OBJECTIVES: the scarcity of cost data is a barrier to health economic evaluations in Middle East and North Africa (MENA). We evaluated the accuracy of estimates when transferring medical costs within MENA via adjustment by GDP per capita (GDPpc).

METHODS: we performed a systematic review of medical costs reported for local populations of Algeria, Bahrain, Egypt, Iraq, Jordan, Saudi Arabia, Kuwait, Lebanon, Libya, Morocco, Oman, Palestine, Qatar, Syria, Tunisia, United Arab Emirates and Yemen (PubMed, 1995-2019). From multi-country studies that employed homogenous costing methods, we estimated each cost item from the relative differences in costs and GDPpc via robust regression. We analysed the relative error (RE) of predictions via linear regression (log-RE, M1) and logistic regression (RE<30%, M2) using cost complexity (drug, unit, disease), country (A) or GDP per capita (B), relative difference of GDPpc between countries, number of available foreign costs, and cost sources (model, primary study, secondary sources) as predictors.

RESULTS: From 3525 cost items (1646 records, 206 full text papers) we identified 142 eligible direct medical costs from 7 multi-country studies. Most accurate estimates were provided by adjusting mean available foreign costs by 0.284 times relative differences in GDPpc. 80% of true costs (10-90 percentile range) fell in the predicted cost -64% / +71% error range. Predictions were most accurate for drug costs (Models 1A, 1B, 2A, 2B), when GDPpc differences were big between countries (Models 1B, 2B) and when more foreign costs were available (Models 2A, 2B); and least accurate for Qatar and Syria (Models 1A, 2A) and for countries with higher GDP per capita (Models 1B, 2B) (p<0.05).

CONCLUSIONS: between MENA countries, medical costs vary proportionally to, but less than GDPpc. When transferring costs, matching costs should be collected from multiple countries, adjusted by GDPpc and used with wide (>75%) error margin in sensitivity analysis.