Beta
Views And Reviews

Free for All: Lessons from the RAND Health Insurance Experiment

BMJ 1994; 308 doi: http://dx.doi.org/10.1136/bmj.308.6945.1724a (Published 25 June 1994) Cite this as: BMJ 1994;308:1724
  1. C Normand

    The results of the RAND Health Insurance Experiment are important in the debates on access to and financing of health services, and direct charges to patients. The study randomly assigned people in six areas in the United States to different health insurance plans for periods of three or five years. Elderly people were excluded (as most already had access to free care through Medicare). Around 2000 families were included in the experiment.

    The aim of the study was to test in experimental conditions the response of people to different charges for using health services. Some people were enrolled in plans that provided care free at the point of use. Others had to pay varying proportions of the cost. For those paying some part of the cost there was usually a maximum annual payment. The study measured use of health services, health status, and the satisfaction of patients. New measurement tools were developed for the study.

    The main results of the experiment are now well known, for the study has produced over 300 reports and publications. Free for All brings together all the findings, and provides detail of study design, data collection, analysis, and commentary on the results.

    The study found that use of health services varies with the level of charges paid by patients directly. Those who faced large copayments used services less. The reduction applies to most types of service, and not just to those of relatively little value. The exceptions are children's hospital services and some mental health services, which are not affected by the price at the point of access. In essence health services behave like other economic goods - their use is increased by low prices and decreased by higher ones.

    The important issue is what difference this change in use of health services makes to health. The general conclusion is none. This is a little surprising given that the reduction in use includes services which can be shown to be highly effective. Part of the explanation may be the reduction in iatrogenic disease, such as side effects from (inappropriate) prescription of antibiotics. There is also a proportion of effective treatments which are used in ultimately self limiting conditions. Whatever the explanation it is clear that the effects of reduced use of health services on health are at most small.

    An important exception to this general finding is that the reduced use by poorer people did have a measurable and harmful effect on health. Although this finding might be used to argue for exemption from charges for poorer people, the authors suggest the alternative of providing a few highly effective free services for poorer people.

    What is the relevance of the findings of the RAND experiment for countries such as Britain where most health services are free at the point of use? The findings would probably apply in most European countries - there would be some reduction in the use of services if charges were introduced, but the effect on health would be small. The evidence also shows that the effects of charges are indiscriminate - they do not eliminate the relatively ineffective treatments and retain the effective ones. As such they are crude rationing devices. A further consideration is the administrative cost of applying charges. Savings from reduced use of services will be offset in part by additional costs of administration.

    The experiment gives some insight into the likely effects of introducing charges, but does not provide a conclusive argument for doing so. It is possible to criticise many details of the conduct of the study, but it remains the most important source of evidence about the behaviour of individuals in response to health service charges.