Decision-makers in low- and middle-income countries (LMICs) often must prioritize health spending without quantitative benchmarks for the value of their purchases. The Tufts Global Health Cost-Effectiveness Analysis (GH CEA) Registry (healtheconomicevaluation.org/GHCEARegistry/) is a freely-available, curated and standardized dataset designed to address this need.

All indexed English-language articles published between 1995 and 2017 are currently included in the GH CEA Registry. Studies are limited to those reporting cost-effectiveness in terms of cost per disability-adjusted life years (DALYs) averted, a commonly-employed metric in global health. Abstracted data include intervention type, comparator(s), country, funding source, study characteristics (e.g., perspective, time horizon), primary study findings, sensitivity analyses, and disaggregated data on costs and DALYs. Study quality is assessed using a numerical scoring system (from 1-7, higher scores indicating better quality) based on accuracy of findings and comprehensive reporting of methods and results.

To date, 620 articles have been included in the GH CEA Registry. Among LMICs, studies have been conducted primarily in Sub-Saharan Africa (41 percent) or South Asia (34 percent), have focused on communicable diseases (67 percent), and have involved immunization, educational, or pharmaceutical interventions (67 percent). As a priority-setting example, seven percent of interventions from higher-quality studies (ratings of 5 or higher) were reported to be cost-saving (i.e., lower costs and greater DALYs than standard care), two-thirds of which involved primary disease prevention (e.g., immunization, educational or behavioral interventions).

The GH CEA Registry is a new tool for decision-makers in LMICs, particularly those without a formal health technology assessment infrastructure but with a remit for providing access to essential, cost-effective health interventions. New functions are under development, including league tables for priority ranking, a repository for shared models, and tools for enhancing transferability between settings.