Direct attending supervision during rounds did not significantly reduce the rate of medical errors but led to subjective decreases in resident autonomy compared to no attending supervision, a recent study found.
On an inpatient general medical service at one large academic medical center, 22 faculty members participated in separate two-week arms in random order over a nine-month period. In the standard supervision arm, attendings joined bedside presentations of newly admitted patients but did not join resident work rounds on established patients. In the intervention arm, they provided more direct supervision by joining both new patient presentations and work rounds on previously admitted patients.
Researchers examined the rate of medical errors, the primary safety outcome, in 1,259 patients (5,772 patient-days). Since most errors were anticipated to be minor, they chose a 40% reduction in errors for clinical significance. They also assessed resident education by evaluating resident participation on rounds and measuring resident and attending educational ratings in surveys. Results were published June 4 by JAMA Internal Medicine.
The medical error rate per 1,000 patient-days was not significantly different between the standard and increased supervision arms (107.6 vs. 91.1; 15% relative reduction; 95% CI, −36% to 9%; P=0.21).In a time-motion study of 161 work rounds, there was no significant difference in mean length of time spent discussing patients between the two models (202 min [95% CI, 192 to 212 min] vs. 202 min [95% CI, 189 to 215 min]; P=0.99).
Interns spoke significantly less when an attending physician joined rounds (64 min [95% CI, 60 to 68 min] vs. 55 min [95% CI, 49 to 60 min]; P=0.008). When attendings were present, a lower proportion of interns reported feeling efficient (41 [55%] vs. 68 [73%]; P=0.02) and autonomous (53 [72%] vs. 86 [91%]; P=0.001). Residents also reported lower rates of feeling autonomous with an attending present (11 [58%] vs. 30 [97%]; P<0.001). However, attendings said the quality of care was higher (20 [100%] vs. 16 [80%]; P=0.04) and that they knew the team's plan of care better (20 [100%] vs. 12 [60%]; P=0.002) when they joined work rounds.
The study authors noted limitations, such as the single-center design and the potential for limited generalizability to other settings. They added that their methodology could have missed relevant errors and reduced the overall power to detect statistically significant reductions.
An accompanying editorial echoed these limitations, calling for more research on the effects of supervision on medical errors and on residency training. “A one-size-fits-all model may not be appropriate. Yet, that may be exactly what we are doing in the current era of internal medicine training in which academic hospitalists are serving as the primary educators,” the editorialists wrote.
Supervision of trainees should not be viewed as a hindrance to education but as a challenge of balancing mentorship and autonomy, they said. “To help guide young physicians toward independent practice, we need as much time with our trainees as we can get, but really only as much as they need,” they wrote. “Our faculty need the training as educators to provide that highly nuanced and balanced approach.”