Abstract
BACKGROUND: Neuraminidase inhibitors were widely used during the 2009-10 influenza A H1N1 pandemic, but evidence for their effectiveness in reducing mortality is uncertain. We did a meta-analysis of individual participant data to investigate the association between use of neuraminidase inhibitors and mortality in patients admitted to hospital with pandemic influenza A H1N1pdm09 virus infection. METHODS: We assembled data for patients (all ages) admitted to hospital worldwide with laboratory confirmed or clinically diagnosed pandemic influenza A H1N1pdm09 virus infection. We identified potential data contributors from an earlier systematic review of reported studies addressing the same research question. In our systematic review, eligible studies were done between March 1, 2009 (Mexico), or April 1, 2009 (rest of the world), until the WHO declaration of the end of the pandemic (Aug 10, 2010); however, we continued to receive data up to March 14, 2011, from ongoing studies. We did a meta-analysis of individual participant data to assess the association between neuraminidase inhibitor treatment and mortality (primary outcome), adjusting for both treatment propensity and potential confounders, using generalised linear mixed modelling. We assessed the association with time to treatment using time-dependent Cox regression shared frailty modelling. FINDINGS: We included data for 29,234 patients from 78 studies of patients admitted to hospital between Jan 2, 2009, and March 14, 2011. Compared with no treatment, neuraminidase inhibitor treatment (irrespective of timing) was associated with a reduction in mortality risk (adjusted odds ratio [OR] 0·81; 95% CI 0·70-0·93; p=0·0024). Compared with later treatment, early treatment (within 2 days of symptom onset) was associated with a reduction in mortality risk (adjusted OR 0·48; 95% CI 0·41-0·56; p<0·0001). Early treatment versus no treatment was also associated with a reduction in mortality (adjusted OR 0·50; 95% CI 0·37-0·67; p<0·0001). These associations with reduced mortality risk were less pronounced and not significant in children. There was an increase in the mortality hazard rate with each day's delay in initiation of treatment up to day 5 as compared with treatment initiated within 2 days of symptom onset (adjusted hazard ratio [HR 1·23] [95% CI 1·18-1·28]; p<0·0001 for the increasing HR with each day's delay). INTERPRETATION: We advocate early instigation of neuraminidase inhibitor treatment in adults admitted to hospital with suspected or proven influenza infection. FUNDING: F Hoffmann-La Roche.