Multi-country analysis of routine data from integrated community case management (iCCM) programs in sub-Saharan Africa

Aim To identify better performing iCCM programs in sub-Saharan Africa (SSA) and identify factors associated with better performance using routine data. Methods We examined 15 evaluations or studies of integrated community case management (iCCM) programs in SSA conducted between 2008 and 2013 and with information about the program; routine data on treatments, supervision, and stockouts; and, where available, data from community health worker (CHW) surveys on supervision and stockouts. Analyses included descriptive statistics, Fisher's exact test for differences in median treatment rates, the Kruskal-Wallis test for differences in the distribution of treatment rates, and Spearman's correlation by program factors. Results The median percent of annual expected cases treated was 27% (1-74%) for total iCCM, 37% (1-80%) for malaria, 155% (7-552%) for pneumonia, and 27% (1-74%) for diarrhoea. Seven programs had above median total iCCM treatments rates. Four programs had above median treatment rates, above median treatments per active CHW per month, and above median percent of expected cases treated. Larger populations under-five targeted were negatively associated with treatment rates for fever, malaria, diarrhea, and total iCCM. The ratio of CHWs per population was positively associated with diarrhoea treatment rates. Use of rapid diagnostic tests (RDTs) was negatively associated with treatment rates for pneumonia. Treatment rates and percent of annual expected cases treated were equivalent between programs with volunteer CHWs and programs with salaried CHWs. Conclusions There is large variation in iCCM program performance in SSA. Four programs appear to be higher performing in terms of treatment rates, treatments per CHW per month, and percent of expected cases treated. Treatment rates for diarrhoea are lower than expected across most programmes. CHWs in many programmes are overtreating pneumonia. Programs targeting larger populations under-five tend to have lower treatment rates. The reasons for lower pneumonia treatment rates where CHWs use RDTs need to be explored. Programs with volunteer CHWs and those with salaried CHWs can achieve similar treatment rates and percent of annual expected cases treated but to do so volunteer programs must manage more CHWs per population and salaried CHWs must provide more treatments per CHW per month.