Objective: To determine if providing general practitioners (GPs) with costing information can change their clinical behaviour and reduce medical costs. Data sources: MEDLINE, CINAHL, Health Plan and EMBASE and citations in review articles were searched for studies published between 1980 and 1996. Study selection: Studies were included if they provided costing information to GPs with the aim of decreasing costs by changing behaviour, included an objective measure of GP performance or clinical care, and used a randomised or quasi-randomised controlled design, crossover design or a controlled time series. Data extraction: Data extracted included study design, intervention used and measure of GP performance/clinical care (including test ordering, drug prescribing, hospital and primary care visits and costs). Data synthesis: Six studies met the inclusion criteria. Computerised feedback on drug costs increased generic prescribing, and ''academic detailing'' decreased inappropriate prescribing of target drugs. Providing costing information also decreased ordering of diagnostic tests. ''Gatekeeper'' physicians reduced use of hospital and specialist services. Only two studies found the changes were sustained for nine months or longer and only one evaluated health outcomes. Conclusion: The provision of costing information can change GP behaviour in all service areas. Sustainability of these changes and linking of cost savings to health outcomes have not been well studied.
