%0 Journal Article %A Talata Sawadogo-Lewis %A Youssouf Keita %A Emily Wilson %A Souleymane Sawadogo %A Ibrahim Téréra %A Hamadoun Sangho %A Melinda Munos %T Can We Use Routine Data for Strategic Decision Making? A Time Trend Comparison Between Survey and Routine Data in Mali %D 2021 %R 10.9745/GHSP-D-21-00281 %J Global Health: Science and Practice %P 869-880 %V 9 %N 4 %X Key FindingsThe direction and magnitude of the time trends of contraceptive prevalence rate, institutional delivery, and 3 doses of the diphtheria, pertussis, and tetanus (DPT3) vaccine from 2001 to 2012 were similar at the national level between data sources.At the regional level, routine data tended to overestimate DPT3 coverage, and underestimate institutional delivery and contraceptive prevalence relative to survey data.Routine data are appropriate for use for program planning. However, it does not seem to be appropriate for use for impact evaluations, particularly at the regional level.Key ImplicationsProgram planners and evaluators can continue to use routine data to detect overall trends but until the quality of routine data is improved, they should refrain from using it for impact evaluations.Background:Countries with scarce resources need timely and high-quality data on coverage of health interventions to make strategic decisions about where to allocate investments in health. Household survey data are generally regarded as “gold standard,” high-quality data. This study assessed the comparability of intervention coverage time trends from routine and survey data at national and subnational levels in Mali.Methods:We compared 3 coverage indicators: contraceptive prevalence rate, institutional delivery, and 3 doses of diphtheria, pertussis, and tetanus (DPT3) vaccine, using 3 Mali Demographic and Health Surveys (DHS 2001, 2006, and 2012–2013) and routine health system data covering 2001–2012. For routine data, we used local health information system (HIS) annual reports and an HIS database. To compare time trends between the data sources, we calculated the percentage point change and 95% confidence interval from 2001–2006 and 2006–2012. We then computed the absolute and relative differences between the 2 data sources for each indicator over time at national and regional levels and assessed their level of significance.Results:The direction and magnitude of the time trends of contraceptive prevalence rate, institutional delivery, and DPT3 vaccine from 2001 to 2012 were similar at the national level between data sources. At the regional level, there were significant differences in the magnitude and direction of time trends for institutional delivery and the DPT3 vaccine; contraceptive prevalence trends were more consistent. Routine data tended to overestimate DPT3 coverage, and underestimate institutional delivery and contraceptive prevalence relative to survey data.Conclusion:Routine data in Mali—particularly at the national level—appear to be appropriate for use to inform program planning and prioritization, but routine time trends should be interpreted with caution at the subnational level. For program evaluations, routine data may not be appropriate to draw accurate inferences about program impact. %U https://www.ghspjournal.org/content/ghsp/9/4/869.full.pdf