Generalising the results of economic evaluations from high-income to middle-income countries

Article type
Authors
Iglesias C, Augustovski F, Garcia S, Manca A, Rubinstein A, Drummond M
Abstract
Background: Methods which enable policy makers to assess the extent to which the economic and clinical results from RCTs are valid in different geographical settings should have a key role in Health Technology Assessment (HTA). More so in the healthcare systems from middle-income and developing countries, where research and development funding for HTA are scarcer.

Objectives: To evaluate the generalisability of the results from published economic evaluation (EE) studies developed and/or partly conducted in North-America and Europe to Latin-American (LA) countries.

Methods: A systematic review of studies published between Jan-1980 and Jan-2004 was conducted. Electronic searches were carried out in NHS-EED, OHE-HEED, LILACS and NEVALAT databases. Trial and model-based partial (cost-consequences) and full-EE studies (cost-effectiveness, cost-utility and cost-benefit) which involved at least one LA country were included. Three independent reviewers extracted relevant data from a predefined checklist validated by regional and international experts.

Results: Of 521 studies, 73 were EEs (38% RCTs, 19% non-randomized-trials, and 43% models). Trial-based analyses were, 41% single-center national, 20% multicenter national, and 26% multinational. Reporting quality was heterogeneous. In 88% of trial-based EEs the health benefit estimate was based on intention-to-treat population. Most multinational trials reported aggregated clinical, resource use, and costs data, which made the assessment of generalisability more difficult. Patient-level resource use was analised in 67% of the cases and was collected only in a sub-sample of countries. Geographical variation by country or region was explored only in one model-based-EE. Cost-utility-analyses used preference weights from other regions, i.e. United Kingdom, USA.

Conclusions: A large proportion of EE studies involving LA countries are trial-based analyses. Nonetheless, important methodological flaws (ie. lack of randomization, selection-bias, poor characterization of patients, arbitrary use of data produced in other jurisdictions, among others) jeopardize their internal and external validity. EE analyses in LA do not closely adhere to existing international guidelines. 'State of the art' methods are mainly undertaken within international collaborations. Additional training is needed for health services researchers operating in LA countries to ensure future research output is sound and generalisable. EE studies conducted in high-income countries should explore and discuss the generalisability of their results to other settings.