There are a number of reasons for the increasing interest in outcomes of health care interventions over the last few years. Foremost is the realization that in all health care systems, resources are limited, and should be directed towards those interventions that are of proven effectiveness, and producing outcomes that are valued by patients. As will be discussed below, deciding upon the outcome to be achieved is a necessary prerequisite for determining whether any treatment is effective.

The next stimulus to outcomes research is the realization that there are large variations in practice, not only throughout the world, but within the same health care system, and indeed in neighbouring cities. For example, Wennberg's comparison between Boston and New Haven, two university cities on the East coast of the United States, showed that in Boston people had twice the chance of having a carotid endarterectomy compared to New Haven, but only half the chance of having coronary bypass surgery (Wennberg et al. 1987). These differences were apparent even when corrections were made for the age and sex distribution and other variables between the two local communities. Another striking example is that rates of hysterectomy correlate more closely with the number of gynaecologists per head of population than they do with the number of women within a population. Such variations indicate that much of what doctors do is a matter of practice style, is imprecise, and is not related to procedures of proven effectiveness. Belatedly, the precision that has been brought to bear in biomedical research is now being brought to health services research.