Estimating the size of family practice populations. Quadratic Odds Estimation.

Abstract

The thorough evaluation of a new health service model requires that questions of effectiveness be answered in the population to whom the service is offered and not just in the subpopulation that uses the service. This creates significant problems for researchers in primary care in which the norm is for a practice to serve an undefined population. With few exceptions, e.g., the British National Health Service, it is rare for patients to specify the primary care practice from which they intend to seek services. Thus, most practices can identify and count the patients that they see, but cannot determine the number of nonattending patients who are also members of their population. The inability to enumerate practice populations was first raised by epidemiologic researchers who needed population figures for the calculation of prevalence rates; hence the coining of the sobriquet denominator problem.'2 Attempts to solve this problem have taken two general approaches:3 the development of patient registers and the application of mathematical models.