Processes and Outcomes of Care for Patients With Community-Acquired Pneumonia:  Results From the Pneumonia Patient Outcomes Research Team (PORT) Cohort Study FREE

COMMUNITY-acquired pneumonia (CAP) affects more than 4 million adults and accounts for more than 1 million hospital admissions each year in the United States.^1- 2 Pneumonia and influenza are the sixth leading cause of death in this country, and age-adjusted mortality attributed to these illnesses is increasing.³ Community-acquired pneumonia is responsible for 64 million days of restricted activity, 39 million days of bed confinement, and 10 million days of work loss annually.⁴

Knowledge of current processes of care is a basic requirement for improving the quality of medical treatment for patients with CAP. Although prior studies^5- 9 of CAP have described the clinical characteristics and microbiologic causes of their study populations based on standard investigational protocols, many processes of care, such as how often laboratory or microbiologic tests are routinely performed, have, to our knowledge, not been described.

Understanding medical outcomes of CAP is important for quality improvement efforts, as well as patient education and satisfaction with care, because it allows physicians to inform their patients about the likelihood of potential complications, natural history of symptom resolution, and expected time to return to usual activities.¹⁰ Few previously published reports of prognosis for CAP have assessed outcomes other than all-cause mortality,¹¹ and few have studied patients treated in the ambulatory setting.^12- 17

As part of the Pneumonia Patient Outcomes Research Team (PORT) project, we conducted a multicenter, prospective, observational study to accomplish 2 objectives: to describe the processes of care and to assess a broad range of medical outcomes for ambulatory and hospitalized patients with CAP. The processes of care, or components of the encounter between a patient and medical care provider, described in this study consist of baseline laboratory and microbiologic tests, length of hospital stay, and treatment within a specialty care unit (eg, intensive care unit) for inpatients.¹⁸

The Pneumonia PORT prospective cohort study was conducted from October 1991 through March 1994 at 3 university hospitals, 1 community teaching hospital, and 1 group-model health maintenance organization, in Pittsburgh, Pa, Boston, Mass, and Halifax, Nova Scotia; methods and criteria have been previously described.¹⁹ This study of immunocompetent and immunocompromised human immunodeficiency virus–negative patients with CAP was approved by the biomedical institutional review boards of all study sites.

Enrolled patients initially treated in an ambulatory setting, including those who were subsequently admitted to a hospital, were classified as outpatients. All patients admitted to a hospital on presentation to a study site facility were classified as inpatients. Both outpatients and inpatients were enrolled from each of the 4 hospital sites; only outpatients were enrolled from the single health maintenance organization study site.

During the study enrollment period, more than 12,500 potential cases of CAP were screened to identify 3964 potential participants who satisfied all eligibility criteria. Overall, 2287 eligible patients (57.7%) were enrolled; 1343 (55.5%) of 2418 eligible inpatients and 944 (61.1%) of 1545 eligible outpatients (site of care was unknown for 1 eligible patient who was not enrolled). The leading reasons for nonenrollment of the 1677 eligible patients were patient or physician refusal (44.1%) and inability to contact the patient within the 6-day period for enrollment (34.8%). Enrollment of eligible patients varied across sites, ranging from 76.0% in Halifax to 45.3% for the 2 Pittsburgh sites. The lower enrollment rate for the Pittsburgh sites resulted from our exclusion of all data collected for 185 patients by 1 research assistant who failed to adhere to study protocols. Analysis of data for age, race, sex, and baseline severity of illness collected independently for these patients provided no evidence that they were systematically different from the remaining patients enrolled from the Pittsburgh study sites (data not shown).

To assess potential selection biases, the same data (age, race, sex, and baseline severity of illness determined by 19 clinical variables from presentation used to determine the risk of short-term mortality in CAP²⁰) were collected for eligible patients who were not enrolled in the study. In comparison with eligible patients who were not enrolled, enrolled patients were slightly younger (mean age, 56 vs 61 years; P<.001), more likely to be white (85.2% vs 76.8%; P<.001), and more likely to be classified as low risk (ie, predicted probability of less than .04) for short-term mortality (68.9% vs 57.3%; P<.001).

For all study patients, baseline sociodemographic characteristics and clinical data were assessed at presentation to a study site by patient interview and review of medical records. A proxy respondent was sought from the patients' caregivers if a patient was unable to participate in a direct interview because of physical barriers (eg, mechanical ventilation), language barriers, or mental confusion.

Clinical data collected included the presence of symptoms and physical examination findings at presentation, comorbid illnesses, and results of pertinent laboratory tests performed within 48 hours of presentation. Five respiratory symptoms (cough, dyspnea, sputum production, pleuritic chest pain, and hemoptysis) and 14 nonrespiratory symptoms (fatigue, fever, anorexia, chills, sweats, headache, myalgia, nausea, sore throat, confusion, inability to eat, vomiting, diarrhea, and abdominal pain) were assessed. During the baseline interview, patients or proxies were also asked to report retrospectively the presence of 5 common CAP-related symptoms (cough, dyspnea, sputum production, pleuritic chest pain, and fatigue) for the month before the onset of illness with CAP. Physical examination data collected at presentation included vital signs and an evaluation of mental status. Laboratory data collected included white blood cell count, hematocrit, blood urea nitrogen level, serum sodium level, liver enzyme levels, arterial blood gas levels, and pulse oximetry.

Severity of illness at presentation was quantified by means of a validated prediction rule for 30-day mortality and medical complications in patients with CAP.²¹ This rule is based on 3 demographic characteristics, 5 comorbid illnesses, 5 physical examination findings, and 7 laboratory and radiographic findings from the time of presentation. The rule classified patients into 5 risk classes with 30-day observed mortality ranging from 0.1% for class 1 (lowest risk) to 27.0% for class 5 (highest risk).

Copies of initial chest radiographs used for the diagnosis of pneumonia at each study site were independently reviewed by a 3-member panel of staff radiologists at the University of Pittsburgh who had no patient-specific clinical information. For all patients with an infiltrate confirmed as possible, probable, or definite by 2 of the panel radiologists, further characterization of the radiograph was performed by 1 panel member using a standardized data collection form. A prior analysis of 232 chest radiographs characterized independently by 2 panel members demonstrated fair to good interrater reliability for the presence of infiltrate (κ=0.37), number of pulmonary lobes with infiltrate (κ=0.51), and pleural effusion (κ=0.46).²² Pleural effusion was quantified by the maximum present in either lung as follows: none, minimal (costophrenic angle blunting only), moderate (less than one third of the pleural space), and large (one third or more of the pleural space). The pattern of infiltrate was categorized as predominantly alveolar, predominantly interstitial, miliary, or mixed alveolar and interstitial.^23- 24 The extent of radiographic infiltrate was determined by the number of pulmonary lobes with any identifiable infiltrate.

Data were collected through chart review for the following microbiologic tests, when ordered at the discretion of the physicians managing these patients: sputum Gram stains and bacterial cultures obtained within 2 days of presentation, blood cultures drawn before initiating antimicrobial therapy, pleural fluid cultures, and acute (within 1 week of presentation) and convalescent (from 1 to 8 weeks after presentation) serologic tests. Results of these tests were reviewed, and a microbiologic cause was assigned independently by 2 of 5 clinical investigators (C.M.C., M.J.F., W.N.K., T.J.M., D.E.S.), using previously published criteria.⁸ The 5-member committee discussed all discrepancies between the initial 2 reviewers, and the final assignment of cause was based on the unanimous consensus opinion of the full committee.

A degree of certainty (definite or presumptive) was assigned to each etiologic diagnosis. Assignment of a definite microbiologic cause required positive blood or pleural fluid cultures or a 4-fold rise in an IgG antibody titer for Legionella species, Mycoplasma pneumoniae, or Chlamydia pneumoniae. A presumptive microbiologic cause was based on heavy or moderate growth of a predominant bacterial pathogen in sputum culture or light growth of a bacterial pathogen in sputum culture confirmed by the presence of a compatible predominant organism in a sputum Gram stain. Definite and presumptive causes were combined for reporting purposes. When 2 or more microbiologic causes were coded as definite or presumptive pathogens for a patient, the cause was classified as multiple pathogens. A patient was considered to have pneumonia of unknown cause if no diagnostic tests were performed or tests were performed but test results did not meet criteria for assigning microbiologic cause.

Based on radiographic data and synopses of clinical data, diagnoses of presumptive aspiration pneumonia and postobstructive pneumonia were also assigned by the clinical committee. Presumptive aspiration pneumonia was diagnosed in patients with a disorder known to alter consciousness, the normal gag reflex, or swallowing mechanism in whom the chest radiograph revealed an infiltrate involving the superior or basilar segments of the lower lobes or the posterior segments of the upper lobes.²⁵ Presumptive postobstructive pneumonia was diagnosed radiographically or bronchoscopically in the presence of a pulmonary infiltrate located distal to an obstructing neoplasm.

The extent of laboratory and microbiologic testing, length of hospital stay, number of specialty care unit (intensive care unit, coronary care unit, and telemetry) admissions, and the reason for the first admission to any specialty care unit were assessed by medical record review. In addition, the dose, route of administration, frequency, and duration of antimicrobial therapy were recorded for all agents prescribed within 30 days of presentation and are reported in detail elsewhere.²⁶ Length of hospital stay was calculated as the date of discharge minus the date of admission.

Survival at 30 days following the radiographic diagnosis of pneumonia was ascertained for all 2287 study patients. Detailed case summaries of all deaths, based on medical record review, autopsy results when available (14 of 113 patients who died underwent autopsies), and interviews with caregivers and family members, were reviewed independently by 2 of the 5 clinical investigators. The cause of death was assigned using World Health Organization criteria.²⁷ Based on these criteria, death was considered to be pneumonia related when CAP was identified as the underlying cause of death (ie, initiated the series of morbid events directly leading to death), the immediate cause of death (ie, immediately preceded death), or a major contributing cause of death (ie, neither the underlying nor immediate cause, but death would not have occurred without it). Each death was discussed by the full 5-member committee until complete agreement for causality was reached. Data concerning all new medical complications, and worsening of previously stable chronic medical conditions (eg, congestive heart failure), that occurred within 30 days of presentation were collected for 96.1% of 944 outpatients and 99.7% of 1343 inpatients by medical record review.

The same 5 CAP-related symptoms (cough, dyspnea, sputum production, pleuritic chest pain, and fatigue) assessed at baseline and retrospectively were assessed at 30 days by patient or proxy interview. Each of these symptoms are reported for all patients who were alive at 30 days and had complete data at 3 time points (retrospective, baseline, and 30 days). Data for individual symptoms were complete for 93.3% to 94.6% of the 938 outpatients and 84.9% to 91.1% of the 1236 inpatients who were alive at 30 days.

Subsequent hospitalization for outpatients and hospital readmission for inpatients were assessed at 30 days after presentation. Subsequent hospitalization for outpatients was attributed to CAP or comorbid illness, based on previously described criteria.²⁸

Additional outcome measures that were evaluated for patients alive at 30 days included return to usual activities and return to work. Initially, these data were collected for only a subset of patients; the protocol was later modified so that these data were collected for all subsequently enrolled patients. Data for return to usual activities were complete for 86.5% of the 938 outpatients and 69.0% of the 1236 inpatients who were alive at 30 days. Data for return to work were complete for 89.8% of the 539 outpatients and 75.8% of the 215 inpatients who were employed at the time of presentation and alive at 30 days.

Descriptive statistics were generated separately for outpatients and inpatients because the focus of this article is to describe processes and outcomes for these 2 groups rather than to compare them. For time-to-event outcomes (ie, return to usual activities and work), Kaplan-Meier estimated probabilities were computed, with inpatients assumed not to be at risk before hospital discharge.²⁹

Most outpatients were between 18 and 44 years of age (59.5%), had more than a high school education (58.7%), and were employed (57.2%) (Table 1). Overall, 204 outpatients (21.7%) received their care through a health maintenance or preferred provider organization, including 69 outpatients who were enrolled from hospital sites. Most inpatients (58.7%) were 65 years or older, about one third (32.9%) had more than a high school education, and relatively few (16.3%) were employed. Most US inpatients (62.0%) had Medicare insurance. Almost all the Canadian patients had Medical Services Insurance, the national health insurance provided in Nova Scotia.

Most outpatients (59.1%) had none of the comorbid conditions listed in Table 1, and 95.7% were in the 3 lowest-severity risk classes. Only 14.4% of inpatients had none of the comorbid conditions listed in Table 1, and half (50.0%) were in the 2 highest-severity risk classes.

Most patients reported multiple respiratory and nonrespiratory symptoms at presentation, with cough and fatigue being most frequently reported by both outpatients and inpatients (Table 2). Vital signs (respiratory rate, heart rate, systolic blood pressure, and temperature) were assessed for 50.5% to 72.8% of outpatients, while mental status was assessed for 99.8%. Clinically meaningful abnormalities in vital signs or mental status were observed in fewer than 5% of the outpatients for whom signs were assessed. Vital signs and mental status were assessed for almost all inpatients; tachypnea (respiratory rate, ≥30/min) was observed for 23.0%, tachycardia (heart rate, ≥125/min) for 13.0%, and altered mental status for 17.3%.

Of the 907 outpatients for whom copies of the baseline chest radiographs were available for review, pulmonary infiltrate was independently confirmed for 84.3% and pleural effusions were present in 3.3% (Table 3). Only 1 lobe was involved for most outpatients with a confirmed infiltrate (76.5%), and bilateral infiltrates were present for 12.9%. The initially identified pulmonary infiltrate was independently confirmed for 89.5% of 1276 inpatients, with pleural effusions present in 10.8%. More than 1 lobe was involved for 50.7% of those inpatients with confirmed infiltrates, and bilateral infiltrates were present for 29.6%.

Fewer than half (47.8%) of the outpatients had any of the laboratory tests listed in Table 4 performed. White blood cell count, hematocrit, and platelet count were ordered most frequently (31.4%-39.8%), followed by creatinine level, blood urea nitrogen level, glucose level, pulse oximetry, and sodium level (18.1%-19.7%). For outpatients with 1 or more laboratory tests performed, the most frequent abnormal findings were elevated white blood cell count (34.6%), hypoalbuminemia (25.0%), and elevated aspartate aminotransferase level (15.6%). Each of the laboratory tests in Table 4 was performed for most inpatients, and elevated white blood cell count and hypoalbuminemia were found in most inpatients in whom these tests were performed (57.2% and 63.0%, respectively). Among inpatients for whom these tests were performed, abnormal liver function test result (aspartate aminotransferase and alkaline phosphatase), blood urea nitrogen level, and creatinine level were observed in about one quarter (24.0%-27.6%) and abnormal PO2 in 36.7%.

Only 29.7% of outpatients had a sputum Gram stain or culture, blood or pleural fluid culture, or serologic or other diagnostic test performed to determine the cause of pneumonia (Table 4). Most inpatients (95.7%) had 1 or more tests performed; 52.5% had a sputum Gram stain, 58.0% had a sputum bacterial culture, and 71.2% had a blood culture obtained before the administration of antimicrobial therapy.

A microbiologic cause was determined for only 5.7% of outpatients, primarily because no diagnostic tests were performed in 70.3% of outpatients (Table 5). A microbiologic cause was established in only 29.6% of inpatients despite 1 or more microbiologic tests being performed for 95.7% of inpatients. The most frequently identified pathogens among both outpatients and inpatients were Streptococcus pneumoniae and Haemophilus influenzae (1.8% and 1.5%, respectively, for outpatients and 9.1% and 4.8%, respectively, for inpatients). Clinical diagnoses of presumptive aspiration (0.4%) and presumptive postobstructive pneumonia (0.2%) were rare among outpatients and were assigned in 7.5% and 3.3% of inpatients, respectively.

Seventy-one outpatients (7.5%) were subsequently hospitalized within 30 days of their initial presentation with pneumonia, including 5 patients who had initially refused hospitalization. Forty of the remaining 66 subsequent hospitalizations were determined to be pneumonia related, while the other 26 were primarily related to comorbid illness. The median length of stay for the subsequently hospitalized patients was 5 days. Of the 9 patients (13.8%) admitted to an intensive care, cardiac care, or telemetry unit during the hospitalization, 6 were admitted for monitoring only, 2 for treatment of respiratory failure, and 1 for treatment of shock.

For inpatients, the median length of stay for the index hospitalization was 8 days, with 5.2% of these patients remaining in the hospital at least 30 days. The most common discharge dispositions were return to home (71.2%), return to nursing home or chronic care facility (10.4%), and new admission to a nursing home or chronic care facility (3.9%). The 248 inpatients (18.5%) who were admitted to an intensive care, cardiac care, or telemetry unit during their initial hospital stay included 115 patients with respiratory failure of which 73 required mechanical ventilation. A total of 121 inpatients (10.1%) were rehospitalized within 30 days of their initial presentation with pneumonia.

Six outpatients (0.6%) died within 30 days (1 in risk class 1 and 5 in risk class 4); 3 deaths were determined to be CAP related (Table 6). Of the 3 deaths that were not CAP related, 2 were due to coronary artery disease and 1 to neoplastic disease. Two (66.6%) of the 6 deaths among outpatients occurred in the hospital, 1 of which was in an intensive care unit. None of the outpatients who died had a do-not-resuscitate code at baseline, but 3 were assigned this code status before death.

Overall, 107 inpatients (8.0%) died within 30 days; 81 deaths were considered to be CAP related. The risk class–specific all-cause mortality rates for inpatients ranged from 0.5% in risk class 1 to 27.1% in risk class 5. Ninety-two (86.0%) of the 107 inpatient deaths occurred in the hospital, of which 23 occurred in an intensive care unit. Fifty-one (47.7%) had a do-not-resuscitate code status at baseline; an additional 44 (41.1%) were assigned this code status before death.

For all patients who died, the leading immediate causes of death were respiratory failure (42.5%), cardiac arrhythmia (8.0%), and sepsis (5.3%).

Of the 901 outpatients who were alive at 30 days and had medical record reviews completed, most (92.0%) had none of the medical complications shown in Table 7. The most prevalent medical complications among the surviving outpatients were rash (2.0%) and vomiting or diarrhea requiring intravenous fluids (1.6%). Three of the 6 outpatients who died within 30 days had 1 or more complications.

Most inpatients alive at 30 days (69.0%) and almost all who died (94.4%) had 1 or more medical complications (Table 7). The most prevalent complications among surviving inpatients and those who died within 30 days were respiratory failure (36.5% and 65.4%, respectively) and congestive heart failure (19.0% and 42.1%, respectively). Among inpatients, 9.0% of those alive and 17.8% of those who died within 30 days had at least 1 potentially treatment-related complication. The most prevalent presumed treatment-related complication was pseudomembraneous colitis or Clostridium difficile–associated diarrhea (3.8% and 5.6% among inpatients alive and dead, respectively, at 30 days).

For outpatients and inpatients, the reported prevalence of each symptom was reduced at 30 days relative to baseline, although not to the level reported retrospectively for the month before the onset of illness with CAP (Figure 1). Most outpatients reported fatigue (56.2%) and a few reported cough (48.9%), sputum production (38.1%), or dyspnea (32.4%) at 30 days. Among outpatients with complete data on all 5 symptoms, 76.0% reported at least 1 of these symptoms at 30 days. Among inpatients, 72.6% reported fatigue and a few reported cough (47.1%), dyspnea (46.5%), or sputum production (42.3%) at 30 days (Figure 1). Most inpatients (86.5%) with complete longitudinal data on all 5 symptoms reported at least 1 of these symptoms at 30 days.

For outpatients who were not employed at presentation, the estimated probability of return to usual activities by 30 days was 88.9% (Figure 2), with a median time to return to usual activities of 7 days. Among outpatients who were employed at presentation, the median times to return to usual activities and to work were 6 days and 7 days, respectively; based on estimated probabilities, 95.6% returned to usual activities and 95.3% returned to work by 30 days. The median time to return to usual activities for inpatients who were not employed at presentation was 24 days, with an estimated probability of 57.2% of returning to usual activities by 30 days (Figure 2). For the employed inpatients, the median times to return to usual activities and to work were 15 days and 22 days, respectively, with corresponding estimated probabilities of returning to usual activities and to work by 30 days of 82.0% and 68.1%, respectively.

The Pneumonia PORT observational cohort study has several strengths that distinguish it from prior studies of CAP. In this study, processes of care instituted at the discretion of physicians in the routine management of their patients were prospectively assessed, whereas most prior studies^5- 9 used standard study protocols to obtain laboratory and microbiologic tests. In contrast to most studies¹¹ of prognosis in CAP that focused on mortality as the sole outcome measure, this study assessed a broad range of medical outcomes describing the natural history of this illness. Furthermore, to our knowledge, this is the first study to report a comprehensive assessment of processes and outcomes of medical care for a large cohort of patients treated in the ambulatory care setting, where approximately three quarters of patients with this illness are managed.

This study demonstrated that some form of laboratory testing, such as measurement of a complete blood cell count, serum electrolyte levels, and/or glucose level, was performed in nearly half of outpatients, but test results were rarely abnormal. In addition, routine microbiologic testing was infrequently performed among outpatients and rarely led to a microbiologic diagnosis. Because 4 of the 5 study sites were hospitals with affiliated medical education training programs where more aggressive diagnostic evaluation is often advocated, the 48% of outpatients who had 1 or more laboratory tests and the 30% who had 1 or more microbiologic tests may be overestimates of the actual frequency of testing that occurs in community-based medical office practices. The infrequent discovery of laboratory abnormalities among outpatients suggests that directed testing (eg, ordering a blood glucose test for a diabetic patient or a serum sodium test for a volume-depleted patient) is preferable to routine laboratory testing in this patient population. The failure to perform a microbiologic evaluation among most outpatients suggests that an empiric management strategy was the standard of care within the study population. Although testing was rarely performed and test results were rarely abnormal, this study did not directly assess the impact of routine laboratory and microbiologic testing on medical outcomes and costs of care for outpatients with CAP.

In contrast to outpatients, virtually all inpatients in this cohort had 1 or more laboratory tests performed, and more than 95% had 1 or more microbiologic tests performed at presentation. Although this study did not determine whether such processes of care had an association with improved patient outcomes, which would be the highest standard for establishing the appropriateness of processes of care, we believe that certain laboratory or microbiologic tests are useful because they help quantify prognosis and guide clinical management decisions. For example, because of the prognostic importance of leukopenia, anemia, hyponatremia, azotemia, and liver function test abnormalities,^11,20- 21 we concur with the American Thoracic Society and Infectious Diseases Society of America pneumonia guidelines that it is appropriate to order these tests for patients with CAP who require hospitalization.^30- 31 A measure of arterial oxygenation (ie, arterial blood gas or pulse oximetry), performed in more than 90% of inpatients in this study, also should be obtained in patients hospitalized with CAP. Hypoxemia at presentation is independently associated with short-term mortality^20- 21 and is a marker for respiratory failure requiring intensive care unit admission in patients with this illness.^30,32 Likewise, blood cultures are warranted by the prognostic importance of bacteremia in CAP,¹¹ and the potential for establishing a definitive microbiologic diagnosis that would allow therapy focused on the antibiotic sensitivities of the isolated organism.^30- 32 Although prior studies^33- 34 have demonstrated a low yield of positive findings for routine blood cultures and failure of physicians to modify clinical practice based on results, performance of blood cultures within 24 hours of presentation has recently been identified as a process measure that is associated with improved 30-day survival in hospitalized patients with CAP.³⁵ In the present study, only 71% of inpatients had blood cultures obtained before the initiation of antimicrobial therapy, identifying an area for quality improvement in the provision of medical care.

Most outpatients in this study had an excellent prognosis. Five of the 6 deaths and all 3 of the pneumonia-related deaths were in outpatients designated as high risk based on a validated prediction rule for prognosis in pneumonia.²¹ Only 8% of outpatients who survived had 1 or more medical complications and only 7.5% of outpatients required subsequent hospitalization. Subsequent hospitalization does not necessarily imply treatment failure for pneumonia; nearly half of the subsequently hospitalized outpatients were admitted because of underlying comorbid illness or because they refused initially recommended hospitalization.²⁸

Overall mortality for inpatients in this study was 8.0%, and nearly 70% of inpatients who survived had 1 or more medical complications. However, there are wide ranges in illness severity, comorbidity, and prognosis among inpatients. In the present study, half of inpatients were in the 3 lowest-severity risk classes, with 30-day all-cause mortality rates ranging from 0.5% to 1.2%. A previous study, as well as recently published Infectious Diseases Society of America guidelines for the management of CAP,^31,36- 37 suggests that many of these patients can be safely managed in the outpatient setting at substantially lower medical care costs. The remaining half of inpatients were in risk classes 4 and 5, with all-cause mortality rates of 9.0% and 27.1%, respectively. The reported mortality may underestimate the actual frequency of this outcome among inpatients with CAP because of the relatively lower enrollment of older and more severely ill patients in this cohort. A previously published meta-analysis¹¹ of prognosis in CAP reported a 13.6% mortality in more than 25,000 hospitalized patients.

This study confirms that pneumonia often occurs in patients with underlying comorbid illness and demonstrates that the pneumonia often results in a worsening of such underlying conditions. One or more medical complications occurred in nearly 70% of inpatients who survived and in 95% of inpatients who died. An example of the interaction between comorbidity and the acute episode of pneumonia is that nearly 17% of inpatients in this study had a history of congestive heart failure at presentation and 19% of survivors and 42% of nonsurvivors had new or worsening congestive heart failure within 30 days of presentation.

Nearly 90% of outpatients had returned to usual activities or work by the 30-day follow-up; 57% of nonworking inpatients and 82% of working inpatients returned to usual activities during this period. Physicians should inform patients and their caregivers that symptoms are likely to persist beyond the acute phase of the illness; more than three quarters of the outpatients and nearly 90% of the inpatients in this study had 1 or more pneumonia-related symptoms at 30 days. Separate analyses of a subgroup of outpatients and inpatients enrolled in this study demonstrated that 28% had dyspnea, 32% had cough, and 51% had fatigue for up to 90 days after presentation.³⁸ Failure of symptom resolution also has an impact on health care resource use. Symptom severity measured 7 and 30 days after presentation was independently associated with pneumonia-related physician office visits at both 30 and 90 days after presentation.³⁹

Based on the microbiologic tests obtained by the managing physicians in this study, only 5.7% of outpatients and 29.7% of inpatients had an assigned microbiologic pneumonia cause. These proportions are much lower than the 53% to 97% reported in studies with diagnostic protocols designed to determine pneumonia cause.^5- 9,39 However, the most common pathogen identified in this study, S pneumoniae, is the same as the leading pathogen identified in many studies of pneumonia cause.^5,8- 9,39 It is likely that atypical pathogens, including Legionella species, M pneumoniae, Coxiella burnetii, C pneumoniae, and viral species, were less commonly identified than in prior studies designed to determine pneumonia cause due to the relative infrequency with which serologic testing was performed in this observational study compared with studies using standard diagnostic protocols.^5- 9,39 The relatively low rate of assignable microbiologic diagnoses among study patients suggests that increased use of diagnostic tests and the institution of epidemiological surveillance programs similar to those established by the Centers for Disease Control and Prevention may be important in determining the complete spectrum of microbiologic pathogens responsible for causing pneumonia among hospitalized patients and identifying the emergence of antimicrobial resistance patterns among these pathogens.³

The present study has limitations that are important to acknowledge. Although the screening and recruitment process was designed to be comprehensive, all consecutive patients were not identified because of infrequent nonenrollment days throughout the 29-month study period. However, it is unlikely that this led to any systematic bias in the identification of patients with CAP from the participating study sites. Moreover, the findings may not be generalizable to all patients with CAP because the study was conducted primarily at urban hospital sites with affiliated medical education training programs. Enrollment of a substantial proportion of patients from a single Canadian hospital site may reduce the generalizability of study findings for patients treated in the United States. Furthermore, the prognosis of these study patients may not reflect the full spectrum of patients with this illness because of selection biases that resulted in decreased enrollment of relatively older and more severely ill patients who met study eligibility criteria. Finally, the process measures assessed may be institution specific and not generalizable to other health care settings. Large variations in certain processes of care, such as antimicrobial therapy, were discovered even within the hospitals participating in the present study.²⁶ It should also be emphasized that the analyses presented herein do not describe associations between process measures and outcomes, which represent the highest standard for defining appropriate processes of care.

Accepted for publication September 3, 1998.

This research was part of the Pneumonia PORT Project funded by the Agency for Health Care Policy and Research, Washington, DC (grant number R01 HS06468). Dr Fine was also supported as a Robert Wood Johnson Foundation Generalist Physician Faculty Scholar.

The authors are indebted to Michael Karpf, MD, for collaborating on study design and method; Elmer Holzinger, MD, for coordinating study activities at St Francis Medical Center; Karen Lahive, MD, for coordinating study activities at the Harvard Community Health Plan, Kenmore Center; Terry Sefcik, MS, for data management; and the following clinical research assistants for cohort study patient enrollment and data collection: Mary Walsh, RN, Donna Polenik Simpson, RN, MPH, and Kathryn Fine, RN, in Pittsburgh; Mary Ungaro, RN, Leila Haddad Borowsky, MPH, and Marian Hendershot, RN, in Boston; Rhonda Grandy, RN, Jackie Cunning, RN, Dawn Menon, GN, Linda Kraft, RN, and Maxine Young, RN, in Halifax.

Corresponding author: Michael J. Fine, MD, MSc, Montefiore University Hospital, Room E-820, 200 Lothrop St, Pittsburgh, PA 15213-2582 (e-mail: mjf1+@pitt.edu).