Decision-making: a comparison of referral practice and primary care.

BACKGROUND: Should care by subspecialist physicians be more costly than care by primary care physicians? This article addresses diagnostic testing, one element of the answer to this question.

METHODS: A theoretical analysis was conducted of the sequences of testing, treatment, or watchful waiting in patients with low, intermediate, or high probabilities of disease. This was followed by a reanalysis of data from a previously published study of patients with chest pain from two referral populations and two primary care populations. The study used a chest pain score as a summary measure of the number of suggestive findings.

RESULTS: The analysis of sequences of testing, treatment, and watchful waiting suggests that patients with intermediate probabilities of disease are most likely to be referred. The study of patients with chest pain shows that the probability of disease for a given chest pain history score is higher in referred patients than it is in primary care patients, as is the proportion of patients with intermediate and high chest pain scores. This result is direct evidence that referral physicians get more patients with suspect but often uncertain histories. In general, the probability of disease given a particular history will be lower in primary care patients, and hence testing will be less fruitful.

CONCLUSIONS: Subspecialists are more likely to see patients who represent a diagnostic puzzle and have intermediate probabilities of disease. Since patients with intermediate probabilities of disease are most likely to benefit from testing, a per capita rate of testing that is higher than in a primary care practice might be appropriate in a subspecialist's practice.