Estimating the Burden of Pneumococcal Pneumonia among Adults: A Systematic Review and Meta-Analysis of Diagnostic Techniques

BACKGROUND Pneumococcal pneumonia causes significant morbidity and mortality among adults. Given limitations of diagnostic tests for non-bacteremic pneumococcal pneumonia, most studies report the incidence of bacteremic or invasive pneumococcal disease (IPD), and thus, grossly underestimate the pneumococcal pneumonia burden. We aimed to develop a conceptual and quantitative strategy to estimate the non-bacteremic disease burden among adults with community-acquired pneumonia (CAP) using systematic study methods and the availability of a urine antigen assay. METHODS AND FINDINGS We performed a systematic literature review of studies providing information on the relative yield of various diagnostic assays (BinaxNOW® S. pneumoniae urine antigen test (UAT) with blood and/or sputum culture) in diagnosing pneumococcal pneumonia. We estimated the proportion of pneumococcal pneumonia that is bacteremic, the proportion of CAP attributable to pneumococcus, and the additional contribution of the Binax UAT beyond conventional diagnostic techniques, using random effects meta-analytic methods and bootstrapping. We included 35 studies in the analysis, predominantly from developed countries. The estimated proportion of pneumococcal pneumonia that is bacteremic was 24.8% (95% CI: 21.3%, 28.9%). The estimated proportion of CAP attributable to pneumococcus was 27.3% (95% CI: 23.9%, 31.1%). The Binax UAT diagnosed an additional 11.4% (95% CI: 9.6, 13.6%) of CAP beyond conventional techniques. We were limited by the fact that not all patients underwent all diagnostic tests and by the sensitivity and specificity of the diagnostic tests themselves. We address these resulting biases and provide a range of plausible values in order to estimate the burden of pneumococcal pneumonia among adults. CONCLUSIONS Estimating the adult burden of pneumococcal disease from bacteremic pneumococcal pneumonia data alone significantly underestimates the true burden of disease in adults. For every case of bacteremic pneumococcal pneumonia, we estimate that there are at least 3 additional cases of non-bacteremic pneumococcal pneumonia.