From this article it would seem that in-house attending physicians have little effect on outcome, whereas a commitment to a fellowship training program does. We must be cautious in interpreting these results. The amount of data collected, though concurrent, is small for the number of centers studied. In each cohort in each trauma center, there are on average less than 20 patients. The injury matching is balanced by Injury Severity Score, but the Abbreviated Injury Score used between centers is not standardized, which introduces variability.
Hoyt DB. Patient Outcomes in Academic Medical Centers—Invited Critique. Arch Surg. 2003;138(1):51. doi:10.1001/archsurg.138.1.51