Managed care: US research evidence and its lessons for the NHS.

OBJECTIVES

To review the high quality US evidence on performance of managed health care organisations and the available US evidence on specific managed care techniques; namely, financial incentives, utilisation management and review, physician profiling and disease management.

METHODS

Literature searches were conducted using numerous databases including Medline, Embase, the Social Sciences Citation Index and the National Health Service (NHS) Centre for Reviews and Dissemination library. For inclusion of evaluations of overall performance, studies had to use a comparison group (typically fee-for-service patients), make appropriate statistical adjustments for differences between groups, and be published in a peer-reviewed journal from 1980 forward. For assessments of techniques, less-demanding inclusion criteria reflected the paucity of generalisable literature; however, more current results were required (1990 forward).

RESULTS

We identified 70 articles for systematic review, covering 18 dimensions of performance (e.g. utilisation, quality of care, consumer satisfaction, equity). The strength of the evidence varied by dimension. It was strongest for utilisation and quality. In general, managed care seems to reduce hospitalisation and use of high-cost discretionary services, to increase preventive screening, and to be neutral in terms of patient outcomes. As for specific techniques, we identified 19 articles for review, but limitations of these studies prevented our drawing any definite conclusions about techniques' effectiveness. This is an important, if somewhat negative, conclusion.

CONCLUSIONS

Applying US evidence is complicated by an irrelevant comparator and a higher baseline of utilisation. Managed care brought Americans the familiar NHS practices of population-based health care and resource management through gatekeeping; hence, changes due to UK adoption of managed care techniques may be modest. US evidence should be used to generate hypotheses, not to predict UK behaviour.