Bayesian Estimation of Pneumonia Etiology: Epidemiologic Considerations and Applications to the Pneumonia Etiology Research for Child Health Study

Abstract

In pneumonia, specimens are rarely obtained directly from the infection site, the lung, so the pathogen causing infection is determined indirectly from multiple tests on peripheral clinical specimens, which may have imperfect and uncertain sensitivity and specificity, so inference about the cause is complex. Analytic approaches have included expert review of case-only results, case–control logistic regression, latent class analysis, and attributable fraction, but each has serious limitations and none naturally integrate multiple test results. The Pneumonia Etiology Research for Child Health (PERCH) study required an analytic solution appropriate for a case–control design that could incorporate evidence from multiple specimens from cases and controls and that accounted for measurement error. We describe a Bayesian integrated approach we developed that combined and extended elements of attributable fraction and latent class analyses to meet some of these challenges and illustrate the advantage it confers regarding the challenges identified for other methods.

Surveillance for incidence and etiology of early-onset neonatal sepsis in Soweto, South Africa

“…The primary analytic objective was to estimate the proportion of sepsis episodes attributed to each pathogen evaluated in the study, referred to throughout as “pathogen proportions.” Because the study captured two specimen samples (blood and respiratory), conducted up to three tests per pathogen (blood culture, blood TAC, respiratory TAC), and because each of these tests had its own true positive rate (TPR or sensitivity) and false positive rate (FPR or 1-specificity) for detection of the true cause of the sepsis episode, we adapted the partial latent class model (pLCM) methods developed by Wu et al[19, 20] for application to our surveillance as detailed in the text in S1 Text. All protocol defined EOS cases with at least one laboratory test result available were included in the model input dataset, along with blood culture final determination (for neonates with sepsis) and observed binary (positive or negative) results of the TAC tests.…”

The epidemiology and estimated etiology of pathogens detected from the upper respiratory tract of adults with severe acute respiratory infections in multiple countries, 2014–2015

“…An additional assumption of the model is that infection with one pathogen does not change the likelihood of carriage of another pathogen. Some organisms, such as S. pneumoniae, which are both frequent colonizers and common etiologies of pneumonia, have demonstrated that convergence and accuracy of the model are improved by obtaining data from more than one type of diagnostic test (e.g., culture and nonculture methods) on multiple specimen types [13,33,[47][48][49].…”

Colonization Density of the Upper Respiratory Tract as a Predictor of Pneumonia—Haemophilus influenzae, Moraxella catarrhalis, Staphylococcus aureus, and Pneumocystis jirovecii

“…We set out to determine if density of URT colonizers predicted pathogen-specific infections among pneumonia cases in the Pneumonia Etiology Research for Child Health (PERCH) study. Provided differences in densities between cases and controls, we evaluated whether pathogen densities offer any value in pneumonia diagnostic algorithms, or provide information beyond presence or absence of positivity alone in analytic models such as the PERCH integrated analysis [20]. The density evaluation for S. pneumoniae is reported elsewhere [21].…”