Meta-analytic methods for health services research: An example from geriatrics

The authors recently published a meta-analysis of controlled trials of comprehensive geriatric assessment (CGA). The results supported the view that efficacy of CGA is strongly related to the patients, objectives, and basic design of CGA programs, and that particular program models and design features are associated with important health outcome improvements (e.g., survival, living at home, and functional improvement atfollow-up). Present objectives include the outline of methods and how they were developed given the condition of the trial database and scientific context. Aspects of the approach, such as (a) survey of primary trialists to recover unpublished information and standardize data, (b) development of a program typology to guide the principal analysis, and (c) incorporation of program design features as covariates where statistical heterogeneity was detected4 proved extremely useful, and have implications for other systematic reviews ofsimilarly complexprnary trials ofnew health care technologies, healthservices, and organizational interventions.