Measuring Health Care Quality

To the Editor. —Drs Jencks and Wilensky¹ say that a reason to develop their Health Care Quality Improvement Initiative was the finding that practice variation cannot be attributed to "known variations in patient sickness" and, hence, presumably may represent poor quality care.

In fact, there are few data to support this statement. The studies usually cited in support of it actually compared the general health of a population sample with the rates of performance of specific procedures in the geographic area from which the sample was drawn. For example, Roos and Roos² showed that the socioeconomic characteristics and health status of samples of aged populations in different areas did not correlate with rates of specific surgical procedures in these areas. Similarly, Wennberg and Fowler³ showed no relationship between population characteristics and procedure rates. Both studies surveyed the general population, not patients with particular problems who might have