Statistical methods for cost-effectiveness analyses that use data from cluster randomized trials: a systematic review and checklist for critical appraisal.

Manuel Gomes; Richard Grieve ORCID logo; Richard Nixon; WJ Edmunds ORCID logo; (2012) Statistical methods for cost-effectiveness analyses that use data from cluster randomized trials: a systematic review and checklist for critical appraisal. Medical decision making, 32 (1). pp. 209-220. ISSN 0272-989X DOI: 10.1177/0272989X11407341
Copy

INTRODUCTION: The best data for cost-effectiveness analyses (CEAs) of group-level interventions often come from cluster randomized trials (CRTs), where randomization is by cluster (e.g., the hospital attended), not by individual. METHODS: for these CEAs need to recognize both the correlation between costs and outcomes and that these data may be dependent on the cluster. General checklists and methodological guidance for critically appraising CEA ignore these issues. This article develops a new checklist and applies it in a systematic review of CEAs that use CRTs. METHODS: The authors developed a checklist for CEAs that use CRTs, informed by a conceptual review of statistical methods. This checklist included criteria such as whether the analysis allowed for both clustering and the correlation between individuals' costs and outcomes. The authors undertook a systematic literature review of full economic evaluations that used CRTs. The quality of studies was assessed with the new checklist and by the "Drummond checklist." RESULTS: The authors identified 62 papers that met the inclusion criteria. On average, studies satisfied 9 of the 10 criteria for the checklist but only 20% of criteria for the new checklist. More than 40% of studies adopted statistical methods that completely ignored clustering, and 75% disregarded any correlation between costs and outcomes. Only 4 studies employed appropriate statistical methods that allowed for both clustering and correlation. CONCLUSIONS: Most economic evaluations that use data from CRTs ignored clustering or correlation. Statistical methods that address these issues are available, and their use should be encouraged. The new checklist can supplement generic CEA guidelines and highlight where research practice can be improved.

Full text not available from this repository.

Atom BibTeX OpenURL ContextObject in Span Multiline CSV OpenURL ContextObject Dublin Core Dublin Core MPEG-21 DIDL EndNote HTML Citation JSON MARC (ASCII) MARC (ISO 2709) METS MODS RDF+N3 RDF+N-Triples RDF+XML RIOXX2 XML Reference Manager Refer Simple Metadata ASCII Citation EP3 XML
Export

Downloads