Accuracy and consistency of quadratic odds estimates.

Abstract

In medical practices that do not have rosters, only the number of patients who come to the practice can be enumerated: the number who might have visited if they had had a reason to do so remains unknown. The Quadratic Odds Estimator is a technique for estimating the total number of patients cared for by a primary care medical practice, including the non-visitors. A revised version of the model is shown to have an error of less than 1% in predicting the number of patients at risk of visiting a primary care medical practice. Aggregate and sex-specific estimates of total practice size are shown to be comparable to within 2%. The model estimates the prevalence of hypertension among the patients of two family practice resdencies as 18 and 11%. The rationale for employing unconventional regression weights and dual regressions is explained.