The ecology of medical care: origins and implications for population-based healthcare research

Health Services Research, April, 1997 by Kerr L. White

An example of the contemporary uses of hospital discharge abstract data is to be found in Vermont. In 1962, I was attracted to an opportunity at the University of Vermont because its relatively well-defined general population would facilitate the study of that state's health services, especially its hospital care. Out department proceeded to install a hospital discharge abstract system in all Vermont hospitals to exploit this opportunity. In 1964, after moving to Johns Hopkins, I interested one of our students, Jack Wennberg, in taking a job in Vermont in order to analyze this gold mine of hospital discharge data (Wennberg and Gittlesohn 1973). His superb contributions over the succeeding decades have shown the remarkable power of small-area comparisons for illuminating the complex factors that influence the content and quality of care under different auspices and in different jurisdictions.

There are equally complex problems associated with the labeling of health and medical problems. Each of the several populations depicted by the "ecology" squares requires somewhat different labels or nomenclatures and different classification schemes to describe that population's health problems or-when records are linked as they have been in Manitoba, Canada-the same populations as they move through life and the healthcare system (Roos and Roos 1992; Wolfson et al. 1993).

The International Classification of Diseases was originally designed for arraying causes of death (World Health Organization 1975). Slight modifications were made over the decades for nonlethal "diseases," and with the Ninth Edition further so-called Clinical Modifications were made (ICD-9-CM) to accommodate the need for labels required for hospitalized patients (Commission on Professional and Hospital Activities 1978).

More recently, the International Classification of Primary Care was developed for use at that level of care and was widely adopted in Europe (Lamberts and Wood 1987; Lamberts, Wood, and Hofmans-Okkes 1993). It recognizes the long-standing observation that 60 percent of the problems brought by general populations to sources of general or primary care cannot be labeled as "diseases" (College of General Practitioners 1966). They are patients' problems, concerns, complaints, symptoms, and assorted "conditions," including a wide variety of social and psychological problems that are the day-to-day fare of general physicians. These also represent the initial manifestations of what later may be labeled by nurses and physicians as "diseases," some of which may eventually lead to death. The unwarranted bias resulting from attempts to cram every patient's unique problem into a "disease" category leads to huge distortions in measuring the burden of illness in different clinical settings. When measures of "severity" and "urgency" are lacking, the distortions are compounded.

In addition, of course, we need labels and classifications for Lay Terms used by the general public in describing their ailments and concerns. Why should the public be required to describe their problems with the Latin or Greek lingo used by the medical profession? We would undoubtedly obtain more accurate and reliable data by using patients' own terms as they move through healthcare systems. I don't expect these changes to come to pass any time soon in this country, but it is of major interest in developing countries. Reasons for Encounter - in other words, for seeking care-constitute yet another classification. The patient may bring a problem to the source of care because he or she read about it in the paper or a spouse insisted that "something be done." And then there are classifications of interventions and procedures of which we have several in vogue; standardization of these is urgently needed.