This paper is unique because it examines the relationship between patient–physician gender concordance and weight-related counseling. Previous research links gender concordance to cardiovascular risk factor control16
and preventive health services17,18
; however, a link has not been established between gender concordance and weight-related counseling. Contrary to the stated hypotheses, patients seeing physicians of a different gender had similar odds of weight-related counseling as patients seeing physicians of the same gender. Also inconsistent with the hypotheses was the finding that obese men seeing male doctors were more likely to receive diet/nutrition and exercise counseling than obese women seeing a female doctor.
The finding that patients in gender-discordant relationships had similar odds of receiving weight-related counseling as their female gender-concordant counterparts may partially result from the perception that women are more likely to be dissatisfied with their weight.23
Overweight women delay seeking health care due to embarrassment about their weight and wanting to avoid being lectured about their body weight.24
Because women may be more sensitive about weight-related discussions, female physicians may choose to avoid weight-related discussions with their obese patients and male physicians may avoid weight-related discussions with their obese female patients. Male physicians may also be reluctant to provide weight-related counseling to obese women if they perceive that women have unrealistic weight-loss goals, as previously reported in the literature.25
It is unclear why obese male patients seeing male physicians had higher odds of receiving weight-related counseling than obese women seeing a female physician. Perhaps societal norms linking physical fitness to masculinity leads male physicians to view obese men as more receptive to weight-related counseling and contributes to open dialogue about weight in male gender-concordant relationships.
The findings of this study should heighten clinicians' awareness of how the personal attributes of physicians and patients may influence obesity care. Future studies should objectively measure weight-related communication (e.g., direct observation, audio tapes) in gender-concordant and gender-discordant patient–physician encounters, and explore the potential role of physicians’ explicit and implicit attitudes regarding obesity and gender, in weight-related counseling for obese patients. In addition, future studies should examine outcomes (e.g., weight loss) of weight-related counseling provided among different gender groups.
This study has some limitations. First, the data are cross-sectional, which may have led to underestimation of weight-related counseling if they were provided during a visit not captured by the data set. However, gender groups should not have been differentially affected. Second, the cross-sectional design limits causal inferences. Third, it was not possible to control for the duration of the patient–physician relationship, which may affect delivery of weight-related counseling. However, the analysis adjusted for whether the patient was previously seen by any provider in the practice and there was no observed association between this variable and weight-related counseling. Fourth, it is not possible to know the extent, quality, or impact (e.g., referral for additional evaluation) of the weight-related counseling patients received or the rationale for which it was provided. Fifth, it was not possible to include other measures which might differentially affect weight-related counseling in gender-concordant and gender-discordant pairs such as physician knowledge, physician attitudes toward obesity management, physician obesity or gender stereotypes. Finally, missing data for some covariates (e.g., physician race) decreased the sample size from 5667 to 3969 in the adjusted analyses. Although missing data could potentially introduce bias in the analyses, they should not have differentially affected the gender-concordant and gender-discordant groups and likely biased the results toward the null.