TY - JOUR T1 - Modeling Outputs Can Be Valuable When Uncertainty Is Appropriately Acknowledged, but Misleading When Not JF - Global Health: Science and Practice JO - GLOB HEALTH SCI PRACT SP - 530 LP - 533 DO - 10.9745/GHSP-D-17-00444 VL - 5 IS - 4 AU - Steve Hodgins Y1 - 2017/12/28 UR - http://www.ghspjournal.org/content/5/4/530.abstract N2 - While modeling approaches seek to draw on the best available evidence to project health impact of improved coverage of specific interventions, uncertainty around the outputs often remains. When the modeling estimates are used for advocacy, these uncertainties should be communicated to policy makers clearly and openly to ensure they understand the model's limits and to maintain their confidence in the process.See related articles by Askew et al., Rodriguez et al., and Jones-Hepler et al.In efforts to further the use of evidence for policy and planning decision making, there has been considerable use in global health—across a variety of technical areas—of quantitative modeling approaches that attempt to project health impact of improved coverage of specific “evidence-based” interventions. This approach has roots in analyses done for the World Bank's World Development Report 1993: Investing in Health,1 which emphasized provision of a “minimum essential package of services” and modeled expected population-level impacts of improved coverage of “evidence-based” interventions in terms of disability-adjusted life years.This issue of GHSP includes 3 papers on the use of such models, one in the family planning field (Askew2), another at the confluence of family planning and HIV/AIDS (Rodriguez3), and the third in maternal and newborn health (Jones-Hepler4).As discussed by Askew et al.,2 when multiple modeling approaches or packages are used to address the same question for the same setting and end up with disparate estimates, policy makers' confidence in the methodology can diminish. Because models used in the same field may be developed with different purposes in mind, there may be entirely valid reasons for them to yield differing estimates. To best serve the policy and program communities, however, ideally there should be some degree of harmonization across models. Askew et al. document one such … ER -