Predictors of in-hospital mortality and length of stay in community-acquired pneumonia: A 5-year multi-centre case control study of adults in a developing country

Background: We investigated predictors of in-hospital mortality and length of hospital stay among adults with community-acquired pneumonia (CAP) in Nigeria in order to provide recommendations to improve CAP outcomes in developing countries.Methods: This was a multi-centre case control study of patients ≥ 18 years who were admitted with CAP between 2008 and 2012. Case notes of 100 consecutive patients who died (cases) and random sample of 300 patients discharged (controls) were selected.Results: Mean ages were 55.4±19.6 (cases) and 49.3±19.2 (controls). Independent predictors of mortality were CURB-65 score ≥ 3: adjusted odds ratio (aOR) 24.3, late presentation: aOR 8.6, co-morbidity: aOR 3.9, delayed first dose antibiotics (>4 hours): aOR 3.5, need for supplemental oxygen: aOR 4.9, multilobar pneumonia: aOR 4.0, non-pneumococcal aetiology: aOR 6.5, anaemia: aOR 3.8 and hyperglycemia: aOR 8.6. CURB-65 ≥ 3 predicted mortality with a high specificity (96.1%) but low sensitivity (75%); positive predictive value of 88.2% and negative predictive value of 90.8%. Care in hospital A and B: aOR 3.3 and 2.2 respectively, male gender aOR 2.1, co-morbidity aOR 3.0, anaemia aOR 2.1 and elevated serum creatinine aOR 6.3 independently predicted length of hospital stay >10 days among survivors.Conclusions: Several modifiable patient-related and process-of-care factors predicted in-hospital mortality, and length of hospital stay among survivors. Our findings should be used to improve CAP outcomes in developing countries.