Pleural Effusions and Mortality Risk from Pneumonia


Source: Hasley PB, et al. Do pulmonary radiographic findings at presentation predict mortality in patients with community-acquired pneumonia? Arch Intern Med 1996;156:2206-2212.

The era of managed care is upon us, and increasingly emergency physicians are being asked to make "appropriate" decisions regarding patient disposition. Hasley et al provide data to help us assess the risk of mortality from community-acquired pneumonia based upon radiographic criteria, thereby helping us to make decisions regarding resource utilization in these patients.

The objective of this multicenter, prospective cohort study of medical outcome in patients with community-acquired pneumonia was to determine which, if any, chest x-ray findings were independently associated with 30-day mortality. Inclusion criteria included: age 18 or older; presence of at least one clinical symptom suggestive of pneumonia; chest x-ray evidence of pneumonia; and acquisition of informed consent. Exclusion criteria were as follows: discharge from the hospital within the preceding 10 days; known chronic or unchanged pulmonary infiltrates; known HIV positivity; and previous enrollment in the study.

Of the original 2287 patients enrolled in the study, 1906 (83%) had confirmed radiographic infiltrates (designated as definite [64%] or probable/possible [36%]) and adequate data collection. Ninety-four (4.9%) of the 1906 patients died within 30 days of presentation, and follow-up data were obtained in all study patients.

Univariate analysis revealed the following radiographic characteristics to be associated with mortality: bilateral pleural effusions, moderate to massive pleural effusion, infiltrates involving two or three lobes, bilateral infiltrates, and bronchopneumonia. Only the presence of bilateral pleural effusions demonstrated an independent association with 30-day mortality by multivariate analysis (relative risk, 2.8; 95% CI, 1.4-5.8). The authors conclude that radiographic presence of bilateral pleural effusions carries an increased risk of 30-day mortality in patients diagnosed with community-acquired pneumonia.


So, are we wiser after reading this study? Perhaps to some degree, but certain critical points need to be raised. First and foremost, clinical criteria used in the diagnosis of pneumonia were not spelled out. That is to say, no clinical gold standard for the diagnosis was established a priori. It is not difficult to imagine that some study patients actually had pulmonary emboli. Second, HIV status was not known in all patients. Certainly the predictive criteria of x-ray findings would be expected to differ between immunocompromised patients and normal hosts. Third, interobserver agreement among radiologists was tested in a subset of patients; kappa values ranged from 0.46 to 0.51 for the presence of pleural effusion and number of involved lobes, respectively. These values represent only fair interobserver agreement, leaving one to wonder how many patients actually had pleural effusions or multilobar involvement. Finally, no clear or readily postulated explanation exists for the findings of this study. Why should the presence of bilateral pleural effusions be the only independent predictor of mortality? What about multilobar or bilateral infiltrates, findings which intuitively would be more worrisome? Either we still have a great deal to learn about the pathophysiology of pneumonia, or the observed findings are of questionable validity.

Does this study affect my clinical decision-making? Despite the study’s limitations, it does point to a semi-objective finding (presence of bilateral pleural effusions) that can be used in conjunction with other patient assessment parameters to make decisions of admit vs. discharge, ICU vs. floor. Other markers of serious illness (e.g., the presence of hypoxemia, hypothermia, or multiorgan involvement) are still important. And don’t forget perhaps the most important assessment parameter: our clinical gestalt and judgment developed over years of experience.