Family physician attitudes in managing obesity: a cross-sectional survey study

Over the last 20 years obesity has risen to epidemic proportions, leading to the latest recommendation by the US Preventive Services Task Force that adults be screened for overweight and obesity (via calculation of the Body Mass Index) and that identified cases be followed up with weight loss counseling and behavioral modifications [1]. Implementation of these recommendations in primary care settings, however, has been difficult. For example, Ma and colleagues found that physicians who had participated in the National Ambulatory Medical Care Survey had only identified 29% of adult obese patients during office visits [2]. Similarly, Yaemsiri and colleagues (2010) have reported that 74% of overweight and 29% of obese individuals have not received a formal diagnosis of overweight or obesity, using data from the 2003 to 2008 National Health and Nutrition Examination Survey. They point out that primary care providers "have unused opportunities to motivate their patients to control and possibly lose weight by correcting weight perceptions and offering counseling on healthy weight loss strategies" [3]. Many barriers that primary care physicians face when implementing the guidelines for the prevention and treatment of obesity have been described in the literature for both adults and children, in a variety of settings, and across international borders. These include negative characterization of obese patients, lack of support services, failure to implement office systems to screen for obesity, and lack of training and self-efficacy in counseling for obesity and general lifestyle management [4–21].

To examine this problem on a local level as a needs assessment for further work in this area, we initiated an attitudinal survey of primary care physicians located in Central New York in the Spring and Summer of 2007, asking about perceived causes of obesity, comfort with and accommodations for obese patients, and barriers to implementation of interventions. The survey was administered as part of an inaugural recruitment effort for a regional practice-based research network (PBRN) centered in Syracuse, New York, and included rural, suburban, small city, and mixed practices.

Approximately 204 survey invitations were sent to regional primary care physicians via both an e-mail invitation to participate in an online version of the survey, as well as a paper copy and letter invitation, with 75 responding. We were able to obtain usable data for 67 respondents with completed surveys, and 5 partially completed surveys. Respondents were all family physicians, with the exception of a single internal medicine specialist. Reported practice settings were rural (N = 35) or suburban (N = 25), with 5 urban and 2 "other" settings specified. The largest urban setting in Central New York is the city of Syracuse, itself relatively small and surrounded by suburban and rural areas. For the purposes of analysis, suburban, urban, and other practice settings were reduced to a single "non-rural" category, due to the level of similarity between urban and suburban practices in our region. No significant differences were found between these settings on any measure.

The mean (SD) number of years in practice for the sample was 20.16 (7.19), and the mean age was 50.72 (8.31). Slightly over 73% of the sample was male (male n = 49, female n = 18), 95% White (64/67) and 99% non-Hispanic (66/67). Self-reported rural physicians were also significantly older than non-rural physicians (rural mean age: 54.80 years, SD = 6.98, non-rural mean age: 46.12 years, SD = 7.27; p < .000), and had more years in practice (rural: 23.97 years, SD = 7.23, non-rural: 16.00 years, SD = 6.46; t = -4.745, p < .000). Of the 75 total responses, 41 were returned online, and 34 were returned on paper. There was no significant difference between the two groups in terms of age (online responder average age was 51.3, vs. 49.8 for mail responders). The gender distribution between response types mirrored the overall gender breakdown, with 30 of 41 online responses coming from males (73.17%), whereas 20 of 34 paper responses came from males (58.82%), a difference which was statistically significant (χ² = 4.836, p = .028).

Correlation analysis between family physician characteristics and factor scores for each of the 7 identified factors described above revealed significant inverse relationships between the behavioral factor and age (r = -.320, p = .017) as well as years in practice (r = -.344, p = .010). The relationship between the physician dissympathy factor and both age (r = .248, p = .067) and years in practice (r = .253, p = .062) approached traditional levels of statistical significance. We also found a significant difference in age (p <. 01) and years of experience (p < .01) between our self-reported rural vs. non-rural physicians. In our sample, the rural family physicians were significantly older than non-rural family physicians (means (SD) = 54.8(6.98) and 46.12(7.27), respectively) and had significantly more years of experience (means {SD} = 23.978 {7.23} and 16.00 {6.46}, respectively).

We performed analysis of covariance (ANCOVA) to compare the seven factor scores between self--reported rural and non-rural physicians using years of experience as a covariate. Our analyses revealed significant differences between self-disclosed rural physicians and non-rural family physicians on the psychobehavioral factor score. Specifically, self-reported non-rural family physicians tended to endorse psychobehavioral contributors to obesity (repeated dieting, psychological problems, repeated restaurant eating) more than rural physicians (p = .013), even after factoring in years of experience. None of the other factor scores were significantly different for self-reported rural vs. non-rural family physicians.

This study describes a general sense of frustration among family physicians in managing obesity - stemming from a lack of adequate compensation for treatment efforts, a general lack of self-efficacy, and a complex set of factors that seem to be perceived as beyond the physician's control. These findings parallel the results of the study by Foster, et al. [8], from whom the survey instrument for this study was adapted.

An interesting finding for this study is the detection of a difference in physician opinions and attitudes regarding the causes of obesity, between self-defined rural and non-rural physicians in our sample. This difference accounted for approximately 19% of variance in the attribution of psycho-behavioral factors to obesity, without the inclusion of any other independent variables. Our sample was admittedly small, limited in geographic scope to one section of New York State, and essentially limited to one medical specialty (Family Medicine), so the generalizability of this specific finding requires verification in other larger and more diverse samples. However, a corollary to the split on this particular factor along rural/non-rural lines is that such a finding may indicate a substantive difference in belief sets between those who practice in rural settings, vs. those who practice in other, non-rural settings, as Rabinowitz and others have maintained [22].

The fact that rural physicians in our sample tended to be older - and consequently have more years in practice - may be tied in some way to the opinion split, although we found no explicit statistical effect indicating this is the case. Nevertheless, it is possible that a more sensitive analysis in a more diverse sample might indicate a stronger effect for age or practice-length differences. Additionally, in our sample, these two variables were so closely correlated as to be nearly indistinguishable statistically. It is not unreasonable to suspect that, in a sample where age and years in practice were not so closely correlated, an age, practice-length, or even a generational effect might be demonstrable. Samples containing a large number of physicians who entered the profession later in life, for example, might indicate generational effects. Regardless, such an analysis was not possible with our current data set. When viewed alongside the fact that we found nearly significant indications of lower empathy and greater discomfort in older and more experienced physicians, it is clear that additional study of demographic effects upon physician beliefs and practices may be warranted.

A potential weakness of the conclusion based on rurality is the limitations of self-report data in defining rural practice. In the case of the population we surveyed here, as well as the particular practices surveyed, we highly suspect that the self-reports of practice setting given by our respondents often reflected the population primarily seen by the practice, as opposed to the physical location of the practice. However, while self-report may be an apparently unreliable source of information about the physical location of the practice, we suspect it is still useful for delineating physicians who view themselves as treating rural patients, as opposed to those who do not perceive themselves to be doing much or any such practice.

There are a number of limits to the generalizability of this study. The sample was relatively small and homogenous, representing essentially one medical specialty in one region. Additionally, the differences between practicing in a small city, a suburb of that city, or the outlying rural areas, may be somewhat small, compared with a national sample that including both very large, major urban centers as well as very remote rural regions. However, this apparent weakness may also serve to further emphasize differences we did find related to self-reported practice setting. The collected sample was over-representative of white/Caucasian physicians, and slightly over-representative of males compared to the membership of the American Academy of Family Physicians (73% male in our sample vs. 63% male AAFP membership) as of 2009 [23]. The survey also did not ask physicians about their own perceived responsibility in managing their patients' weight. A number of studies have identified questions on the part of primary care providers about the extent of physician responsibility for patient weight loss, and this seems to be an underlying factor at play in the current analysis [13, 18, 24, 25]. Finally, the study did not ask respondents to comment upon their use of or referral to commercial weight loss products and programs. As this industry continues to grow, the use, implementation and acceptance of such products, services and systems warrants additional study.

The current study indicates that, regardless of setting, primary care physicians appear to need more precise guidelines and better tools for screening and management of obesity, more referral options, better reimbursement for obesity-related services, improved coordination with non-physician providers, and reimbursement of patients for activities such as dietician consultation and fitness activities. Additionally, issues such as culturally appropriate care and discomfort with the subject of obesity may be partially ameliorated with training. For example, better word choice may improve the discussion when physicians are uncomfortable broaching the subject of obesity [26].

More globally, the barriers identified align with those that would be addressed by the adaptation of the principles of the Chronic Care Model and the Patient-Centered Medical Home, both by providers and insurers [27–31]. The Chronic Care Model, originally championed by Wagner [32, 33], involves the use of patient education, systems-level disease management protocols, goal setting, and patient self-management. When incorporated into a broader framework of patient-centered practice, CCM-based management of obesity (and reimbursement of the primary care team for it) would likely be optimal in addressing the recommendations listed above.

This study on the barriers to obesity management in primary care reveals a number of areas requiring intervention as well as further study, but was a pilot study, made possible with seed funding for the creation of a new PBRN. With only 75 responses from a limited geographic region, we have reinforced previous study in the area of barriers, and we believe our results offer preliminary evidence of attitudinal differences toward obesity and obese patients between those physicians who self-identify as rural, vs. those who do not. More research in this area is clearly needed. While greater education of physicians in the topic of obesity causation and management is needed, a more systemic approach to improving resources, patient self-management and the healthcare system's response would more completely address the barriers found in our study.