Testing for “Schools of Thought” in Physician Practice Patterns

The “schools of thought” model, first advanced by Phelps and Mooney (1993), posits that a physician’s practice patterns are formed during residency training and persist afterwards.  Regional variations in practice patterns result from similarly-trained physicians remaining near their residency programs.  Empirical evidence on “schools of thought” has been both limited and mixed.  I seek to extend our understanding of this phenomenon by testing three distinct aspects of the “schools of thought” model.

First, I examine the associations between intensity of care at VA hospitals, affiliated academic hospitals, and unaffiliated non-VA hospitals in the same regions.  Intensity of care is measured as the mean number of inpatient days per patient during the last two years of life among patients with nine serious chronic conditions.   If there are “schools of thought” we would expect that practice styles at VA hospitals would be influenced more strongly by their academic affiliates, particularly those that are nearby and share many residency programs, than they would be by unaffiliated hospitals in the same market.  Preliminary evidence shows that the influence of unaffiliated hospitals in the same market on average was roughly equal to the influence of a large and close academic affiliate hospital.

Second, I examine the evolution of practice styles of housestaff trainees on the medicine service at a large academic medical center.  Practice style is measured as mean number of routine tests ordered per patient day, adjusted for patient characteristics, clinical setting, and other members of the care team.  If there are “schools of thought” we would expect that trainees’ test ordering propensities would show signs of convergence as their residencies advanced.  Preliminary findings do not show any evidence of convergence at least through the intern year.

Third, I examine the influence of the aggregate practice style at the hospital at which a physician trained on that physician’s practice style in her first job after training and compare that to the association between the physician’s initial practice style and the aggregate practice style of other physicians in the same hospital.  Practice style is measured as the propensity to use drug-eluting coronary stents vs. bare metal stents, which varied greatly across centers and changed dramatically during the study period for exogenous reasons.  If there are “schools of thought” we would expect that the style where a physician trained would have more influence on the physician’s own style than would the aggregate style of the physician’s new peers.

Our findings to date, which are preliminary, cast doubt on the power of medical training programs to impose consistent practice patterns on their trainees.