Mortality among patients due to adverse drug reactions that occur following hospitalisation: a meta-analysis.

Abstract

To estimate the prevalence of mortality among patients that develop adverse drug reactions during hospitalisation (ADRIn), to examine heterogeneity through subgroup analysis and to identify system-organ class (SOC) and their causative drugs. Two investigators searched PubMed, Google Scholar and related bibliography for studies reporting ADRIn-related mortality data. The primary outcome was to compute overall prevalence of fatal ADRIn (95% CI) using double arcsine method. We explored the heterogeneity (I2) in its estimation based on study design, study population and data collection methods. The secondary outcomes were the pattern of fatal reactions and their causative drugs. PROSPERO register number-CRD42018090331. Out of 349 full text assessed, 48 studies satisfying the selection criteria were included. The fatal ADRIn prevalence was 0.11% (95% CI 0.06-0.18%; I2= 93%). The fatal ADRIn prevalence ranged from 0.03% (I2= 0%) in all ages to 0.27% (I2= 90%) in elderly population studies. Elderly studies varied for all study characteristics. Among study wards, a higher trend of prevalence was observed in 'internal medicine and ICU' (0.46%, I2= 51%) and 'neonatal/paediatric ward and ICU' (0.34%, I2= 58%) studies. The commonly involved SOC were 'gastrointestinal disorders' (28.79%), 'blood and lymphatic system disorders' (19.69%) and 'renal and urinary disorders' (13.64%). Most commonly observed causative drug-fatal ADRIn pairs were antithrombotics and nonsteroidal anti-inflammatory drugs induced gastrointestinal bleeding, and antineoplastic agents induced cytopenia. ADRIn is an important cause of mortality. Age groups and study wards have important influence on prevalence of fatal ADRIn and its heterogeneity across studies. Few class drugs contribute to sizable proportion of ADRIn-related mortality.