While a majority of patients in our study reported being counseled, their physicians did not address most of the 5As, thus confirming the need to improve physicians’ quality of obesity counseling.8
Our study did not find significant bivariate differences in obesity counseling between residents in the intervention and control groups. However, when theoretically-supported patient, visit and provider characteristics were taken into account, the obesity counseling curriculum was associated with higher quality of counseling. This is encouraging giving the challenges of determining the direct effects of training on clinical care and supports efforts to train physicians to deliver high quality obesity counseling. This study also highlights that there are many factors besides physician training that influence the quality of care a patient receives.
Patients with a higher BMI were more likely to receive counseling. This is not surprising since both patients and physicians are more likely to identify obesity as a problem when patients weigh more.31
Higher self-reported health status was associated with higher quality physician counseling which supports previous work showing that counseling is most likely to take place when there are fewer competing issues.32
Having a female physician was associated with higher quality of obesity counseling; female physician gender has been shown to predict higher rates of preventive care counseling in other studies.33
Greater continuity of care with the physician was associated with higher quality of counseling--physicians may have more time to counsel patients they have seen previously. Paradoxically, considering the physician to be one’s primary care provider was associated with lower quality counseling—future research is needed to confirm and elucidate whether and why this might be true. Unfortunately, female gender of the patient was associated with decreased quality of obesity counseling. This is concerning and warrants further study because obesity is more common in females.1
While this was a largely negative study, it has important strengths. We tested the impact of our curriculum on patient care. Few studies have examined the impact of medical education interventions on real-world physician performance34
possibly because attempts to show a significant impact of medical education are hindered by the relatively small anticipated effect of curricula on outcomes35
when compared to the myriad of other factors that impact patient care32
. Further, we developed a conceptual model and used it to identify important variables which could be expected to influence obesity counseling. While our study did not account for all potentially important factors (e.g., year of resident training, patient activation, language and/or cultural concordance) it demonstrated the need to develop more nuanced models to better identify evidence of educationally sensitive variance in outcomes. This may guide the design of future medical education studies to capture the impact of training interventions on patient outcomes.
There are several limitations to this study. Given the small sample size, our study was underpowered to detect small differences. Residents who did not receive the curriculum may have learned from their peers in the intervention group, diminishing the measured differences between the groups. Generalizability is limited because we studied only one training program, although we anticipate using a more diverse sample of residents would yield substantially larger curriculum effects. Our residents had a higher rate of counseling in both groups than is reported in the literature (>70% vs. 20–30% in other studies (14, 31)), which may have produced a ceiling effect, diminishing the detectable difference between the two groups. Post hoc power calculations suggest that we would have needed a sample size of >700 to have minimum power (0.80) to detect the simple bivariate difference in quality of counseling we found between the curricular groups (about 5%).
There are also limitations in our study design and analysis. We could not randomize the residents to the intervention due to fixed rotation schedules. While patients were blinded to intervention status, the residents were not. There may have been some variability in the quality and accuracy of weight and height measurements. The study was cross-sectional, and by not collecting data from multiple visits, we may have missed important curriculum effects. Our findings may not hold up with different patient populations although we suspect that many of the same core patient and visit characteristics will be important.
Our measures are also limited. Ideally, we would have liked to have assessed more physician and visit characteristics but were limited both by the availability of data (determined by respondent burden and resource constraints) and by our small sample size which precluded the inclusion of too many variables. We relied on patient report to assess the quality of obesity counseling which may have produced a recall bias. While patient report has been shown to be superior to physician report and chart reviews as a method to assess physician practice36
, we cannot determine the accuracy of patient reporting in our study.
This study suggests avenues for further research. We plan to test the impact of this obesity curriculum using larger numbers of residents from multiple institutions. We also plan to conduct longitudinal follow-up of patients to see if training physicians to deliver higher quality obesity care leads to improved outcomes such as patient activation, intention to lose weight, and weight loss.