality in several studies19,33,40. Interestingly, 1 study did not find an association with age4, but the authors excluded nursing home patients, an elderly group of individuals at increased risk of dying from pneumonia19,40. Male sex has also consistently been found to influence mortality, although the reasons for this remain unclear19,33,41. Of course, age and sex are also included in the PSI score. We have now demonstrated that the predictive capability of the PSI extends far beyond 30 days, suggesting that the PSI may in addition be a very accurate measure of susceptibility to adverse events, or physiologic compromise or "frailty." Indeed, although there is no consensus as to the definition of frailty35, one attempt describes frailty as "losses of physiologic reserve that increase the risk of disability," and regards frailty as a precursor state to disability5,36. Thus, the PSI may be a marker of an individual's overall ill health and risk of death from an episode of acute illness as opposed to a true marker of "pneumonia severity." We hypothesized that this might be the case because so many of the variables included in the PSI predate the development of pneumonia.
Yende et al41 examined 1-year and 5-year mortality in 106 patients hospitalized with CAP and speculated that hospitalization with CAP triggers pathophysiologic processes that persist beyond recovery, possibly due to increased concentrations of pro-inflammatory cytokines or persistent abnormalities in the innate immune response. Based on our findings, we believe it may be more likely that needing to be hospitalized for pneumonia is a more general surrogate measure for physiologic compromise or frailty5,36. Attempts to adjust for frailty are clearly difficult and have not been very successful19. At least for patients with pneumonia, the current study suggests that the PSI itself might be a suitable candidate worthy of testing in other populations.
Frailty ought to predispose to morbidity also, a finding supported by the current study. Almost three-quarters of patients who survived the initial hospitalization were hospitalized again, with a median number of 2 admissions. The reasons for hospitalization varied, but included pneumonia and other respiratory conditions, cardiac conditions, and fall-related injuries. Perhaps CAP (like hip fracture, heart failure, or delirium) can be viewed as a sentinel event that identifies patients at high risk of mortality and future hospitalization22,39. If this were so, a targeted follow-up strategy would allow for more careful surveillance of this population, with implementation of key preventative measures including influenza and pneumococcal vaccination, management of comorbidities, and assessment of fall risks. This is analogous to other multidisciplinary approaches to managing the more classic chronic diseases such as heart failure, coronary disease, and diabetes30,31,37.
Despite a number of strengths, several limitations of the current study should be acknowledged. First, although this was a large population-based study, certain patients were excluded, such as those who were immunocompromised or pregnant. Second, the long-term outcomes were only for patients hospitalized with pneumonia, and cannot necessarily be extrapolated to those patients treated for pneumonia on an outpatient basis. Last, enrollment was limited to a single large health region in 1 country, which may limit the generalizability of our findings.
In conclusion, results of the current study demonstrate that the long-term morbidity and mortality following hospitalization for pneumonia is very high. We believe this knowledge will help physicians with prognostication, and, at the least, our findings emphasize the need for far better surveillance and closer follow-up after discharge for older patients with the highest calculated PSI scores.
REFERENCES
1. Ai-Ping C, Lee KH, Lim TK. In-hospital and 5-year mortality of patients treated in the ICU for acute exacerbation of COPD: a retrospective study.
Chest. 2005;128:518-524.
2. Aronsky D, Haug PJ, Lagor C, Dean NC. Accuracy of administrative data for identifying patients with pneumonia.
Am J Med Qual. 2005;20:319-328.
3. Basi SK, Marrie TJ, Huang JQ, Majumdar SR. Patients admitted to hospital with suspected pneumonia and normal chest radiographs: epidemiology, microbiology, and outcomes.
Am J Med. 2004;117:305-311.
4. Brancati FL, Chow JW, Wagener MM, Vacarello SJ, Yu VL. Is pneumonia really the old man's friend? Two-year prognosis after community-acquired pneumonia.
Lancet. 1993;342:30-33.
5. Buchner DM, Wagner EH. Preventing frail health.
Clin Geriatr Med. 1992;8:1-17.
6. Capital Health Home Page. Capital Health website.
http://www.capitalhealth.ca. Accessed April 8, 2008.
7. Carriere KC, Jin Y, Marrie TJ, Predy G, Johnson DH. Outcomes and costs among seniors requiring hospitalization for community-acquired pneumonia in Alberta.
J Am Geriatr Soc. 2004;52:31-38.
8. Census of Population, 2006. Statistics Canada website.
www12.statcan.ca/english/census/index.cfm. Accessed April 8, 2008.
9. Collins TC, Petersen NJ, Menke TJ, Souchek J, Foster W, Ashton CM. Short-term, intermediate-term and long-term mortality in patients hospitalized for stroke.
J Clin Epidemiol. 2003;56:81-87.
10. Discharge Abstract Database Abstracting Manual 2000-2006. Canadian Institute for Health Information website.
http://secure.cihi.ca/cihiweb/dispPage.jsp?cw_page=RC_57_E. Accessed April 8, 2008.
11. Fine MJ, Auble TE, Yealy DM, Hanusa BH, Weissfeld LA, Singer DE, Coley CM, Marrie TJ, Kapoor WN. A prediction rule to identify low-risk patients with community-acquired pneumonia.
N Engl J Med. 1997;336:243-250.
12. Fine MJ, Singer DE, Hanusa BH, Lave JR, Kapoor WN. Validation of a pneumonia prognostic index using the MedisGroups Comparative Hospital Database.
Am J Med. 1993;94:153-159.
13. Fine MJ, Stone RA, Singer DE, Coley CM, Marrie TJ, Lave JR, Hough LJ, Obrosky DS, Schulz R, Ricci EM, Rogers JC, Kapoor WN. Processes and outcomes of care for patients with community-acquired pneumonia: results from the Pneumonia Patient Outcomes Research Team (PORT) cohort study.
Arch Intern Med. 1999;159:970-980.
14. Fry A, Shay DK, Holman RC, Curns AT, Anderson LJ. Trends in hospitalizations for pneumonia among persons aged 65 years or older in the United States, 1988-2002.
JAMA. 2005;294:2712-2719.
15. Halm E, Fine MJ, Marrie TJ, Coley CM, Kapoor WN, Obrosky DS, Singer DE. Time of clinical stability in patients hospitalized with community-acquired pneumonia: implications for practice guidelines.
JAMA. 1998;279:1452-1457.
16. Hedlund JU, Ortqvist AB, Kalin ME, Granath F. Factors of importance for the long term prognosis after hospital treated pneumonia.
Thorax. 1993;48:785-789.
17. Jiang HX, Majumdar SR, Dick DA, Moreau M, Raso J, Otto DD, Johnston DW. Development and initial validation of a risk score for predicting in-hospital and 1-year mortality in patients with hip fractures.
J Bone Miner Res. 2005;20:494-500.
18. Johnstone J, Marrie TJ, Eurich DT, Majumdar SR. Effect of pneumococcal vaccination in hospitalized adults with community-acquired pneumonia.
Arch Intern Med. 2007;167:1938-1943.
19. Kaplan V, Clermont G, Griffin MF, Kasal J, Watson RS, Linde-Zwirble WT, Angus DC. Pneumonia: still the old man's friend?
Arch Intern Med. 2003;163:317-323.
20. Koivula I, Sten M, Makela PH. Prognosis after community-acquired pneumonia in the elderly: a population-based 12-year follow-up study.
Arch Intern Med. 1999;159:1550-1555.
21. Lave JR, Fine MJ, Sankey SS, Hanusa BH, Weissfeld LA, Kapoor WN. Hospitalized pneumonia: outcomes, treatment patterns and costs in urban and rural areas.
J Gen Intern Med. 1996;11:415-421.
22. Leslie DL, Marcantonio ER, Zhang Y, Leo-Summers L, Inouye SK. One-year health care costs associated with delirium in the elderly populations.
Arch Intern Med. 2008;168:27-32.
23. Levy D, Kenchaiah S, Larson MG, Benjamin EJ, Kupka MJ, Ho KK, Murabito JM, Vasan RS. Long-term trends in the incidence of and survival with heart failure.
N Engl J Med. 2002;347:1397-1402.
24. Majumdar SR, McAlister FA, Eurich DT, Padwal RS, Marrie TJ. Statins and outcomes in patients admitted to hospital with community-acquired pneumonia: population based prospective cohort study.
BMJ. 2006;333:999-1001.
25. Marrie TJ. Community-acquired pneumonia in the elderly.
Clin Infect Dis. 2000;31:106.
26. Marrie TJ, Carriere KC, Jin Y, Johnson DH. Mortality during hospitalization for pneumonia in Edmonton, Alberta is associated with physician volume.
Eur Respir J. 2003;22:148-155.
27. Marrie TJ, Lau CY, Wheeler SL, Wong CJ, Vandervoort MK, Feagan BG. A controlled trial of a critical pathway for treatment of community acquired pneumonia.
JAMA. 2000;283:749-755.
28. Marrie TJ, Michalyk D, Man P. A critical pathway for treating community-acquired pneumonia.
Can J Contin Med Educ. 2001;13:43-59.
29. Marrie TJ, Wu LL. Factors influencing in-hospital mortality in community acquired pneumonia: a prospective study of patients not initially admitted to the ICU.
Chest. 2005;127:1260-1270.
30. McAlister FA, Lawson FME, Teo KK, Armstrong PW. Randomised trials of secondary prevention programmes in coronary heart disease: systematic review.
BMJ. 2001;323:957-962.
31. McAlister FA, Stewart S, Ferrua S, McMurray JJ. Multidisciplinary strategies for the management of heart failure patients at high risk for admission: a systematic review of randomized trials.
J Am Coll Cardiol. 2004;44:810-819.
32. Mokdad AH, Marks JS, Stroup DF, Gerberding JL. Actual causes of death in the United States, 2000.
JAMA. 2004;291:1238-1245.
33. Mortensen EM, Kapoor WN, Chang CH, Fine MJ. Assessment of mortality after long-term follow-up of patients with community-acquired pneumonia.
Clin Infect Dis. 2003;37:1617-1624.
34. Ortqvist A, Sterner G, Nilsson JA. Severe community-acquired pneumonia: factors influencing need of intensive care treatment and prognosis.
Scand J Infect Dis. 1985;17:377-386.
35. Rockwood K. Frailty and its definitions: a worthy challenge.
J Am Geriatr Soc. 2005;53:1069-1070.
36. Rockwood K, Fox RA, Stolee P, Robertson D, Beattie BL. Frailty in elderly people: an evolving concept.
Can Med Assoc J. 1994;150:489-495.
37. Shojania KG, Ranji SR, McDonald KM, Grimshaw JM, Sundaram V, Rushakoff RJ, Owens DK. Effects of quality improvement strategies for type 2 diabetes on glycemic control.
JAMA. 2006;296:427-440.
38. Skull SA, Andrews RM, Byrnes GB, Campbell DA, Kelly HA, Brown GV, Nolan TM. ICD-10 codes are a valid tool for identification of pneumonia in hospitalized patients aged 65 years.
Epidemiol Infect. 2008;136:232-240.
39. Stewart S, Marley JE, Horowitz JD. Effects of a multidisciplinary, home-based intervention on planned readmission and survival among patients with chronic congestive heart failure: a randomized controlled study.
Lancet. 1999;354:1077-1083.
40. Waterer GW, Kessler LA, Wunderink RG. Medium-term survival after hospitalization with community-acquired pneumonia.
Am J Crit Care. 2004;169:910-914.
41. Yende S, Angus DC, Ali IS, Somes G, Newman AB, Bauer D, Garcia M, Harris TB, Kritchevsky SB. Influence of co-morbid conditions on long-term mortality after pneumonia in older people.
J Am Geriatr Soc. 2007;55:518-525.
42. Zweig S, Lawhorne L, Post R. Factors predicting mortality in rural elderly hospitalized for pneumonia.
J Fam Pract. 1990;30:153-159.
