Predictors of In-Hospital vs Postdischarge Mortality in Pneumonia

CHEST Original Research Predictors of In-Hospital vs Postdischarge Mortality in Pneumonia Mark L. Metersky, MD, FCCP; Grant Waterer, MBBS; Wato Nsa, MD, PhD; and Dale W. Bratzler, DO, MPH CHEST INFECTIONS Background: Many patients who die within 30 days of admission to the hospital for pneumonia die after discharge. Recently, 30-day mortality for patients with pneumonia became a publicly reported performance measure, meaning that hospitals are, in part, being measured based on how the patient fares after discharge from the hospital. This study was undertaken to determine which factors predict in-hospital vs postdischarge mortality in patients with pneumonia. Methods: This was a retrospective analysis of a database of 21,223 patients on Medicare aged 65 years and older admitted to the hospital between 2000 and 2001. Multivariate logistic regression analyses were performed to determine the association between 26 patient characteristics and the timing of death (in-hospital vs postdischarge) among those patients who died within 30 days of hospital admission. Results: Among the 21,223 patients, 2,561 (12.1%) died within 30 days of admission: 1,343 (52.4%) during the hospital stay, and 1,218 (47.6%) after discharge. Multivariate logistic regression demonstrated that seven factors were significantly associated with death prior to discharge: systolic BP, 90 mm Hg, respiration rate. 30/min, bacteremia, arterial ph, 7.35, BUN level. 11 mmol/l, arterial P O 2, 60 mm Hg or arterial oxygen saturation, 90%, and need for mechanical ventilation. Some underlying comorbidities were associated with a nonstatistically significant trend toward death after discharge. Conclusions: Of elderly patients dying within 30 days of admission to the hospital, approximately one-half die after discharge from the hospital. Comorbidities, in general, were equally associated with death in the hospital and death after discharge. CHEST 2012; 142(2):476 481 Abbreviations: CAP 5 community-acquired pneumonia; CMS 5 Centers for Medicare & Medicaid Services; CURB-65 5 confusion, uremia, respiratory rate, BP, age 65 y; Sp o 2 5 oxygen saturation as measured by pulse oximetry 476 Community-acquired pneumonia (CAP) remains a frequent cause of morbidity and mortality. It is the leading infectious disease cause of death in the world and the third leading cause of death overall. 1 In addition to the deaths within the hospital, patients admitted to the hospital with pneumonia are at an increased risk of death for months to years after discharge, relative to age-matched control subjects. 2-4 Because of the high incidence and impact, significant efforts have been expended over the past decade by payers, regulatory agencies, and health-care providers to improve the care and outcomes of patients with CAP. Efforts to decrease mortality are complicated by the fact that up to 50% of CAP-associated mortality is not directly due to the infection, as cardiovascular complications and death due to other comorbidities cause a substantial proportion of CAPassociated mortality. 5,6 It is believed that these deaths may be the more common cause of death in patients who die after discharge from the hospital. 5 Furthermore non-pneumonia-related mortality tends to occur much later than pneumonia-related mortality, with the vast majority occurring well after 7 days. 7 There are various tools that reliably predict 30-day mortality risk for patients with CAP. These include the pneumonia severity index 8 ; confusion, uremia, respiratory rate, BP, age 65 y (CURB-65) 9 ; and, more recently, a model based on administrative data used by the Centers for Medicare & Medicaid Services (CMS) to identify high- and low-performing hospitals. 10 These results are publicly available on the CMS Hospital Compare website. Since the median hospital length of stay for patients on Medicare admitted for pneumonia is 5 days, 11 hospitals are being measured in Original Research

part based on what happens after patients leave the hospital. In addition to patient-related factors, 30-day mortality rate may be related in part to the quality of care in the hospital but may also be dependent on the quality of health care received after discharge (either at home or in other institutions), circumstances over which hospitals may have little control. This study was performed to determine if patient characteristics can help distinguish those who are at risk for mortality before vs after discharge from the hospital. Such knowledge might help physicians and hospitals select high-risk patients for specific interventions and might be the initial step in creating methodologies to determine if a hospital s postdischarge mortality rates are related to identifiable risk factors or to problems with postdischarge quality of care. These data could also inform the discussion 12 as to whether 30-day mortality rate is an appropriate measure of hospital quality, since hospitals may have little control over care after the initial acute care hospitalization. Sample Selection Materials and Methods Data analyzed in this study were part of the Medicare National Pneumonia Project, a component of CMS s Quality Improvement Program. As such, Human Subjects Committee approval is not required. Eligible patients were fee-for-service Medicare beneficiaries aged 65 years who had been discharged from the hospital during calendar years 2000 and 2001 with a principal diagnosis of pneumonia ( International Classification of Disease, Ninth Edition, Clinical Modification codes 480.0-483.99, 485-486.99, or 487) or a principal diagnosis of septicemia or respiratory failure (codes 038.XX or 518.81) and a secondary diagnosis of pneumonia, as defined here. Cases were considered for inclusion if the admission note demonstrated that the physician s working diagnosis was pneumonia at the time of admission. Cases were excluded if the patient was younger than 65 years of age; there was not a confirmatory Manuscript received September 19, 2011; revision accepted January 4, 2012. Affiliations: From the Division of Pulmonary and Critical Care Medicine (Dr Metersky), University of Connecticut School of Medicine, Farmington, CT; University of Western Australia (Dr Waterer), Perth, WA, Australia; Oklahoma Foundation for Medical Quality (Drs Nsa and Bratzler); and the University of Oklahoma Health Sciences Center (Dr Bratzler), College of Public Health, Oklahoma City, OK. Funding/Support: The analyses upon which this publication is based were performed under funding by the Centers for Medicare & Medicaid Services, an agency of the US Department of Health and Human Services [Contract Number HHSM-500-2008-OK9THC]. Correspondence to: Mark L. Metersky, MD, FCCP, Division of Pulmonary and Critical Care Medicine, University of Connecticut Health Center, 263 Farmington Ave, Farmington, CT 06030-1321; e-mail: Metersky@nso.uchc.edu 2012 American College of Chest Physicians. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians. See online for more details. DOI: 10.1378/chest.11-2393 chest radiograph indicating a new infiltrate; missing data elements prevented analysis; or if the patient was transferred from another acute care facility, was discharged on the day of admission, or had been discharged from an acute care facility during the previous 14 days. Patients who left the hospital against medical advice or who were admitted for comfort measures only were also excluded, as were patients who had a history of HIV/AIDS or organ transplant. If there was more than one admission for pneumonia during the sampling period, only the first was included. Patients who were discharged or died on the day of admission and those who stayed in the hospital for. 30 days were also excluded from this study. Data Collection Each medical record was subjected to explicit review using an electronic data collection instrument with predefined instructions. Patient demographics, comorbidities, physical and laboratory findings, type and timing of antibiotic treatment, and selected outcomes were collected by experienced abstractors in the Medicare Clinical Data Abstraction Center. Most of the clinical variables were derived from the pneumonia severity index of Fine et al. 8 Comorbid illnesses were defined based on a documented history as opposed to examination or laboratory findings. To test data reliability, 957 charts were subjected to duplicate abstraction. k Statistics were calculated for pneumonia confirmation and exclusion, clinical characteristics, and blood culture results and ranged from 0.58 (chest radiograph showed pneumonia) to 1.00 (transfer from another acute care hospital) for exclusion criteria, and from 0.45 (liver disease) to 0.96 (systolic BP) for clinical characteristics. Statistical Analysis Descriptive statistics of the clinical and demographic characteristics of the entire cohort of patients (N 5 21,223) were calculated. A randomly chosen half sample (n 510,611), the exploratory sample, was used to develop a predictive model based on the logistic regression analysis. The other half sample (n 5 10,612), the validation sample, was used to verify how well the predictive model from the exploratory sample could produce similar results when applied to a different sample of patients. Three multivariate logistic regression analyses were performed to examine the relationship between patient characteristics and the timing of death among those patients who died within 30 days of hospital admission (in-hospital vs after discharge from hospital). The first logistic regression was designed to identify factors that were associated with in-hospital death as compared with death after discharge, using the subset of patients who died during either period (n 5683 and 620 for in-hospital and after-hospital death, respectively). This initial model included all 26 patient characteristics available in the data set. The second logistic regression was performed on the same subset of deceased patients, but the model only included the seven factors that were identified from the first model. The third logistic regression model was designed to apply the seven-factor model to the entire exploratory sample (n 5 10,611) and examine how the seven factors behave in a population that includes both deceased and surviving patients, as the prediction model would be most useful if it could be applied to all admitted patients. The parameter estimates from the third logistic regression analysis were used to approximate the contribution of each factor to the model. A score was assigned to each of the seven factors by converting the parameter estimates into more manageable values, which were obtained by multiplying the parameter estimates by 10 and rounding them into whole integers. Specifically, we assigned 13 points to mechanical ventilation, 13 points to bacteremia, 5 points to respiration rate. 30/min, 7 points to systolic BP, 90 mm Hg, 8 points to arterial ph, 7.35, journal.publications.chestnet.org CHEST / 142 / 2 / AUGUST 2012 477

10 points to BUN level. 11 mmol/l, and 4 points if there was documentation of either an arterial P o 2, 60 mm Hg or oxygen saturation as measured by pulse oximetry (Sp o 2 ), 90. For each patient, a composite score was calculated by summing the scores of factors that were present in the patient. The minimum score was zero if no factors were present, and the maximum was 60 if all seven factors were present. The composite was then applied to the validation sample to demonstrate its ability to produce similar results in a different sample of patients. The in-hospital death rate was calculated for each level of the composite score for both samples. The exploratory and the validation samples were stratified into three outcome groups: those who died in the hospital, those who survived the hospitalization but died within 30 days of the day of admission, and those who survived. 30 days after the day of admission; and summary statistics (mean, median, SD, and SE) of the composite score of the three groups of patients were calculated for both samples. Analyses were conducted using the software package Statistical Analysis System (SAS version 9.2, SAS Institute). Statistical significance was accepted at P,.05. Results There were 37,123 patients included in the initial sample. The most common reasons for exclusion were Table 1 Patient Demographic and Clinical Characteristics Patient Characteristics No., N 5 21,223 Death During Hospital Stay, Death Between Hospital Discharge and 30 d After Admission, Survival to 30 d After Hospital Admission, n 51,343 (6.3) n 51,218 (5.7) n 518,662 (87.9) White 18,658 1,140 (6.1) 1,103 (5.9) 16,415 (88.0) Black 1,298 100 (7.7) 66 (5.1) 1,132 (87.2) Other 1,267 103 (8.1) 49 (3.9) 1,115 (88.0) Age 65-74 y 5,930 231 (3.9) 210 (3.5) 5,489 (92.6) Age 75-84 y 8,990 554 (6.2) 444 (4.9) 7,992 (88.9) Age 85 y 6,303 558 (8.9) 564 (8.9) 5,181 (82.2) Female sex 11,351 685 (6.0) 591 (5.2) 10,075 (88.8) Male sex 9,872 658 (6.7) 627 (6.4) 8,587 (87.0) From nursing home 4,304 563 (13.1) 524 (12.2) 3,217 (74.7) ICU admission within 24 h 2,119 347 (16.4) 136 (6.4) 1,636 (77.2) Mechanical ventilation 485 157 (32.4) 23 (4.7) 305 (62.9) Bacteremia 1,372 327 (23.8) 104 (7.6) 941 (68.6) First antibiotic within 8 h 18,470 1,148 (6.2) 1,047 (5.7) 16,275 (88.1) History of neoplasm 1,565 139 (8.9) 152 (9.7) 1,274 (81.4) History of liver disease 268 25 (9.3) 18 (6.7) 225 (84.0) History of congestive heart failure 7,174 535 (7.5) 520 (7.2) 6,119 (85.3) History of cerebrovascular disease 2,920 219 (7.5) 181 (6.2) 2,520 (86.3) History of chronic renal disease 954 84 (8.8) 58 (6.1) 812 (85.1) Altered mental status 4,647 571 (12.3) 447 (9.6) 3,629 (78.1) Respiration rate. 30/min 4,067 474 (11.7) 335 (8.2) 3,258 (80.1) Systolic BP, 90 mm Hg 736 162 (22.0) 54 (7.3) 520 (70.7) Temperature, 35 C or. 40 C 405 54 (13.3) 24 (5.9) 327 (80.7) Pulse. 125/min 2,078 222 (10.7) 149 (7.2) 1,707 (82.1) Arterial ph, 7.35 1,188 244 (20.5) 88 (7.4) 856 (72.1) BUN. 11 mmol/l 6,255 815 (13.0) 600 (9.6) 4,840 (77.4) Serum sodium, 130 mmol/l 1,215 83 (6.8) 79 (6.5) 1,053 (86.7) Glucose. 14 mmol/l 1,328 147 (11.1) 89 (6.7) 1,092 (82.2) Hematocrit, 30% 1,863 171 (9.2) 163 (8.7) 1,529 (82.1) Arterial P o 2, 60 mm Hg or Sp o 2, 90% 6,081 492 (8.1) 386 (6.3) 5,203 (85.6) Pleural effusion 5,531 472 (8.5) 394 (7.1) 4,665 (84.3) PSI class 2 a 1,437 12 (0.8) 13 (0.9) 1,412 (98.3) PSI class 3 4,779 79 (1.7) 78 (1.6) 4,622 (96.7) PSI class 4 10,353 461 (4.5) 578 (5.6) 9,314 (90.0) PSI class 5 4,654 791 (17.0) 549 (11.8) 3,314 (71.2) CRB-65 class 1 b 13,217 468 (3.5) 552 (4.2) 12,197 (92.3) CRB-65 class 2 6,639 579 (8.7) 504 (7.6) 5,556 (83.7) CRB-65 class 3-4 1,367 296 (21.7) 162 (11.9) 909 (66.5) CURB-65 class 1 c 6,206 115 (1.9) 144 (2.3) 5,947 (95.8) CURB-65 class 2 9,360 448 (4.8) 531 (5.7) 8,381 (89.5) CURB-65 class 3 4,566 507 (11.1) 398 (8.7) 3,661 (80.2) CURB-65 class 4-5 1,091 273 (25.0) 145 (13.3) 673 (61.7) Data are presented as No. or No. (%). CRB 5 confusion, respiratory rate, BP; CURB-65 5 confusion, uremia, respiratory rate, BP, age 65 y; PSI 5 pneumonia severity index; Sp o 2 5 oxygen saturation as measured by pulse oximetry. a Note that there are no patients in PSI class 1, as all patients were 65 y of age. b Note that there are no patients in CRB-65 class 0, as all patients were 65 y of age. c Note that there are no patients in CURB-65 class 0, as all patients were 65 y of age. 478 Original Research

the lack of a working diagnosis of pneumonia at the time of admission (4,114), age younger than 65 years (3,478), lack of a confirmatory chest radiograph (3,241), prior hospital discharge within 14 days (1,548), and admission for comfort measures only (1,505). After all exclusions were applied, 21,223 cases were included in the analysis. Of the 2,561 patients (12.1%) who died within 30 days of admission, 1,343 (52.4%) died during the hospital stay and 1,218 (47.6%) between the time of discharge and 30 days. Table 1 demonstrates the characteristics of patients who died in the hospital, those who died after discharge from the hospital and before 30 days from admission, and those who survived at least 30 days from admission. Table 2 demonstrates the risk of death in the hospital vs death between the day of discharge and 30 days from admission, associated with the predictive factors. For each factor, an OR of 1.0 means that the factor was equally associated with risk of death in the hospital and after discharge, whereas an OR. 1.0 means the factor predicted that death was more likely prior to discharge than after discharge. Seven factors that reflect acute severity of illness in patients with pneumonia predicted that death was more likely to occur in the hospital. As can be seen, none of the baseline patient factors was significantly associated with the timing of death, although for several, there was a trend toward predicting death after discharge (male sex, admission from a nursing facility, history of neoplasm, history of heart failure, and anemia). The same seven factors remained significantly asso ciated with in-hospital death when all other variables were removed from the model (data not shown). When the seven-factor model was applied to the entire exploratory sample, including deceased and surviving patients, all seven factors still remained significantly associated with in-hospital mortality, although the values of the ORs and parameter estimates changed to reflect the change in the population of patients ( Table 3 ). In a three-level grouping, the composite score was much higher among patients who died in the hospital than those who died after discharge and those who survived 30 days after admission ( Fig 1 ), with mean values (SD) of 16.1 (11.8), 10.2 (10.0), and 6.0 (4.0), respectively, in the exploratory sample and almost identical results in the validation sample (results not shown). However, there was a great deal of overlap between the three groups, limiting the usefulness of the composite score. Discussion Although risk factors for mortality in patients with pneumonia have been investigated extensively, there have been few studies comparing patient-specific Table 2 Potential Predictors of Inpatient Mortality vs Postdischarge Mortality Within 30 d of Admission Potential Predictors OR (95% CI) P Value Intercept n/a.279 Age by 5-y increment 1.01 (0.93-1.09).901 Black vs white 1.12 (0.65-1.92).686 Other race/ethnicity vs white 1.29 (0.76-2.19).341 Male vs female 0.86 (0.67-1.10).223 From nursing home 0.84 (0.64-1.09).182 ICU admit within 24 h 1.31 (0.90-1.91).155 Mechanical ventilation 4.31 (2.02-9.19),.001 Bacteremia 2.57 (1.80-3.66),.001 First antibiotic within 8 h 0.91 (0.65-1.27).567 History of neoplasm 0.83 (0.56-1.24).368 History of chronic liver disease 1.12 (0.46-2.73).799 History of congestive heart failure 0.86 (0.66-1.13).282 History of cerebrovascular disease 1.23 (0.86-1.76).259 History of chronic renal disease 1.49 (0.88-2.53).140 Altered mental status 0.99 (0.76-1.28).929 Respiration rate. 30/min 1.33 (1.02-1.73).038 Systolic BP, 90 mm Hg 2.43 (1.49-3.96),.001 Temperature, 35 C or. 40 C 1.04 (0.52-2.12).904 Pulse. 125/min 1.12 (0.78-1.61).545 Arterial ph, 7.35 1.84 (1.21-2.79).004 BUN. 11 mmol/l 1.55 (1.20-2.01),.001 Serum sodium, 130 mmol/l 1.08 (0.66-1.78).758 Glucose. 14 mmol/l 1.42 (0.93-2.17).104 Hematocrit, 30% 0.83 (0.57-1.21).331 Arterial P o 2, 60 mm Hg or 1.59 (1.23-2.04),.001 Sp o 2, 90% Pleural effusion 1.14 (0.89-1.47).303 OR of inpatient vs postdischarge mortality. n/a 5 not applicable. See Table 1 legend for expansion of other abbreviations. factors for mortality before and after discharge from the hospital. In this analysis of patients on Medicare admitted to the hospital with pneumonia, we found that factors associated with the acute severity of pneumonia were predictive of in-hospital death but that the timing of death was unrelated to baseline patient demographic factors and comorbidities. Our results are somewhat surprising when one considers the results of prior studies investigating mortality from pneumonia. Capelastegui et al 13 studied 90-day mortality in patients discharged from the hospital after pneumonia. They found that baseline functional status was the most important predictor of mortality after discharge. Marrie and Wu 7 found several factors that discriminated between early and late in-hospital mortality. Like Capelastegui et al, 13 they found that baseline functional status was highly associated with later mortality. Mortensen et al 5 found that the cause of death among patients with pneumonia who died within 30 days of admission were generally directly related to pneumonia, whereas deaths between 30 and 90 days were unrelated to the pneumonia. Other studies have demonstrated excess risk from cardiovascular events for a long period of time after apparent recovery from journal.publications.chestnet.org CHEST / 142 / 2 / AUGUST 2012 479

Table 3 The Seven-Factor Predictive Model Applied to the Exploratory Sample Predictive Factors Parameter Estimate OR (95% CI) P Value Intercept 23.7175 n/a,.001 Mechanical ventilation 1.2533 3.50 (2.49-4.92),.001 Bacteremia 1.3245 3.76 (3.05-4.64),.001 Respiration rate. 30/min 0.5372 1.71 (1.43-2.05),.001 Systolic BP, 90 mm Hg 0.7498 2.12 (1.57-2.85),.001 Arterial ph, 7.35 0.7693 2.16 (1.65-2.82),.001 BUN level. 11 mmol/l 1.0170 2.76 (2.34-3.27),.001 Arterial P o 2, 60 mm Hg or Sp o 2, 90% 0.3667 1.44 (1.22-1.71),.001 See Table 1 and 2 legends for expansion of abbreviations. pneumonia.3 Although these studies investigated different time intervals from those in our study, they imply that since earlier and later mortality often have different causes, different risk factors might be identifiable. However, our results suggest that the factors that Mortensen et al 5 and others 3 have identified as a factor in late deaths increase the risk of mortality during the hospital stay to a similar extent. Yende et al 3 reported that many patients with pneumonia have persistently elevated levels of circulating markers of inflammation at the time of discharge and that these patients had a higher risk of death subsequent to discharge. Therefore, even late, apparent nonpneumonia-related deaths might have a root cause in the inflam mation caused by the pneumonia. Risk-adjusted 30-day mortality is a publicly reported CMS performance measure, which relies on patient characteristics derived from preadmission administrative data, largely reflecting comorbidities that correlate highly with 30-day mortality. 10 Our finding that comorbidities were equally predictive of in-hospital and postdischarge mortality provides indirect support for the assertion that the 30-day mortality measure Figure 1. Composite score by mortality timing. Box and whisker plot with box representing 25th and 75th percentiles, line representing median, 1 representing mean, and end bars representing maximum and minimum values. based on administrative data measures factors that are relevant to inpatient outcomes. Furthermore, since we found that patient characteristics were not predictive of predischarge vs postdischarge deaths, it is possible that across institutions, the ratio of predischarge to postdischarge deaths might be similar. This ratio might be a simple metric that could help hospitals benchmark themselves to identify whether it would be more fruitful to target inpatient care vs care transitions/postdischarge care to reduce 30-day mortality. There are limitations to this study. These analyses are based on data that are approximately 10 years old. There have been significant changes in pneumonia care and outcomes since 2001. For example, the 30-day mortality of Medicare patients admitted to the hospital with pneumonia has declined from 16.3% in 2000 to 11.1% during 2006 to 2009. 14,15 However, we are not aware of any other database that is as large and contains the same detailed clinical and demographic data as the one we used. Thus, it is unlikely that a similar powered study could be done with more current data. We could only study those variables that had been abstracted from the chart. Potentially important predictors, such as vaccine status, functional status, tobacco use, and coronary artery disease, and most treatment-related variables were not included. In addition, the patients were all on Medicare, so younger patients are not represented. We did not study hospital characteristics, which may be important determinants of patient outcomes. However, the goal of the study was to study patient-related factors. Despite the limitations of the study, there are several strengths as well. Our data came from a large sample of patients with pneumonia, thereby allowing robust comparisons. Use of a national database avoids the potential bias that could be introduced by the inclusion of patients from only one hospital or small group of hospitals. In addition, the data were abstracted by experienced abstractors with the use of rigorous quality-control procedures. In summary, our data provide important messages for clinicians and administrators trying to improve the outcome of CAP. Among elderly patients admitted to the hospital with pneumonia, of those who died 480 Original Research

within 30 days of admission, approximately one-half died after discharge. We have demonstrated that, in general, comorbidities are equally important predictors of mortality before discharge and during the immediate postdischarge period. Clinicians, those involved with care coordination/discharge planning, and quality improvement officers should be aware of these results when considering interventions to decrease postdischarge mortality in patients with pneumonia. Acknowledgments Author contributions: Dr Metersky is responsible for the integrity of the manuscript and is the guarantor. Dr Metersky: contributed to study design, data analysis, and drafting the manuscript. Dr Waterer: contributed to study design and drafting the manuscript. Dr Nsa: contributed to study design, data analysis, critical review, and revision of the manuscript. Dr Bratzler: contributed to obtaining funding, data analysis, critical review, and revision of the manuscript. Financial/nonfinancial disclosures: The authors have reported to CHEST the following conflicts of interest: Dr Metersky has served as a consultant to the Centers for Medicare & Medicaid Services and to Qualidigm (Connecticut s Medicare Quality Improvement Organization) on various quality improvement and patient safety initiatives. His employer has received remuneration for some of these activities. Drs Waterer, Nsa, and Bratzler have reported that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. Role of sponsors: The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the US government. The sponsors had no role in the design of the study, the collection and analysis of the data, or in the preparation of the manuscript. The authors assume full responsibility for the accuracy and completeness of the ideas presented. References 1. World Health Organization. The top 10 causes of death. World Health Organization website. http:www.who.int/mediacentre/ factsheets/fs310/en/index.html. Accessed September 5, 2011. 2. Mortensen EM, Kapoor WN, Chang CC, Fine MJ. Assessment of mortality after long-term follow-up of patients with community-acquired pneumonia. Clin Infect Dis. 2003 ;37 (12 ): 1617-1624. 3. Yende S, D Angelo G, Kellum JA, et al ; GenIMS Investigators. Inflammatory markers at hospital discharge predict subsequent mortality after pneumonia and sepsis. Am J Respir Crit Care Med. 2008 ;177 (11 ):1242-1247. 4. Waterer GW, Kessler LA, Wunderink RG. Medium-term survival after hospitalization with community-acquired pneumonia. Am J Respir Crit Care Med. 2004 ;169 (8 ):910-914. 5. Mortensen EM, Coley CM, Singer DE, et al. Causes of death for patients with community-acquired pneumonia: results from the Pneumonia Patient Outcomes Research Team cohort study. Arch Intern Med. 2002 ;162 (9 ):1059-1064. 6. Ramirez J, Aliberti S, Mirsaeidi M, et al. Acute myocardial infarction in hospitalized patients with community-acquired pneumonia. Clin Infect Dis. 2008 ;47 (2 ):182-187. 7. Marrie TJ, Wu L. Factors influencing in-hospital mortality in community-acquired pneumonia: a prospective study of patients not initially admitted to the ICU. Chest. 2005 ;127 (4 ): 1260-1270. 8. Fine MJ, Auble TE, Yealy DM, et al. A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med. 1997 ;336 (4 ):243-250. 9. Lim WS, van der Eerden MM, Laing R, et al. Defining community acquired pneumonia severity on presentation to hospital: an international derivation and validation study. Thorax. 2003 ;58 (5 ):377-382. 10. Bratzler DW, Normand SL, Wang Y, et al. An administrative claims model for profiling hospital 30-day mortality rates for pneumonia patients. PLoS ONE. 2011 ;6 (4 ):e17401. 11. Centers for Medicare & Medicaid Services. Medicare Hospital Quality Chartbook. Centers for Medicare & Medicaid Services website. http://www.cms.gov/hospitalqualityinits/20_outcome Measures.asp. Accessed November 1, 2011. 12. Penfold RB, Dean S, Flemons W, Moffatt M. Do hospital standardized mortality ratios measure patient safety? HSMRs in the Winnipeg Regional Health Authority. Healthc Pap. 2008 ;8 (4 ):8-24. 13. Capelastegui A, España PP, Quintana JM, et al. Development of a prognostic index for 90-day mortality in patients discharged after admission to hospital for community-acquired pneumonia. Thorax. 2009 ;64 (6 ):496-501. 14. Metersky ML. Should management of pneumonia be an indicator of quality of care? Clin Chest Med. 2011 ;32(3):575-589. 15. Lindenauer PK, Bernheim SM, Grady JN, et al. The performance of US hospitals as reflected in risk-standardized 30-day mortality and readmission rates for medicare beneficiaries with pneumonia. J Hosp Med. 2010 ;5 (6 ):E12-E18. journal.publications.chestnet.org CHEST / 142 / 2 / AUGUST 2012 481