The use of preventable adverse outcomes to study the quality of child health care. A new application of case-control research.

Because cohort methods are insensitive in detecting rare outcomes, the authors used the more sensitive case-control technique to investigate whether pediatricians or nonpediatric generalists are better able to recognize severe acute illness or to avoid preventable complications. We selected "indicator" outcomes for four types of common acute illness and enrolled patients who had contacted a physician more than 24 hours prior to an index emergency room visit for the same illness. Case (N = 103) were patients with adverse outcomes, i.e. potentially preventable complications or delayed diagnosis, treatment, or referral. Controls (N = 103) were those with acceptable outcomes (uncomplicated illness or prompt diagnosis, treatment, or referral) and were matched according to age, SES, and illness type. The overall results indicate no evidence of better care by pediatricians: prior contract with a pediatrician was associated with an estimated relative risk (omega) of adverse outcome of 1.32 (95% confidence interval, 0.76-2.29). Within the two largest illness categories (gastroenteritis and pneumonia-respiratory), the results were similar: omega = 1.16 (0.68-1.99) and 2.23 (0.82-6.04), respectively. Potential confounders were controlled for by multiple logistic regression analysis, but omega remained virtually unchanged. Nor were the findings altered by elimination of outliers or restriction of "contact" to office visits only. The authors conclude that study pediatricians and generalists provide equivalent acute illness care to children and that the case-control method provides a feasible and highly suitable approach to the study of the quality of medical care.