Effectiveness of smartphone-based Community Case Management on urgent referral, re-consultation and hospitalization of children under-5 in Malawi: Results of a cluster-randomized, stepped-wedge trial

Background Integrated community case management (CCM) has led to reductions in child mortality in Malawi from illnesses such as malaria, pneumonia and diarrhoea. However, adherence to CCM guidelines is often poor, potentially leading to inappropriate clinical decisions and outcomes. We determined the impact of an electronic version of CCM (eCCM) application on referral, re-consultation and hospitalization rates of children presenting to village clinics in Malawi. Methods A stepped-wedge cluster-randomized trial compared paper-based CCM (control) with and without use of an eCCM app on smartphones from November 2016 to April 2017.A total of 102 village clinics from two districts in Northern Malawi were assigned to one of six clusters which were randomized to the sequencing of crossover from the control to the intervention phases, as well as the duration of exposure in each phase. Children ≥2 months to <5 years presenting with acute illness were enrolled consecutively by Health Surveillance Assistants (HSAs). The primary outcome of urgent referrals to higher-level facilities was evaluated using multi-level mixed effects models. A logistic regression model with random effect of cluster and fixed effect for each step was fitted. Adjustment for potential confounders included baseline factors, such as patient’s age, sex, and geographical location of village clinics. Calendar time was adjusted for in the analysis. Results A total of 6965 children were recruited, 3421 in the control and 3544 in the intervention phase. After adjusting for calendar time, children in the intervention phase were more likely to be urgently referred to a higher-level health facility compared with children in the control phase (OR 2.02, 95% CI 1.27-3.23; p<0.01). Overall, children in the intervention arm had lower odds of attending a repeat HSA consultation (OR 0.45, 95% CI 0.34-0.59; p<0.01) or hospital admission (OR 0.75, 95% CI 0.62-0.90; p0.05). Conclusions Addition of eCCM decision support led to a greater proportion of children being referred to higher-level facilities, with no apparent increase in hospital admissions or repeat consultations in village clinics. Our findings provide support for the implementation of eCCM tools in Malawi and other Low and Middle Income Countries (LMIC), with a need for ongoing assessment of effectiveness and integration with national digital health strategies. Trial registration http://ClinicalTrials.gov ; NCT02763345 . Registered 3 May 2016