Impact of monovalent rotavirus vaccine on diarrhoea-associated post-neonatal infant mortality in rural communities in Malawi: a population-based birth cohort study

Abstract

Background: Rotavirus is a major contributor to child mortality. The effect of rotavirus vaccine on diarrhoea mortality has been estimated in middle-income but not low-income settings, where mortality is high and vaccine effectiveness in reducing admissions to hospital is lower. Empirical population-based mortality studies have not been done in any setting. Malawi introduced monovalent rotavirus vaccine (RV1) in October, 2012. We aimed to investigate the impact and effectiveness of the RV1 vaccine in reducing diarrhoea-associated mortality in infants aged 10-51 weeks.

Methods: In this population-based cohort study, we included infants born between Jan 1, 2012, and June 1, 2015, in Mchinji, Central Malawi and analysed data on those surviving 10 weeks. Individual vaccination status was extracted from caregiver-held records or report at home visits at 4 months and 1 year of age. Survival to 1 year was confirmed at home visit, or cause of death ascertained by verbal autopsy. We assessed impact (1 minus mortality rate ratio following vs before vaccine introduction) using Poisson regression. Among vaccine-eligible infants (born from Sept 17, 2012), we assessed effectiveness (1 minus hazard ratio) using Cox regression.

Findings: Between Jan 1, 2012, and June 1, 2015, we recruited 48 672 livebirths in Mchinji, among whom 38 518 were vaccine-eligible and 37 570 survived to age 10 weeks. Two-dose versus zero-dose effectiveness analysis included 28 141 infants, of whom 101 had diarrhoea-associated death before 1 year of age. Diarrhoea-associated mortality declined by 31% (95% CI 1-52; p=0·04) after RV1 introduction. Effectiveness against diarrhoea-mortality was 34% (95% CI -28 to 66; p=0·22).

Interpretation: RV1 was associated with substantial reduction in diarrhoea-associated deaths among infants in this rural sub-Saharan African setting. These data add considerable weight to evidence showing the impact of rotavirus vaccine programmes.

Funding: Wellcome Trust and GlaxoSmithKline Biologicals.

Figure 1

STROBE study profile of the participating vaccine-eligible cohort, Site 1 *Completion of follow-up means sufficient information was obtained by 1 year of age to determine whether the participant could be included in analysis or excluded for the reasons outlined in the figure.

Figure 2

12-month weighted moving average smoothed trend* for all-cause and diarrhoea-associated mortality and dose 3 pneumococcal and dose 2 rotavirus vaccine coverage in 10–51-week old infants (A) Site 1; January, 2012, to June, 2015. (B) Site 2; January, 2004, to June, 2016. RV1=monovalent rotavirus vaccine. PCV=pneumococcal conjugate vaccine. WASH=water and sanitation. *See appendix.

Figure 3

Survival analysis of diarrhoea-associated death in the vaccine-eligible cohort, site 1 (A) Kaplan-Meier survival curve and 95% CIs, by vaccine receipt. (B) Fully parametric hazard rate over survival time, by vaccine receipt. (C) Vaccine effectiveness over survival time. (D) Hazard rate difference and 95% CIs (between vaccinated and unvaccinated infants) over survival time. RV=rotavirus vaccine. *Number at risk is the total number of surviving infants and infants who died with diarrhoea. 15 zero-dose and 209 two-dose recipients contributed to survival time until censoring for reasons other than death.