In addition to analyzing individual attitude items, we created a summary Attitude Toward Gatekeepers scale. To create this scale, individual attitude items worded in a negative direction (eg, the gatekeeper undermines my relationship with patients) were scored so that a score of 4 indicated maximal disagreement. Items worded in a positive direction (eg, the gatekeeper increases the likelihood that patients will receive preventive care) were scored with 4 representing maximal agreement. The summary attitudes toward gatekeeper score was then computed by calculating the mean of the 5 separate gatekeeper items for each physician. This summary scale had a range of 1 to 4, with 2.5 indicating a neutral summary attitude. The Cronbach a for the summary scale was 0.75, indicating acceptable scale properties.
Analysis
Mean scores for the summary Attitude Toward Gatekeeper scale were compared according to physician demographics and practice characteristics using t tests and analysis of variance. For these unadjusted analyses, the physician payment variable was classified into 3 mutually exclusive categories: salaried, capitated for 40% or more of practice income, or fee-for-service payment accounting for 61% or more of practice income. We based the 40% threshold for capitated income on the assumption that this degree of capitation would be sufficient to change the underlying financial incentive experienced by physicians in regard to referral visit volume.
We also performed least squares regression analysis to investigate the independent association of physician and practice variables with the summary Attitude Toward Gatekeeper scale. All physician demographic and practice variables were entered into the regression equation, regardless of their significance on unadjusted analysis. For the regression model, payment mode was categorized in a manner different from that used for the unadjusted analyses. First, a dummy variable was created indicating whether the physician was salaried or nonsalaried. A second variable was included in the model indicating the percentage of practice income attributable to capitated payment. For salaried physicians, the value of this continuous capitation income variable was set at 0. This approach results in an interpretation of the coefficient for the salaried variable indicating the change in gatekeeper attitude scale score for salaried physicians relative to nonsalaried physicians with only fee-for-service payment.
Data were also analyzed after being weighted to account for the oversampling of nonwhite physicians and for the differences in sampling proportions among the different specialties relative to the overall population of physicians in each specialty in the study counties. The results were almost identical when we used the weighted and unweighted data; we therefore present results only from the more simple unweighted analyses.
The University of California, San Francisco, Committee on Human Research reviewed and approved the study protocol.
Results
Of the initial sample of 1750 physicians, 258 were subsequently determined to be ineligible, primarily due to death, retirement, or moving out of the study counties. Completed questionnaires were obtained from 979 of the 1492 eligible specialist physicians (66%). Sixteen of those responding worked in public clinics or other practice settings, such as schools or jails. Given the uniqueness of their practice settings and the small number of physicians there, we excluded these 16 and analyzed the responses of the remaining 963.
The characteristics of the physician respondents are shown in Table 1. Most (73%) were in solo or small office-based group practices of 2 to 10 physicians. Most had fee-for-service payment as their dominant payment method. One fourth were paid on a salaried basis, and 16% had at least 40% of their practice income paid by capitation.
Attitudes toward primary care physicians in the gatekeeper role were mixed Table 2. Almost half (44%) agreed that the gatekeeper undermines a specialist’s relationship with patients. Fifty-six percent agreed that the gatekeeper makes it more difficult to order expensive tests or procedures, and two thirds agreed that the gatekeeper decreases the freedom of the specialist to make clinical decisions. In response to the attitude items positing beneficial effects of a primary care gatekeeper arrangement, 40% agreed that the primary gatekeeper improves coordination of care, and half agreed that the gatekeeper increases the likelihood that the patient will receive preventive care. When all 5 questions were combined into a single summary scale, the general attitude of the specialist physicians toward primary care gatekeepers was essentially neutral, with a mean among all specialists of 2.4 (standard deviation=0.69) on a scale of 1 to 4.
On unadjusted analyses, practice setting and payment method were the strongest predictors of the summary Attitude Toward Gatekeeper score Table 3. Specialists in solo practice exhibited the most negative attitudes. The attitudes of specialists in small (2-10 physicians) and medium-sized (11-50) group practice settings were only slightly more favorable. Attitudes were much more positive among specialists working in large practice settings (>50 physicians) and especially among physicians working in group-model health maintenance organizations (P <.001) for overall difference across practice settings. Method of payment was also significantly associated with specialist attitudes (P <.001) for differences across payment categories. Salaried physicians demonstrated the most favorable attitudes toward gatekeepers and fee-for-service specialists the least favorable attitudes. Those specialists classified as capitated were on average neutral in their views of gatekeepers.