Modelling the effect of a mass radio campaign on child mortality using facility utilisation data and the Lives Saved Tool (LiST): findings from a cluster randomised trial in Burkina Faso.

Background: A cluster randomised trial (CRT) in Burkina Faso was the first to demonstrate that a radio campaign increased health-seeking behaviours, specifically antenatal care attendance, health facility deliveries and primary care consultations for children under 5 years. Methods: Under-five consultation data by diagnosis was obtained from primary health facilities in trial clusters, from January 2011 to December 2014. Interrupted time-series analyses were conducted to assess the intervention effect by time period on under-five consultations for separate diagnosis categories that were targeted by the media campaign. The Lives Saved Tool was used to estimate the number of under-five lives saved and the per cent reduction in child mortality that might have resulted from increased health service utilisation. Scenarios were generated to estimate the effect of the intervention in the CRT study areas, as well as a national scale-up in Burkina Faso and future scale-up scenarios for national media campaigns in five African countries from 2018 to 2020. Results: Consultations for malaria symptoms increased by 56% in the first year (95% CI 30% to 88%; p<0.001) of the campaign, 37% in the second year (95% CI 12% to 69%; p=0.003) and 35% in the third year (95% CI 9% to 67%; p=0.006) relative to the increase in the control arm. Consultations for lower respiratory infections increased by 39% in the first year of the campaign (95% CI 22% to 58%; p<0.001), 25% in the second (95% CI 5% to 49%; p=0.010) and 11% in the third year (95% CI -20% to 54%; p=0.525). Diarrhoea consultations increased by 73% in the first year (95% CI 42% to 110%; p<0.001), 60% in the second (95% CI 12% to 129%; p=0.010) and 107% in the third year (95% CI 43% to 200%; p<0.001). Consultations for other diagnoses that were not targeted by the radio campaign did not differ between intervention and control arms. The estimated reduction in under-five mortality attributable to the radio intervention was 9.7% in the first year (uncertainty range: 5.1%-15.1%), 5.7% in the second year and 5.5% in the third year. The estimated number of under-five lives saved in the intervention zones during the trial was 2967 (range: 1110-5741). If scaled up nationally, the estimated reduction in under-five mortality would have been similar (9.2% in year 1, 5.6% in year 2 and 5.5% in year 3), equating to 14 888 under-five lives saved (range: 4832-30 432). The estimated number of lives that could be saved by implementing national media campaigns in other low-income settings ranged from 7205 in Burundi to 21 443 in Mozambique. Conclusion: Evidence from a CRT shows that a child health radio campaign increased under-five consultations at primary health centres for malaria, pneumonia and diarrhoea (the leading causes of postneonatal child mortality in Burkina Faso) and resulted in an estimated 7.1% average reduction in under-five mortality per year. These findings suggest important reductions in under-five mortality can be achieved by mass media alone, particularly when conducted at national scale.