Abstract
Purpose:
To determine whether primary care physicians can accurately assess body mass index (BMI) by visual inspection and to assess barriers related to the diagnosis and management of obesity.
Design:
Prospective Survey Study.
Setting:
Hospitals and Clinics.
Subjects:
Primary care providers in the fields of Internal Medicine, Family Medicine and Obstetrics/Gynecology.
Measures:
Measures investigated included providers visual assessment of BMI, BMI knowledge, diagnosis and management of obese patients, and perceived barriers to treatment.
Analysis:
Top and bottom quartiles and total scores were determined for responses regarding the reported management of obesity, reported comfort with care, and reported barriers to care and used as the cut point. Statistical analyses were utilized to examine relations and compare groups.
Results:
206 (74%) of the 280 eligible providers completed the survey. The accuracy of visual assessment of BMI was 52%. Physicians were more likely to underestimate BMI than overestimate (36% ± 4% vs 12% ± 6%, respectively, P < .001). Although 91% of providers report routinely calculating BMI, only 61% routinely discuss BMI. Providers feel comfortable providing exercise (72%) and dietary counseling (61%). However, fewer are comfortable prescribing medical (16.4%) and surgical options (36%).
Conclusion:
Visual assessment of BMI is not reliable. Primary care physicians in our study population do not consistently discuss obesity with their patients and many report insufficient knowledge with regard to treatment options. Further studies are needed to determine whether these results are valid for other physicians in various practice settings and to mid-level providers. In addition, research is needed that investigate how collaboration with providers outside the medical field could reduce the burden on physicians in treating patients with overweight or obesity.
Purpose
Obesity is a pressing public health concern in the United States 1 -6 and underlies many of the leading causes of preventable death including cardiovascular disease, stroke, type 2 diabetes, and cancer of the breast, endometrium, and thyroid. 3,6 -11,12,13
Despite its importance, health-care providers do not consistently address the issue of overweight or obesity directly with their patients. 8,14,15,16 Less than half of obese patients are informed of their status from a health-care professional, and even fewer obese patients receive weight-related counseling. 3,8,14,17,18 Yet, even minimal discussion of weight (when at-risk) by their provider can impact patient behavior. 19 Overweight or obese patients who are told they are overweight by their doctors have more realistic perceptions of their own weight, a greater desire to lose weight, and have more recently attempted weight loss. 10,20
The purpose of this current study was to better understand factors related to the diagnosis and management of overweight and obesity among primary care physicians. We hypothesized that physicians have difficulty diagnosing overweight and obesity by visual inspection alone and therefore may not be prompted to address it during a patient care visit. Our secondary hypothesis was that physician-, patient-, and system-related barriers experienced by providers may interfere with their ability to help patients manage and/or improve their BMI.
Methods
Survey Design
Men and women were recruited by study investigators (B.A.P. and A.H.) at a public venue in a racially diverse urban neighborhood on the south side of Chicago. Following informed consent, participants were escorted to a private location where they were photographed in front of a white backdrop with markings at 5” and 6” followed by height and weight measurements (rounded to nearest half inch and half pound). Photographs and measurements were performed by study investigators (B.A.P. and A.H.). Individuals were selected to achieve a full range of BMI categories, that included underweight (< 18.5), normal (18.5-24.99), overweight (≥25), obese (≥30), and morbidly obese (≥40). 21,22 Participants were also selected to achieve a gender balance. Photographs were deidentified by blurring facial features using photoediting software (Figure 1).
Exemplary photographs.
Survey questions were created based on a combination of clinical obesity treatment expertise and a careful review of the literature to address key topics in the identification and management of overweight and obesity. The questions included provider demographics such as race, ethnicity, gender, training background, years of experience, type of workplace, and patient population were collected. To evaluate providers’ ability to assess BMI via visual inspection, 19 separate photographs were presented (Figure 1), and participants were asked to identify the correct BMI category. Subsequent questions focused on BMI knowledge where respondents were asked to correctly identify the BMI category when presented with a specific BMI (eg, “A patient with a BMI of 41 is 1 = underweight, 2 = normal weight, 3 = overweight, 4 = obese, and 5 = morbidly obese); practice behaviors focusing on the diagnosis and management of overweight and obesity (eg, If my patient has an elevated BMI, I document overweight or obese in the problem list or as a diagnosis: 1 = almost never, 2 = rarely, 3 = sometimes, 4 = often, 5 = almost always); as well as perceived barriers (eg, It is hard for me to talk about overweight and obesity with my patients who are overweight or obese: 1 = strongly disagree; 2 = disagree; 3 = slightly disagree; 4 = somewhat agree; 5 = agree; 6 = strongly agree), and comfort levels in the management of overweight or obese patients (eg, I feel comfortable discussing possible surgical interventions for the treatment of morbid obesity: 1 = strongly disagree; 2 = disagree; 3 = slightly disagree; 4 = somewhat agree; 5 = agree; 6 = strongly agree).
Providers were also given free text options to further elaborate on their responses for questions concerning their actions and behaviors, attitudes, as well as comfort with treatment options. The final survey was administered via the Web-based survey instrument, Research Electronic Data Capture (REDCap). 23
Sample
The survey was sent to primary care providers in the fields of Internal Medicine, Family Medicine and Obstetrics/Gynecology practicing in urban and suburban centers in Illinois. Subspecialists in any of the fields were excluded. The survey was administered via survey e-mail link or via paper survey per provider preference. All responses were entered into the REDCap survey platform.
Measures and Data Analysis
We used SPSS Version 23 in all analyses. Following the calculation of descriptive statistics and frequency of responses, we calculated total summed scores for barriers to obesity care (15 items), comfort providing obesity care (7 items), and comprehensive obesity care (11 items), which was comprised of items representing diagnosis and management practice behaviors. Item stems for these 3 scales may be found in Tables 1 –3.
Physician Report of Management of Obesity.
Abbreviation: BMI, body mass index.
Physician-Reported Barriers to Management of Obesity.
Physician-Reported Comfort With Management of Obesity.
In order to evaluate preliminary psychometric properties for these scales, we used Classical Test Theory–based analyses of reliability and validity. First, we determined the internal consistency reliability of these summed scale scores by calculating the Cronbach α statistic, which is estimated on a scale from 0 to 1. Items that “hang well together” as a latent construct score ≥.70 on Cronbach α. We also evaluated item-total correlations, of which a general rule of thumb is that items ≥.40 are considered acceptable for inclusion in a scale. Finally, using Spearman Correlation Coefficient, we examined the associations of these 3 scales with one another and also with BMI Knowledge (total number correct), hypothesizing that Total Comprehensive Care would be moderately positively associated with comfort and knowledge, and moderately negatively associated with barriers.
Next, we created top and bottom quartiles for the 3 scales, which were utilized to discriminate between providers at the top and bottom of the spectrum to better evaluate the characteristics and differences between these groups. Chi-square analysis, Fisher exact test, Student t-test, and Spearman correlation coefficient were utilized to examine relations and compare groups. A P value of <.05 was considered statistically significant.
Results
Survey Respondents
The survey was administered to a total of 285 physicians, and 206 (73.6%) of the 280 eligible physicians responded (Figure 2). The majority of participants were Caucasian (70.9%), non-Hispanic (96.1%), female (59.7%), and worked at either an integrated health system (47.1%) or a university-affiliated practice (43.7%). The majority were in the field of internal medicine (51.9%). Nearly half (46.7%) reported practicing medicine for greater than 11 years, while 33.5% were resident physicians. A substantial proportion of physicians reported that they cared for a mix of publicly and privately insured patients (32%) and a racially and ethnically diverse patient population (42.7%, Table 4).
Completed surveys for analysis.
Demographics of Respondents.
Preliminary Psychometric Properties of New Summed Scale Scores
All 3 scales demonstrated high internal consistency reliability (Barriers α = .91; Comfort α = .88; Comprehensive Care α = .87). Item-total correlations were ≥.40 for all items outside of 2 items for the Barriers scale (eg, too little time: ρ = .38; language barriers: ρ = .37), and 2 items for the Comprehensive Care scale (eg, use BMI to determine abnormal weight: ρ = .32; BMI is calculated for all patients: ρ = .35). Despite this, we chose to retain all items as they did not have any effect on internal consistency values (eg, omitting them would not have led to superior estimates). With the exception of BMI knowledge (which was not significantly associated with any other scale), correlation coefficients were as hypothesized: Comprehensive Care was moderately positively associated with Comfort (ρ = .54, P < .01) and moderately negatively associated with Barriers (ρ = −.45, P < .01). Overall, we felt comfortable proceeding with planned analyses given the performance of these summed scale scores.
Body Mass Index Assessment Accuracy by Visual Inspection
The accuracy of assessment of BMI by visual inspection was 51.9%. Providers were able to assess normal weight with more accuracy (71.8%) compared to other categories (Table 5). Physicians underestimated BMI more frequently (36.2% ± 4%) than they overestimated BMI (12.4% ± 6%).
Visual Assessment Results.
Abbreviation: BMI, body mass index.
Body Mass Index Knowledge and Practice Behaviors
Physicians were able to correctly select the appropriate BMI category the majority of the time: underweight (71.8%), normal (91.7%), overweight (98.5%), obese (94.2%), and morbidly obese (93.2%).
The majority of physicians reported that they often or almost always calculate BMI (91.3%), while only 82.6% reported often or almost always utilizing BMI in the management of their patients. For the population of overweight or obese patients, 60.7% reported routinely discussing BMI with their patient and 59.1% reported providing weight loss counseling (59.1%). Even fewer physicians (40.2%) reported often or almost always documenting BMI in the patient chart (Table 1).
Yet, only a minority of providers agreed or strongly agreed that they experienced barriers to providing obesity care (Table 2). Although the majority of providers reported high comfort levels with dietary and exercise counseling, fewer reported high comfort with medical or surgical interventions (Table 3).
Comprehensive obesity care was defined by those providers who fell within the upper quartile (scores ≥ 46). There was no significant difference between physicians who provided comprehensive care compared to those who did not with regard to age, gender, race or ethnicity, experience, practice insurance or practice race, or ethnicity (data not shown). However, there was a significant difference between physician specialty and reported comprehensive obesity care (P < .002). Family Medicine (43%) and Internal Medicine (33%) physicians were more likely than Obstetricians and Gynecologists (10%) to provide comprehensive obesity care.
High barriers to obesity care was defined by those providers who fell within the upper quartile (scores ≥ 44). There were no differences in those physicians who reported high barriers to care as compared to those who did not with regard to any of the physician characteristics such as age, race, gender, specialty, or physician experience (data not shown).
High Comfort Level for Management of Overweight and Obesity
High comfort level was defined by those providers who responded in the upper quartile (scores ≥ 31). There were no significant differences between those physicians who reported high comfort levels as compared to those who did not with regard to race and ethnicity, practice insurance or practice type, or ethnicity (data not shown). Men were more likely than women to report higher comfort levels in the management of overweight and obesity (40.7% vs 21.5%, respectively, P < .003). In addition, older physicians (45-54, 58.5%) and physicians aged 55 and older (39%) were more likely to report higher levels of comfort as compared to younger physicians aged 35 to 44 (26.3%) and aged 25-34 (11.0%, P < .001). Physicians who work in Integrated health systems are significantly more likely to report high comfort in managing obesity (43.6%) as compared to those at a University-affiliated practices (14.6%), community health center (20%), or other (28.6%, P < .001). Family medicine physicians are more likely to report high comfort levels (42.5%) as compared to Obstetricians and Gyncecologists and Internal Medicine physicians (8.6% and 35.9%, respectively, P < .001).
Association Between Comprehensive Care, Barriers, and Comfort
Comfort was inversely correlated with barriers (ρ = −0.36, P < .001) and positively correlated with comprehensive care (ρ = .54, P < .001). There was no significant correlation between comprehensive care and ability to correctly assess BMI by visual inspection P > .05). Among respondents who provided comprehensive care, a minority agreed or strongly agreed that they experienced difficulty discussing their patient’s weight (6.4%), or felt uncomfortable with this discussion (5.1%), believed patient discomfort was a barrier (4.2%) or believed their patients would feel stigmatized by the discussion (4.2%). However, 25.8% strongly agreed or agreed that time was a barrier to providing obesity care. Further, comprehensive care providers agreed or strongly agreed that they felt comfortable with dietary counseling (67.5%) or exercise counseling (93.2%) and fewer strongly agreed or agreed to being familiar with the indications (46.6%) and contraindications (18.1%) for weight loss medication (46.6%). Only 27.6% of these physicians agreed or strongly agreed that they felt comfortable prescribing weight loss medication.
Discussion
This study sought to better understand the potential barriers to the diagnosis and management of overweight and obesity among primary care physicians. We were particularly interested in a physician’s ability to diagnosis overweight or obesity by visual inspection alone. The visual assessment of an individual in a provider’s office may serve as a prompt to discuss relevant health issues. Our findings confirm that visual assessment of BMI is inaccurate and calculation of a patient’s BMI is essential in properly diagnosing and managing obesity. This finding has implications for clinical practice and emphasizes the importance of accurate measurement of height and weight for each patient. Similar to previous studies, our study demonstrated that physicians on average have a 51.9% accuracy in the visual assessment of BMI. 3,24 This inaccuracy exists even for providers who reportedly provide comprehensive obesity care. Furthermore, we demonstrate that providers are more likely to underestimate weight than to overestimate it. Our results further support the recommendation that BMI should be calculated for every patient to provide appropriate care. 3,24
Although providers demonstrated high levels of knowledge with regard to BMI, their reported management of overweight/obesity was inconsistent. The results demonstrate that the majority of physicians reported consistently calculating BMI as well as screening patients with overweight and obesity for diabetes and metabolic syndrome. Yet, these same providers often failed to document an elevated BMI in the medical record or discuss abnormal BMI with their patients. These results support our main hypothesis that physicians have difficulty diagnosing overweight and obesity by visual inspection alone, and therefore, may not be prompted to address it during a patient care visit. Physicians were only 51.9% accurate at assessing BMI visually, and this correlated with a BMI documentation rate of 40.2% and a weight loss counseling rate of only 59.1%. These findings suggest potential targets for physician education and electronic medical record modifications to achieve more consistent identification and counseling of overweight or obese patients.
Interestingly, despite the reported discrepancy in diagnosis and management/treatment of obesity, only the minority of physicians agreed or strongly agreed to experiencing barriers to the provision of care. 25 Time was the most commonly cited barrier and only 35% of physicians agreed or strongly agreed that time was a barrier. When queried, few physicians explicitly cited system-related barriers; however, many expressed a lack of comfort with treatment options. Although many physicians were comfortable with diet and exercise counseling, fewer were comfortable with prescribing weight loss medications or discussing surgical interventions for eligible patients. Even physicians who reported providing comprehensive obesity care noted a lack of comfort with prescribing weight loss medication and surgical treatment and were not familiar with the contraindications or side effects of weight loss medications. These items may be reasonable targets for physician education programs to help facilitate consistent and complete provision of care to the obese patient population.
Our study has several limitations. Our study involved physicians predominantly from a health-system and a university-based practice. Few community health centers were included and the providers themselves did not represent the diversity of the field (5% of physicians surveyed were African American). These study characteristics limit the generalizability or external validity of our study. Although the findings may be applicable to other health systems or university-based programs with similar ethnic and racial makeup, they may not be generalizable to providers in other health settings. Similarly, they are not generalizable to all providers in that the survey did not include advanced practice registered nurses and physician assistants who may have a large role in the provision of obesity care.
With regard to internal validity, we created new brief measurement scales for “Barriers to Obesity Care,” “Comfort Providing Obesity Care,” and “Comprehensive Obesity Care,” and future confirmation of their psychometric properties is warranted. That said, we adhered to a systematic and rigorous development and testing methodology of each scale’s internal consistency reliability (of which all scales were quite high), examination of item-total correlations for each scale (of which the vast majority of items met standard inclusion thresholds), and calculation of correlation coefficients, all of which were in expected directions and magnitudes between scales.
Finally, our study is also limited in that it does not specify or assess the full spectrum of treatment options for the care of overweight and obese women (eg, referrals to registered dieticians, exercise physiologists, trainers, psychologists, social workers, weight loss support, and other options that may be available within a university setting or an integrated health system). Although the focus of the present study was to assess provider practices and barriers to the provision of care, subsequent studies would be enriched through an assessment of availability and utilization of resources. The utilization of these potential resources has educational implications for providers, especially considering that time was found to be a common barrier among physicians. Others within the system or community can be recommended as options for additional support with the goal of providing comprehensive multidisciplinary care for these patients. Future research into collaborating with other providers outside of the medical field could reduce the stress placed on physicians to do it all alone. Resources should also be allocated in reinforcing and recommending the training of other health professionals with experience in root causes of overweight or obesity in patients. Thus, further studies are needed that investigate how collaboration with providers outside the medical field could reduce the burden on physicians in treating patients with overweight or obesity. In addition, further research is needed to determine whether these results are valid for other physicians in various practice settings and to mid-level providers.
Our study has several strengths. We had a large sample size with a high response rate. To our knowledge, it is the largest study designed to assess a provider’s ability to evaluate BMI based on visualization alone. Another strength of the study is that the physicians varied in practice setting and primary care medical specialties, which adds to the generalizability of the study findings.
“So What?”
Our findings suggest that primary care physicians do not consistently identify overweight or obese patients, document an elevated BMI, or discuss an elevated BMI with their patients.
Our study also demonstrates that the majority of providers report discomfort with the full range of treatment options for obesity including medical and surgical management.
Additional studies will be needed to determine whether findings are similar for other groups of physicians and mid-level providers who may practice in different setting. However, in our large sample of obstetricians and internal medicine and family medicine physicians, our findings suggest potential targets to improve primary care physician practice patterns.
One possible intervention might be to create a licensing or an accreditation based on these obesity education courses that physicians would be encouraged to acquire when they work in specialties that provide obesity counseling most often.
Additionally, ensuring physicians utilize and are educated regarding existing resources aimed to help in treatment of overweight or obesity would be a key element to propel the care of these patients forward.
Our findings also demonstrate that decisions about obesity care need to be made with the input of both the provider and the patient. Through this patient-centered care, advances in obesity care can be made in the coming years.
Footnotes
Acknowledgments
We would like to acknowledge Pawel Truszkowski who assisted with the development of the survey.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Institute of Diabetes and Digestive and Kidney Diseases [grant numbers 5T35DK062719-28] and the University of Chicago.