Assessment and Management of Adult Obesity in a Primary Care Practice

Authors

  • Marietta Orlowski,

  • Sherry Adkins,

  • Sylvia Ellison,

  • Audrey Choh,

  • Nancy Terwoord,

  • Richard Schuster


Abstract

The purpose of this project was to describe primary care physician adherence to National Heart, Lung, and Blood Institute (NHLBI) Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults, using a measurement tool developed purposefully for the project, and to explore patient characteristics associated with physician assessment and management behaviors. This study sought to add to the body of existing knowledge by analyzing physician clinical behaviors in a more descriptive and systematic manner. The researchers hypothesized that measuring physician behavior for assessing and managing overweight and obesity as a series of steps, as outlined in the NHLBI guidelines, would identify specific gaps in physician action and possibly identify other relationships between patient characteristics and physician behavior. A chart abstraction of 99 randomly selected adult patients with at least one visit to a primary care practice during a 12-month period was completed. Three out of four health center patients were clinically overweight or obese, yet despite high rates of weight status measurement, only 25 percent of overweight and obese patients received such a diagnosis. High levels of weight measurement, including BMI calculations, did not correspond to a weight diagnosis. The majority of patients did not receive any dietary (72 percent) or physical activity (69 percent) management. When dietary management was introduced, patients received either information (68 percent) or a goal (32 percent), and none received a goal with an accompanying plan. In cases where physical activity management was introduced, patients received a goal (52 percent) or information (39 percent). Patients of higher BMIs were more likely to be diagnosed and managed with regard to their weight. Improvements in future weight-related counseling may be found in encouraging physicians' willingness to make weight and weight-related diagnoses. Results from this analysis of assessment highlight that measuring weight status (via BMI) and making a weight-related diagnosis are not the same practice.

Introduction

Overweight and obesity are exacting an enormous toll on the American population. Currently, two thirds of the population is overweight and or obese and excess weight is a contributing factor for four of the leading causes of death: heart disease, malignancies, cerebrovascular disease, and diabetes mellitus (National Center for Health Statistics, 2011). Yet, the Centers for Disease Control and Prevention consider obesity a “winnable battle” in public health whose solution lies in comprehensive and evidence-based approaches (Centers for Disease Control and Prevention, 2012b).

Health care providers are a valuable component of a comprehensive solution to obesity. Physicians and related clinicians are viewed as health experts and trusted sources of health information (U.S. Department of Health and Human Services, 2010). Furthermore, primary care providers have established relationships with patients and are able to observe changes in health and identify opportunities for health behavior intervention. This trust and contact frequency allows health care providers to tailor messages and interventions to individual patient knowledge, attitudes, and readiness. Physician-delivered health education has been associated with greater attempts for patient smoking cessation, dietary changes, alcohol reduction, and physical activity initiation (Balasubramanian et al., 2008; Fleming et al., 2002; Jepson, Harris, Platt, & Tannahill, 2010; Kreuter, Chheda, & Bull, 2000; Loureiro & Nayga, 2006). Population-level changes in obesity levels can benefit from the participation of health care providers.

The purpose of this project was to describe primary care physician adherence to National Heart, Lung, and Blood Institute (NHLBI) Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults (1998), using a measurement tool developed purposefully for the project, and to explore patient characteristics associated with physician assessment and management behaviors. The NHLBI guidelines provide an evidence-based approach for examining and treating patients with regards to overweight (Body Mass Index [BMI] 25–29.9 kg/m2) and obesity (BMI > 30 kg/m2) (National Heart, Lung and Blood Institute & National Institute of Diabetes and Digestive and Kidney Diseases, 1998). The guidelines reflect a two-step process in physician behavior: assessment and management. Assessment for overweight and obesity involves determination of degree of obesity, determination of patient's absolute risk status, assigning weight-related diagnosis, and determination of patient readiness. Management of overweight and obesity involves dietary and physical activity goal development, dietary and physical activity behavioral therapy, and arranged patient follow up. Lastly, weight loss drugs are recommended for consideration as part of a comprehensive weight management program in patients with BMI >30 or >27 kg/m2 in the presence of concomitant risk factors (family history, smoking, and age) or diseases like type 2 diabetes and hypertension. Weight loss surgery is also a treatment option for patients with clinically severe obesity (BMI ≥40 or 35 kg/m2 with associated comorbidities) after failure of less invasive methods and presence of high risk for morbidity and mortality.

In studies of physician behavior regarding select actions within the guidelines, assessment, and management behaviors were low. Approximately half of primary care patients were screened for obesity or overweight (Ma, Xiao, & Stafford, 2009; Melamed, Nakar, & Vinker, 2009; Rose, Turchin, Grant, & Meigs, 2009; Smith et al., 2011) and less than a third of overweight and obese patients were diagnosed as such (Bardia, Holtan, Slezak, & Thompson, 2007; Davis, Emerenini, & Wylie-Rosett, 2006; Ma et al., 2009; Melamed et al., 2009; Ruser et al., 2005). BMI chart documentation increased with higher patient age, obese status, presence of obesity-related comorbidities, and use of chronic medication (Melamed et al., 2009).

The assessment and management of patient weight involves a series of physician actions. The NHLBI guidelines outline these actions. Yet, previous studies have reported assessment and management as binary variables (Bardia et al., 2007; Bleich, Pickett-Blakely, & Cooper, 2011; Boardley, Sherman, Ambrosetti, & Lewis, 2007; Rose et al., 2009; Ruser et al., 2005; Schuster, Tasosa, & Terwoord, 2008; Sciamanna, Tate, Lang, & Wing, 2000; Simkin-Silverman et al., 2005; Waring, Roberts, Parker, & Eaton, 2009). Despite the understanding of the progressive nature of health behavior development and cessation, methodological barriers can inhibit such progressive measurement. For example, data extracted from large national data sets may not have adequate measurement to create a composite variable. In other instances, researchers may not have been able to collect variability in physician behavior due to underdeveloped instrumentation and/or lack of physician documentation. This study sought to add to the body of existing knowledge by analyzing physician clinical behaviors in a more descriptive and systematic manner. Researchers hypothesized that measuring physician behavior for assessing and managing overweight and obesity as a series of steps, as outlined in the NHLBI guidelines, would identify specific gaps in physician action and possibly identify other relationships between patient characteristics and physician behavior. Findings could be used to clarify next steps in supporting physician behavior, and in designing policy supports for the assessment and management of weight status in a primary care setting.

Methods

Setting and Sample

This study was a retrospective chart review to analyze physician adherence to adult obesity guidelines in a family practice setting. A large family health center, located on the campus of an urban hospital, was selected as the site for this study. At this center, staff members (including Family Medicine physicians and residents, registered nurses, a clinical psychologist, and a pharmacist educator) offer primary care services including pediatrics, adolescent and geriatric care, psychology, women's health, sports medicine, and minor office procedures. The center staffs approximately 30 residents and attending physicians, who see over 20,000 visits per year with a mix of 70 percent Caucasian and 30 percent African American patients. The family health center is located in the limits of a mid-sized metropolitan city in Ohio. Ohio ranks 13th in adult obesity prevalence. Data from the Behavioral Risk Factor Surveillance System indicates that one third of Ohio residents (29.7 percent) and adults residing in the metropolitan study area (29.7 percent) are obese (Centers for Disease Control and Prevention, 2012a).

Charts were randomly selected for review from the active patient database. An electronic list of patient visits over 12 months prior to list development was obtained from hospital information technology staff. The list was filtered to include only unique medical record number (MRN) identifiers, and random numbers were assigned to each MRN. The list was sorted by random number, and charts were reviewed in ascending order. Eligible charts included those of patients who had at least one visit in the previous 12 months, were 18 years of age or older, were not pregnant during the year of review, and had a body mass index of 25 or greater.

For regression analysis planning, an a priori minimum sample size of 80 was calculated, taking into account generalizability and power (Draper & Smith, 1981; Tabachnick & Fidell, 2001). Additionally, in order to undertake factor analysis, a ratio of at least 10 subjects for each variable has been described as suitable for sample generalization (Pett, Lackey, & Sullivan, 2003). The Physician Obesity Guideline Behavior Scale consists of a total of eight measures. Therefore, a minimum of 80 subjects is appropriate for this study.

Instrument Development

The Physician Obesity Guideline Behavior Scale, a new data collection tool, was developed to assess physician adherence to the NHLBI guidelines. A multidisciplinary team developed the Physician Obesity Guideline Behavior Scale. The team consisted of a primary care physician, a dual medical and public health student, a primary care nurse manager, a health behavior researcher, an epidemiologist, and a statistical consultant. A key objective of the instrument development process was to be able to capture variability in physician behavior. The instrument consisted of two subscales, assessment and management. There were four items per subscale, each one representing a recommended action in the guidelines. The individual items were coded from 0 to 3, with 3 representing the recommended action within the NHLBI guidelines. Conceptually, scores of 1 and 2 represented intermediate, progressive steps toward the recommended guideline behavior. A score of zero indicated that nothing was recorded in the chart. A score of 1 indicated that something related to that item was recorded, and a score of 2 indicated an action close to the recommended guideline behavior was recorded. A pilot chart review was completed in May 2009. One research assistant, a medical and public health student, abstracted all information from the medical record (n = 25). Following research assistant feedback and team discussion, the physician behavior scale was adjusted. All instrument changes involved clarifying the terms used to describe the item scores. Table 1 includes the instrument used for data collection.

Table 1. Physician Obesity Guideline Behavior Scale Used for Data Collection
Assessment Minimal ActionModerate ActionAt or Near Guidelines
Measure0123
 No height or weightHeight or weight (any time)BMI/waist (any time)BMI/waist (last 12 months)
 OR   
 No waist circumference  Waist circumference
Comorbidities Risk Status0123
  Some comorbidities on problem listComorbidities with reference to O/OO/O classification and associated risk
Diagnosis0123
  In visit notesOn problem listDiagnosis with mention of class/severity
Patient Readiness0123
  Discussion of barriers and prior attempts notedBarriers and prior experiences listedLevel (stage) of readiness assessed
Management Minimal ActionModerate ActionAt or Near Guidelines
Dietary goal0123
  Patient given dietary informationa) Goal, no action orMeasurable, attainable goal with action plan
   b) partial goal, or 
   c) just action plan, includes referral 
Physical activity0123
  Patient given activity informationa) Goal, no action orMeasurable, attainable goal with action plan
   b) Partial goal, or 
   c) Just action plan, includes referral 
Patient record keeping0123
  Patient given record/tracking formPatient log asked about goalsPatient log (goals) reviewed
Follow-up and monitoring0123
  Patient advised to follow-upPatient follow-up appointment on weight loss goal at least oncePatient follow-up appointment on weight loss, monthly

Data Collection

All study variables were abstracted from the patient medical record. Patient characteristic variables included age, sex, race, and health insurance type. Assessment measures were coded from documentation of height, weight, body mass index, waist circumference, related comorbidities, weight-related diagnosis, and patient readiness. Height was not recorded in 16 charts. If body mass index was not available by chart, female patients with weights of >145 pounds and male patients >169 pounds were deemed eligible, based on average height data for the American population (female 64 in., male 69 in.).

Management measures were coded from documentation on dietary and physical activity goal setting, referral, patient record keeping, follow-up visits, and use of weight loss medications or referral for surgery. This abstraction process was carried out between May and July 2009. Charts were reviewed at a rate of 5–25 minutes per chart, with variability based on number of office visits per patient in the 12 months prior to review. The university's Institutional Review Board for Human Subjects approved this study.

Data Analysis

Descriptive statistics were calculated for all study variables. After descriptive analysis, health insurance type was recoded to a binary variable of Medicaid: Yes/No. A principal component analysis (PCA) was performed using SAS version 9.2 (SAS Institute, Inc., Cary, NC) to characterize the subscales obtained from Physician Obesity Guideline Behavior Scale instrument. For each individual, the first principal component (PC1) score is calculated as a linear combination given by:

display math

where X1 through Xp are the instrument subscales, and a11 through a1p are the loadings or eigenvectors, which represent the degree of the correlation between the instrument variables and PC1 (Manly, 1994). The magnitude and direction of the loadings are used to interpret the orthogonal principal components.

Multiple linear regression models were tested separately for the assessment and management behavior scales using the backward procedure in IBM SPSS (Chicago, IL) v. 19. Because assessment and management are different physician behaviors, and outlined in the guidelines as two separate composite steps, separate regressions for assessment and management were completed. Variables were removed from the models based on significance level (p > 0.10) and overall variability explained. Covariates approaching statistical significance (0.05 < p < 0.10) in simple or multiple regression models were retained in the final models. Multicollinearity among predictor variables was assessed prior to conducting linear regression.

Results

Characteristics of the Sample

Of 138 charts opened for review, 99 (71.7 percent) were eligible for inclusion. Thirty-five patients were excluded because they were not overweight or obese. Three patients were excluded due to pregnancy, and one who was seen only for mental health counseling. The mean age of patients was 50.2 with a standard deviation (SD) of 16.1 years, and 31.3 percent patients were male. Similar to the health center's full patient mix, 29 percent of sample patients were black. Half of patients (51.5 percent) had commercial insurance, while 19.2 percent had Medicare and 21.2 percent Medicaid. Eight patients had no current form of health insurance.

The mean BMI was 32.9 (SD 6.7) kg/m2. A third of patients (31.3 percent) were overweight (BMI between 25.0 and 29.9), and 62.7 percent were obese (BMI ≥ 30.0). Sixteen patients did not have BMI values available but were eligible for study inclusion based on weight and consideration of average height for U.S. males and females. Frequent comorbid conditions included hypertension (48.5 percent), hyperlipidemia (41.4 percent), and diabetes (18.2 percent). Forty-five percent of patients had a family history of premature heart disease, and 21.2 percent were smokers. Nearly 20 percent of patients had an “other” related comorbidity, including primarily osteoarthritis, but also gynecological abnormalities, gallstones, or urinary stress incontinence. Based on risk factor data collected in this study including age, sex, family history, smoking status, presence of key comorbidities, and serum glucose and cholesterol, almost half of patients (45.5 percent) were at increased risk for weight-related mortality (see Table 2).

Table 2. Sample Patient Characteristics
 Percentage or M (SD)n
  1. aPersonal history of myocardial infarction (MI), angina pectoris, or coronary artery procedure.
  2. bMI or sudden death by first-degree relative (before age 55 for male relatives or age 65 for female relatives).
  3. cGynecological abnormalities, osteoarthritis, gallstones, and stress incontinence.
  4. dCalculated based on NHBLI guidelines.
Sex
Female68.768
Male31.331
Age (years)
18–244.04
25–3416.216
35–4924.224
50–6438.438
65–8416.216
85+1.01
Race
White68.768
Black29.329
American Indian1.01
Other1.01
Insurance
Medicare19.219
Medicaid21.221
Private51.551
Pending HCAP (no current insurance)8.18
Comorbidities
Hypertension48.548
Diabetes18.218
Hyperlipidemia41.441
Coronary heart diseasea16.216
Smoker21.221
Family historyb45.545
Sleep apnea8.18
Other relatedc19.219
Disease risk leveld
Very high29.329
High16.216
Less than high54.554
BMI (kg/m2)
≥30.0 (obese)62.752
25.0–29.9 (overweight)31.331

Qualitative Nature of Chart Documentation

Chart documentation regarding overweight and obesity included measurements, diagnoses, and discussion of management. When present, measurements and diagnoses were both prominently displayed in the patient chart. Diagnoses were entered primarily by physicians and displayed in the chart on the problem list. Documentation regarding management was nonspecific, brief, and located in individual office visit notes by physicians. Assessment of patient readiness was documented in description of (1) patient habits, (2) weight loss barriers or triggers, (3) patient motivation, and (4) past or present weight loss attempts. Documentation regarding dietary goals was variable, and included mention of general advice (most common), written handouts or professional referral (occasional), and specific dietary advice (rare). Documentation concerning physical activity goals reflected general guidance, specific recommendations, and professional referral. Physical activity recommendations tended to be more specific than those related to diet. Documentation regarding patient follow-up was nonspecific.

Assessment and Management Subscales

After data collection, a principal component analysis (PCA) was also completed and both the assessment and management final subscales were modified slightly. There was no documentation of any patient record keeping and thus, the patient record keeping item was excluded from further analysis. For statistical and clinical reasoning, patient readiness was included in the final management subscale. The patient readiness item loaded on both PCA1 and PCA2 subscales, but was higher on the management scale (PCA2). Furthermore, in the guidelines, patient readiness was described as the initial conversation about motivations, perceived barriers, and previous experiences with dietary and physical activity changes and thus, conceptually fit on the management subscale (National Heart, Lung and Blood Institute & National Institute of Diabetes and Digestive and Kidney Diseases, 1998).

Four items had moderately high loading on PC1: readiness (0.374), diet (0.442), physical activity (0.553), and follow-up (0.552). PC2 included two measures with high loading values: measure (0.700) and diagnosis (−0.556). The readiness item also loaded on PC2 with a moderate value (0.340), but the loading was higher on PC1 (Draper & Smith, 1981; Tabachnick & Fidell, 2001). A third single item component emerged from the analysis—comorbidities. Comorbidities had a loading factor of 0.953, but descriptively the data lacked variability. Of the total charts, 83 patient records had a comorbidity score: 77 of those were scored as a 1, the remaining scores were a zero. After detailed analysis of the descriptive notes, the researchers hypothesized that the instrument collected a patient characteristic of comorbidities, and did not capture a physician assessment behavior. Thus, comorbidities were excluded from the final analysis.

Rates of Assessment and Management

Based upon the PCA, assessment was re-categorized to two items: measure plus diagnosis. Scores could range from 0 to 6. The following categories were created to give meaning to the values: no assessment (score of 0); minimal assessment (score 1–2), moderate assessment (score 3–4), and at or near guideline assessment (score 5–6). Management was re-categorized to include four items: patient readiness, diet, physical activity, and follow-up. Scores could range from 0 to 12 and the following categories were created to give meaning to the values: no management (score of 0), minimal management (score 1–4), moderate assessment (score 5–8), and at or near guideline assessment (score 9–12). Most patients received moderate weight-related assessment but minimal or no management. All patients received some level of assessment, with most (60 percent) receiving scores of 3–4 out of 6. In contrast, half of patients received no weight-related management at all. Eighteen percent of patients received moderate management, and only one received management that was at or near guideline standards (see Table 3).

Table 3. Distribution of Physician Obesity Behavior Scale Variable Values
VariableScore
NoneMinimalModerateAt or Near GuidelineM (SD)
  1. aSingle scale items have a range of 0–3. Assess, the sum of measure and diagnosis, has a range of 0–6.
  2. bManage, the sum of readiness, diet, physical activity, and follow-up, has a range of 0–12.
Assess (0–6)a02960102.9 (1.1)
Measure01822592.4 (0.8)
Diagnosis7532100.5 (0.8)
Manage (0–12)b49311812.0 (2.5)
Readiness63221400.5 (0.7)
Diet7119900.4 (0.6)
Physical activity68121630.5 (0.9)
Follow-up/monitoring6914880.5 (1.0)

Assessment was moderate and highly variable. Despite a high percentage of patients being measured at or near guideline standards, most patients were not diagnosed. The Physician Obesity Guideline Behavior Scale item measure had a mean score of 2.4 out of 3, indicating measurement behaviors at or near guideline standards. All patients had a weight in the chart, and most (84 percent) had a height. Of 99 eligible patients, 82 percent had documented BMIs, with 60 percent measured in the last 12 months. There were no waist circumference measurements recorded during this study. Relevant comorbidities were located in the chart problem list for 77 out of 83 patients with applicable comorbidities (93 percent). Comorbidities, however, were never described with reference to overweight, obesity, or a formal classification of weight-related risk. The diagnosis item had a mean score of 0.5 out of 3 suggesting poor diagnosis of overweight and obese patients. Less than 25 percent of all patients received a weight-related diagnosis; 38 percent of obese patients and less than 3 percent of overweight patients received a weight-related diagnosis. Most diagnoses were displayed on the patient's problem list.

In contrast to assessment, weight management was minimal. Assessment of patient readiness occurred for 36 percent patients and consisted mostly of discussion regarding barriers and prior weight loss attempts. No patients were assessed using a formal readiness scale. Only 28 percent of patients were counseled regarding diet and 31 percent regarding physical activity. Thirty-six percent of patients received some counseling regarding diet or physical activity, with 26 percent receiving both types of counseling activities. Two patients were counseled regarding diet alone, and eight patients regarding physical activity alone. Of those who received dietary advice, patients received either information (68 percent) or a goal (32 percent), but none received a measurable, attainable goal with an accompanying action plan. In cases where physical activity management was introduced, 39 percent of patients received information, while more patients received an activity goal (61 percent). Three patients received measurable, attainable goals with associated action plans. Patient record keeping was not documented in any cases. Finally, 30 percent of patients were at least advised to follow up concerning diet, physical activity, or weight, with eight patients receiving counseling in two office visits and another eight in three or more office visits (Figure 1).

Figure 1.

Level and Type of Management Received by Percent of Patients.

Relationship of Patient Characteristics to Physician Behavior

Assessment varied significantly by patient BMI, controlling for patient race and Medicaid insurance (p < 0.01). A 1 unit (kg/m2) increase in BMI corresponded to 0.056 unit increase in assessment score. Thus, an obese patient with a BMI of 30 was one third of a point (0.28) closer to assessment near guidelines (0–6 scale) than an overweight patient with BMI of 25 and similar race and health insurance. Patient insurance, race, and BMI accounted for 19.6 percent of the variability in physician assessment scores (Table 4). Age and sex were not individual significant predictors of assessment or management, and were already incorporated in determination of patient risk.

Table 4. Regression Statistics for Assessment Model
VariableUnstandardized CoefficientsStandardized CoefficientstSig.
BSEBeta
BMI0.0560.0140.4053.942<0.0001
African American0.3830.2160.1911.7720.080
Medicaid−0.3920.254−0.169−1.5460.126
Model R2 = 0.196

Management varied significantly by patient BMI, controlling for Medicaid insurance and obesity-related risk (p < 0.01). A 1 unit (kg/m2) increase in BMI corresponded to 0.127 unit increase in management score. Thus, an obese patient with a BMI of 30 was more than half a point (0.64) closer to being managed near guidelines (0–12 scale) than an overweight patient with BMI of 25 and similar health insurance and risk. Patient insurance, risk level, and BMI accounted for 16.8 percent of the variability in management scores (Table 5). Based upon previous research, the relationship of diagnosis to obesity management was explored. Diagnosis was not significantly related to management.

Table 5. Regression Statistics for Management Model
VariableUnstandardized CoefficientsStandardized CoefficientstSig.
BSEBeta
BMI0.1270.0400.3343.1880.002
Risk0.9280.5210.1841.7840.078
Medicaid−1.3020.665−0.205−1.9600.054
Model R2 = 0.168

Discussion and Conclusion

Assessment of weight status is a progressive act. It begins with documentation of a patient's height and weight, followed by the calculation of a BMI, and concludes with the making of weight and weight-related diagnoses. In this study, all overweight and obese patients had weight measurements and 82 percent of overweight and obese patients had documented BMIs. Previous researchers found that screening and BMI calculations occur in approximately half of adult primary care visits (Boardley et al., 2007; Ma et al., 2009; Melamed et al., 2009; Rose et al., 2009; Smith et al., 2011; Waring et al., 2009). This practice did use electronic medical records (EMR) and the EMR system may have contributed to BMI calculation rates higher than previous reports. The EMR prompts for a height and weight as well as calculates a BMI. In contrast, no patients had waist circumference measurements, highlighting the absence of this practice in routine patient care and eliminating the opportunity for further risk assessment and management related to waist.

Diagnosis is an important milestone in the assessment and subsequent management of obesity. Previous researchers documented higher rates of management when identification or diagnosis occurs (Bardia et al., 2007; Bleich et al., 2011; Boardley et al., 2007; Ruser et al., 2005; Waring et al., 2009). Bleich et al. (2011) found that if a patient was diagnosed as obese, he or she was more than twice as likely to receive diet or exercise counseling. Similarly, Bardia et al. (2007) found that patients with a diagnosis of obesity were 2.5 times more likely to have an obesity management plan. The high rates of weight status measurement in this study, however, did not transfer to high rates of diagnosis. As in previous research, only 25 percent of overweight and obese patients in this study were diagnosed as such (Bardia et al., 2007; Bleich et al., 2011; Ma et al., 2009; Melamed et al., 2009). In this study, diagnosis was measured on a progressive four-point scale, yet most overweight and obese patients (75 percent) had no reference to weight on the problem list or in the office notes. The low rate of diagnosis, given the progressive manner in which diagnosis was scored, again was surprising. Physicians may be unwilling to record an actual diagnosis, yet one would expect most obese or overweight patients to have weight appear in office visit notes or on the problem list. Diagnosis occurring rarely across studies, but with a strong relationship to management, appears as an important target behavior to shape with physicians. Improving rates of diagnosis may in turn improve rates of management.

Barriers to increasing rates of diagnosis of overweight and obesity include physician beliefs regarding the nature of overweight and obesity, as well as the expectation for patient success. Weight management is often framed as a method for managing risk related to other comorbidities. Descriptive detailed results from this study confirm the medicalizing of obesity. In this study, documentation of weight management was often located in the assessment/plan portion of the office visit note under the heading of a comorbidity, most commonly dyslipidemia, hypertension, and diabetes mellitus type 2. This suggests that physicians may conceptualize overweight and obesity as factors impacting risk related to other diseases, rather than as primary disorders in and of themselves. Others have demonstrated that the physician practice of medicalizing weight disorders is one way in which doctors enter into weight-related management, suggesting that a potential strategy for improving management behaviors among physicians is to alter beliefs regarding the medical nature of overweight and obesity (Forman-Hoffman, Little, & Wahls, 2006). Also, physicians report being prepared to treat obesity (Hayden, Dixon, Piterman, & O'Brien, 2008), yet uncomfortable, and having low expectations for weight loss success (Schuster et al., 2008). Thus, poor adherence to guidelines may reflect a physician's own conceptualization of weight as a health issue with efficacious treatment options.

Physicians demonstrated poor delivery of specific, behavioral advice and referral to other professionals. Rates for dietary and physical activity counseling were similar to previous published values, leaving two thirds of overweight and obese patients without any level of management (Anderson, Konz, Frederich, & Wood, 2001; Bleich et al., 2011; Ma et al., 2009; Simkin-Silverman et al., 2005; Smith et al., 2011). Physicians did not use patient record keeping as a weight management tool, despite its use for managing other disorders like hypertension and diabetes. As observed previously, physicians also rarely referred patients or recommended pharmacologic or surgical treatments (Ferrante, Piasecki, Ohman-Strickland, & Crabtree, 2009; Hayden et al., 2008; Phelan, Nallari, Darroch, & Wing, 2009; Shiffman et al., 2009). It is possible that these shortcomings reflect a lack of knowledge and skills regarding weight management as well as beliefs regarding management efficacy. Plourde and Prud'homme recently suggested that the adoption of a stage matched intervention, the 5 As (assess, advise, agree, assist, and arrange follow-up), might assist physicians in counseling patients about weight (Plourde & Prud'homme, 2012). The 5 As have been used successfully with smoking cessation in primary care and are a recommended tobacco management action within the U.S. Preventive Services Task Force Guidelines.

Patients with higher BMIs are more likely to be assessed as well as receive weight management advice and resources. The assessment and management care of an obese patient (BMI = 30) was one third to half a point closer to guidelines than an overweight patient (BMI = 25) of similar race, health insurance, and risk. Others have documented a similar, and stronger relationship of BMI to physician behavior (Bleich et al., 2011; Boardley et al., 2007; Ma et al., 2009; Melamed et al., 2009; Sciamanna et al., 2000; Waring et al., 2009) thus implying an important category of overweight patients are not receiving appropriate care. Patient characteristics, however, accounted for low variation in physician assessment (19.6 percent) and management (16.8 percent), implying other factors mediating adherence to treatment guidelines.

Policy and Practice Implications

Obesity is a medical diagnosis. Improvements in future weight-related counseling may be found in encouraging physicians' willingness to make weight and weight-related diagnoses. Results from this analysis of assessment highlight that measuring weight status (via BMI) and making a weight-related diagnosis are not the same practice. Future clinical guidelines or policies should encourage and prompt the last step of assessment—that of diagnosis. A simplified message as to the importance of diagnosing overweight and obesity can be delivered to physicians, while system level factors like EMR reminders can be enacted. Technology, via electronic medical records, is being used successfully to prompt the assessment of weight and BMI, and rates of BMI assessments have improved. When a diagnosis of overweight or obesity is documented, patients are more likely to receive counseling and/or referral.

A second opportunity to improve obesity care is in the assessment and counseling of overweight, but not yet obese, patients. Physicians should be encouraged to complete both assessment and management in patients of lower, but overweight, BMIs. Currently, when physicians do take action, it is with patients of higher weight and comorbid conditions. As noted above, it is often through conversation about the comorbid condition that strategies about weight management are discussed. Much like discussing elevated blood pressure or impaired fasting glucose, drawing attention to smaller patient weight gains may encourage more timely patient action.

Recent statements by the U.S. Preventive Services Task Force (USPSTF) do not recommended diet and physical activity behavioral counseling into primary care for adults without a chronic condition, namely diabetes, high blood pressure, elevated cholesterol levels, and heart disease. USPSTF does, however, recommend such counseling for adults with these medical conditions (Moyer, 2012). The rationale to the recommendation is that there is insufficient evidence of benefit in a general “healthy” population. We hope that researchers continue to explore this relationship, and explore the efficacy of counseling in primary care with overweight and obesity as the primary diagnosis.

Limitations

This study measured physician adherence to NHLBI guidelines, using an instrument developed purposefully to measure the two-step process described in the guidelines. The study has a small sample size (99) drawn from one clinical practice, limiting statistical manipulations and our ability to generalize. In addition, many items on the Physician Obesity Guideline Behavior Scale exhibited limited variability, particularly patient record keeping and physician documentation of comorbidities. When combined with low variability, the small sample size resulted in a reduced ability to reliably detect moderate effects. Low variability also led to a redefining of assessment and management. In this study, we were unable to determine significant relationships between patient characteristics other than BMI and the assessment and management behaviors; however, the relationship of patient age and sex found by previous researchers should not be dismissed. Finally, the method of chart abstraction to measure physician behavior is inherently limited. Direct observation provides a more accurate picture of clinical behavior and may have given insight into physician reasoning for diagnosis and counseling behaviors.

Conclusion

Evidence-based guidelines for assessing and managing adult overweight and obesity are available, but physician adherence is poor. Through a progressive measurement tool, we identified infrequent physician diagnosis of overweight and obesity as a gap in the process of managing weight. Three out of four health center patients were clinically overweight or obese, yet despite high rates of weight status measurement, only one out of four patients received such a diagnosis. High levels of weight measurement, including BMI calculations, did not correspond to a weight diagnosis. A means to improving the act of diet and physical activity counseling may be through first encouraging patient diagnosis.

Biographies

  • Marietta Orlowski, PhD, MCHES, is an associate professor in the Department of Community Health at Wright State University and serves as the director of the heath promotion and education concentration in the Master of Public Health program.

  • Sherry Adkins, MD, MPH, is a Family Medicine resident at Clinton Memorial Hospital in Wilmington, Ohio.

  • Sylvia Ann Ellison, MA, MPH, is a Research Instructor in the Department of Community Health at Wright State University.

  • Audrey C. Choh, PhD, is a research assistant professor in the Division of Epidemiology, Lifespan Health Research Center, within the Department of Community Health at Wright State University.

  • Nancy Terwoord, RN, BS, CPHQ, is the Director of Clinical Quality for the Ohio Association of Community Health Centers. She is certified in professional health care quality and is a licensed registered nurse.

  • Richard J. Schuster, MD, MMM, FACP, is the Director of the Center for Global Health at the University of Georgia and a Professor of Health Policy and Management in the College of Public Health. Dr. Schuster also serves as a Visiting Professor at the University of Haifa School of Public Health.

Ancillary