Kathryn I. Pollak, PhD

The US Preventive Services Task Force (USPSTF) recommends that physicians screen patients for obesity and offer intensive counseling and behavioral interventions to promote sustained weight loss.¹ Evidence suggests that physician counseling, including advice, can help patients to lose weight, increase physical activity, and improve diet.^2-9 However, little is known about what specific types of weight loss advice physicians give to patients, and whether some types are more effective than others at influencing behavior change.

We analyzed physician weight loss advice delivered in primary care visits and measured changes in patients’ dietary intake, physical activity, and body weight. We examined both the type of weight loss advice delivered and the impact of type of advice on weight and behavior change.

Methods

This study analyzed audio recordings from Project CHAT – Communicating Health: Analyzing Talk. The project was approved by the Duke University Medical Center Institutional Review Board.

Recruitment Physicians. We obtained consent from 40 primary care physicians in community-based practices and told them the study would examine communication around preventive health topics, not weight specifically.

Patients. We identified potential participants by reviewing scheduled appointments 3 weeks in advance. Eligible participants were at least 18 years of age, English-speaking, overweight or obese (body mass index [BMI] ≥25 kg/m²), cognitively competent, and not pregnant. After we obtained consent, a remotely located research assistant started a digital audio recorder as the patient entered the exam room. Immediately after the encounter, the research assistant administered a post-encounter survey to the patient and recorded the patient’s vital signs (N=461). Three months later, the research assistant met with the participant to record vital signs and administer a survey assessing changes in dietary fat intake and exercise (N=426).

Data coding
We coded advice into 4 broad categories: (1) nutrition advice, (2) physical activity advice, (3) specific weight loss advice, and (4) nonspecific weight loss/weight-related advice. We transcribed each piece of advice verbatim.

Nutrition advice consisted of 9 sub-categories: calorie/portion control, meal timing/planning, commercial diet plans, negative diet plans, increase fruits/vegetables, reduce sugar/carbohydrates, reduce fat/cholesterol, other micronutrient recommendations, and specific food items from multiple categories.

Physical activity advice consisted of 6 subcategories: walking, aerobic exercise, anaerobic exercise, exercise intensity, exercise duration, and exercise for comorbid conditions.

Specific weight loss advice consisted of 3 categories: weight loss behavior, weight loss for comorbid conditions, and referrals.

Nonspecific weight loss advice also consisted of 3 subcategories in which physicians provided no details about the general topics of nutrition, physical activity, or weight loss.

Two independent coders (CBT and MEC) assessed each piece of advice and double coded 20% of conversations for reliability. Cohen’s kappa was used to calculate inter-rater reliability for each code using Landis and Koch’s classification (0.21-0.40=fair agreement; 0.41-0.60=moderate agreement; 0.61-0.80=substantial agreement; 0.81-1.0=near-perfect agreement).¹⁰ Three advice categories achieved near perfect agreement: nutrition (kappa= 0.94; 95% confidence interval [CI] 0.82-1.0; 99.2% agreement), physical activity (kappa=0.91; 95% CI, 0.84-0.99; 98.6% agreement), and weight loss (kappa=0.95; 95% CI, 0.82-1.0; 99.7% agreement). The nonspecific weight loss advice category had slightly lower agreement but still achieved near-perfect agreement (kappa=0.82; 95% CI, 0.62-1.0; 99.2% agreement).

After all advice was coded, we placed conversations into 1 of 6 categories: (1) no advice given; (2) nonspecific advice only; (3) nutrition only; (4) physical activity only; (5) weight loss only; or (6) combination of nutrition, physical activity, and/or weight loss.

Measures

Dietary fat and fiber intake. We assessed dietary fat intake at baseline and at 3 months using the 22-item Fat- and Fiber-Related Diet Behavior Questionnaire.^11,12 Questions about frequency of food selections included, “When you ate dessert, how often did you eat only fruit?” and “When you ate chicken, how often did you take off the skin?” We averaged responses into a total score wherein 1 reflected higher fiber, lower fat food choices; a score of 4 reflected lower fiber, higher fat choices (α=0.74 at baseline and α=0.77 at 3-month follow-up).

Physical activity. We measured physical activity (baseline, 3 months) using the Framingham Physical Activity Index.¹³ Participants recalled the average number of hours spent engaged in various daily activities (sleeping, working, leisure) and the level of activity for each (sedentary, slight, moderate, or heavy). The composite score accounts for activity duration and intensity.

Anthropometrics. We measured patient weight (baseline, 3 months) and height (baseline only) using a calibrated scale and portable stadiometer. Patients removed shoes, outerwear, and belongings from their pockets before being weighed.

Analysis
We analyzed data using SAS (SAS Institute, Inc., Cary, NC). We assessed the association between type of advice and weight loss, improvement in dietary fat intake behaviors, and increase in physical activity between baseline and the 3-month follow-up visit. We used PROC MIXED to fit general linear models; we incorporated responses into these models from all participants who provided measurements for at least one time point. This modeling framework yields unbiased estimates when missing data are unrelated to the observed variable.¹⁴

Primary predictors: (1) type of advice (none, nonspecific, nutrition, physical activity, weight loss, and combination), (2) time since baseline visit, and (3) time by type of advice interaction. All models included a priori defined patient, physician, and visit-related covariates that were theoretically or empirically related to changes in the outcomes (weight, physical activity, or dietary fat in-take). The 14 patient covariates were sex; age; race; high school education; economic security (enough money to pay monthly bills); over-weight (BMI, 25-29.9 kg/m²) or obese (BMI ≥30 kg/m²); actively trying to lose weight (yes/ no); motivated to lose weight (Likert scale 1-7); comfortable discussing weight (Likert scale 1-5); confident about losing weight (Likert scale 1-5); and patient-reported comorbid conditions of diabetes, hypertension, arthritis, and hyperlipidemia.

The 9 physician covariates were sex; race; years since medical school graduation; specialty (family vs internal medicine); self-efficacy (Likert scale 1-5); barriers for weight counseling (Likert scale 1-5); comfort discussing weight (Likert scale 1-5); insurance reimbursement concerns (Likert scale 1-5); and prior training in behavioral counseling (yes/no). Finally, 2 visit-level covariates were included: minutes spent addressing weight issues and visit type (preventive vs chronic).

Results

Sample characteristics
Of the 40 physicians, 19 were family physicians and 21 were internists. More than half of the physicians were female (60%), and 85% were white. Mean age was 47.2 years and mean BMI was 24.9 kg/m². Of the 461 patients, 66% were female, 65% were white, 35% were African American, and two-thirds had post-high school education (TABLE 1). Mean patient age was 59.8 years; only 4% of the patients were new to their physicians.

TABLE 1
Patient characteristics (N=461)

Frequency of advice
Physicians gave some type of weight-related advice in 63% of the encounters. They combined types of advice in 34% of all conversations, provided physical activity advice only in 13%, nutrition advice only in 8%, nonspecific advice in 5%, and weight loss advice only in 3%. Many times when physicians gave advice, it was centered on self (eg, “I need you to do X” or “What will it take for me to get you to do Y?”).

Nutrition advice most commonly pertained to specific food items from multiple categories (27% of conversations). Physicians also advised patients to reduce sugar/carbohydrates, control calories and portions, add other micronutrients, eat more fruits/vegetables, and eat meals more frequently.

Walking was the physical activity topic discussed most frequently, followed by exercise duration, exercise for comorbidities, aerobic activities, exercise intensity, and anaerobic exercise. The most common specific weight loss topic was weight loss behavioral advice, followed by weight loss for comorbid conditions. Physicians rarely provided referrals to weight-loss programs.

Effect of type of advice on fat and fiber diet behavior score
Receipt of nutrition advice only was not associated with reduction in fat intake (P=.43, TABLE 2). However, those who received combined types of advice exhibited a significantly greater reduction of fat intake compared with those who received no advice (Fat- and Fiber-Related Diet Behavior Questionnaire score reduction of 0.15 vs 0.05; P=.02).

TABLE 2
How types of physician advice affected dietary fat intake, physical activity, and weight

Effect of type of advice on Framingham Physical Activity score
No type of advice, including physical activity advice, led to a change in Framingham Physical Activity scores at the 3-month visit (overall P=.76; TABLE 2).

Effect of type of advice on weight loss
Patients who received physical activity advice gained significantly more weight than patients who received no advice (1.41 kg gained vs 0.18 kg lost; P=.02). Patients who received combined advice lost more weight than patients who received no advice, but the difference did not reach statistical significance (0.55 kg lost vs 0.18 kg lost; P=.08).

Discussion

Physicians typically took an “all or nothing” approach to weight-related issues, giving no advice (37%) or a combination of nutrition, physical activity, and weight loss advice (34%). It seems when physicians do give advice, most of them follow the USPSTF guidelines by addressing nutrition and physical activity together.¹⁵

Providing advice alone did not predict a change in patient behavior. For instance, we found no significant association between dietary fat reduction and having received only nutrition advice. Possible explanations include the following:

• Although physicians advised patients to reduce fat/cholesterol intake in 28% of conversations, they did so mostly in combination with other types of advice. Nutrition-only advice occurred in only 8% of conversations. Thus, there may have been insufficient power to detect the impact of this specific type of advice.
• With nutrition-only advice, the most common recommendation was to reduce carbohydrates/sweets, which should not affect fat intake.

Advising patients solely on physical activity led to unintended weight gain overall. Other data have shown that exercise without dietary changes, though beneficial in many ways, is not substantially effective for weight loss.¹⁵ People may eat more when they exercise, either to reward themselves or to satiate increased appetite from increased energy expenditure. Or, if physicians recommend the standard goal of 150 minutes of intensive physical activity per week, normally sedentary patients may see that as unattainable and become too discouraged to try.^1,16,17

Combining types of advice seemed to help patients reduce their fat intake. Overall, however, simple, brief advice from a physician may not be enough to promote healthy lifestyle changes.

Also notable was that physicians rarely provided referrals, even though this is a strong recommendation from the National Institutes of Health, the American Diabetes Association, and the USPSTF.^1,16,17 It could be that many physicians believe referrals are not covered by insurance. Yet, the low frequency of referrals may suggest an important missing component of weight loss therapy, especially given that physician advice alone seems an inadequate intervention.

Avoid physician-centered appeals. Advice was often given in a physician-centered way. There are 3 possible explanations for such phrasing:

1. In the absence of clear evidence about how to deliver weight loss advice, physicians may be formulating advice based on their personal or clinical experiences.
2. Physicians either assume or sense that patients lack internal motivation to make lifestyle changes for themselves and instead request that patients make changes for the doctor-patient relationship.
3. Physicians might be trying to invoke authority in the hope that patients will respond accordingly.

Whatever the reason, the literature on self-centered physician talk indicates that patients are less satisfied when physicians make the visit more about themselves than about patients.¹⁸ A better strategy might be to use Motivational Interviewing¹⁹ that supports patient autonomy and attempts to elicit and build on internal motivation.

The take-away message is that behavior change is complex and that knowledge is a necessary but insufficient agent for change. Following the tenets of Social Cognitive Theory,²⁰ physicians might also need to address patient motivation, confidence, outcome expectations, and skills to help promote behavior change.

Strengths and limitations of this study
We recorded conversations rather than relying on physician or patient recall. Additionally, these primary care patients were not enrolled in a weight-loss trial and, therefore, were not self-selected to be highly motivated to lose weight. Because of this, and the large and ethnically diverse sample, our results should be generalizable to many clinical settings.

One limitation is that few younger, lower-income patients were included in the sample, which limits generalizability to those populations. Also, the study was observational. Although we adjusted for a broad set of patient, physician, and visit covariates, unmeasured confounding variables may still account for at least part of the observed associations. The analysis is limited by the use of self-reported dietary fat intake and physical activity measures. A food diary and accelerometer would have been more accurate; however, such involved measures could invoke changes in behavior, which would have confounded our ability to assess the effect of physician advice on weight loss.

Acknowledgements

The authors thank all of the physicians and patients who participated in this study, the study project managers Gretchen Yonish and Iguehi Esoimeme, and research assistant Justin Manusov.

CORRESPONDENCE
Stewart C. Alexander, PhD, Department of Medicine, Duke University School of Medicine, P.O. Box 3140 Medical Center, Durham, NC 27710; [email protected]

The US Preventive Services Task Force (USPSTF) recommends that physicians screen patients for obesity and offer intensive counseling and behavioral interventions to promote sustained weight loss.¹ Evidence suggests that physician counseling, including advice, can help patients to lose weight, increase physical activity, and improve diet.^2-9 However, little is known about what specific types of weight loss advice physicians give to patients, and whether some types are more effective than others at influencing behavior change.

We analyzed physician weight loss advice delivered in primary care visits and measured changes in patients’ dietary intake, physical activity, and body weight. We examined both the type of weight loss advice delivered and the impact of type of advice on weight and behavior change.

Methods

This study analyzed audio recordings from Project CHAT – Communicating Health: Analyzing Talk. The project was approved by the Duke University Medical Center Institutional Review Board.

Recruitment Physicians. We obtained consent from 40 primary care physicians in community-based practices and told them the study would examine communication around preventive health topics, not weight specifically.

Patients. We identified potential participants by reviewing scheduled appointments 3 weeks in advance. Eligible participants were at least 18 years of age, English-speaking, overweight or obese (body mass index [BMI] ≥25 kg/m²), cognitively competent, and not pregnant. After we obtained consent, a remotely located research assistant started a digital audio recorder as the patient entered the exam room. Immediately after the encounter, the research assistant administered a post-encounter survey to the patient and recorded the patient’s vital signs (N=461). Three months later, the research assistant met with the participant to record vital signs and administer a survey assessing changes in dietary fat intake and exercise (N=426).

Data coding
We coded advice into 4 broad categories: (1) nutrition advice, (2) physical activity advice, (3) specific weight loss advice, and (4) nonspecific weight loss/weight-related advice. We transcribed each piece of advice verbatim.

Nutrition advice consisted of 9 sub-categories: calorie/portion control, meal timing/planning, commercial diet plans, negative diet plans, increase fruits/vegetables, reduce sugar/carbohydrates, reduce fat/cholesterol, other micronutrient recommendations, and specific food items from multiple categories.

Physical activity advice consisted of 6 subcategories: walking, aerobic exercise, anaerobic exercise, exercise intensity, exercise duration, and exercise for comorbid conditions.

Specific weight loss advice consisted of 3 categories: weight loss behavior, weight loss for comorbid conditions, and referrals.

Nonspecific weight loss advice also consisted of 3 subcategories in which physicians provided no details about the general topics of nutrition, physical activity, or weight loss.

Two independent coders (CBT and MEC) assessed each piece of advice and double coded 20% of conversations for reliability. Cohen’s kappa was used to calculate inter-rater reliability for each code using Landis and Koch’s classification (0.21-0.40=fair agreement; 0.41-0.60=moderate agreement; 0.61-0.80=substantial agreement; 0.81-1.0=near-perfect agreement).¹⁰ Three advice categories achieved near perfect agreement: nutrition (kappa= 0.94; 95% confidence interval [CI] 0.82-1.0; 99.2% agreement), physical activity (kappa=0.91; 95% CI, 0.84-0.99; 98.6% agreement), and weight loss (kappa=0.95; 95% CI, 0.82-1.0; 99.7% agreement). The nonspecific weight loss advice category had slightly lower agreement but still achieved near-perfect agreement (kappa=0.82; 95% CI, 0.62-1.0; 99.2% agreement).

After all advice was coded, we placed conversations into 1 of 6 categories: (1) no advice given; (2) nonspecific advice only; (3) nutrition only; (4) physical activity only; (5) weight loss only; or (6) combination of nutrition, physical activity, and/or weight loss.

Measures

Dietary fat and fiber intake. We assessed dietary fat intake at baseline and at 3 months using the 22-item Fat- and Fiber-Related Diet Behavior Questionnaire.^11,12 Questions about frequency of food selections included, “When you ate dessert, how often did you eat only fruit?” and “When you ate chicken, how often did you take off the skin?” We averaged responses into a total score wherein 1 reflected higher fiber, lower fat food choices; a score of 4 reflected lower fiber, higher fat choices (α=0.74 at baseline and α=0.77 at 3-month follow-up).

Physical activity. We measured physical activity (baseline, 3 months) using the Framingham Physical Activity Index.¹³ Participants recalled the average number of hours spent engaged in various daily activities (sleeping, working, leisure) and the level of activity for each (sedentary, slight, moderate, or heavy). The composite score accounts for activity duration and intensity.

Anthropometrics. We measured patient weight (baseline, 3 months) and height (baseline only) using a calibrated scale and portable stadiometer. Patients removed shoes, outerwear, and belongings from their pockets before being weighed.

Analysis
We analyzed data using SAS (SAS Institute, Inc., Cary, NC). We assessed the association between type of advice and weight loss, improvement in dietary fat intake behaviors, and increase in physical activity between baseline and the 3-month follow-up visit. We used PROC MIXED to fit general linear models; we incorporated responses into these models from all participants who provided measurements for at least one time point. This modeling framework yields unbiased estimates when missing data are unrelated to the observed variable.¹⁴

Primary predictors: (1) type of advice (none, nonspecific, nutrition, physical activity, weight loss, and combination), (2) time since baseline visit, and (3) time by type of advice interaction. All models included a priori defined patient, physician, and visit-related covariates that were theoretically or empirically related to changes in the outcomes (weight, physical activity, or dietary fat in-take). The 14 patient covariates were sex; age; race; high school education; economic security (enough money to pay monthly bills); over-weight (BMI, 25-29.9 kg/m²) or obese (BMI ≥30 kg/m²); actively trying to lose weight (yes/ no); motivated to lose weight (Likert scale 1-7); comfortable discussing weight (Likert scale 1-5); confident about losing weight (Likert scale 1-5); and patient-reported comorbid conditions of diabetes, hypertension, arthritis, and hyperlipidemia.

The 9 physician covariates were sex; race; years since medical school graduation; specialty (family vs internal medicine); self-efficacy (Likert scale 1-5); barriers for weight counseling (Likert scale 1-5); comfort discussing weight (Likert scale 1-5); insurance reimbursement concerns (Likert scale 1-5); and prior training in behavioral counseling (yes/no). Finally, 2 visit-level covariates were included: minutes spent addressing weight issues and visit type (preventive vs chronic).

Results

Sample characteristics
Of the 40 physicians, 19 were family physicians and 21 were internists. More than half of the physicians were female (60%), and 85% were white. Mean age was 47.2 years and mean BMI was 24.9 kg/m². Of the 461 patients, 66% were female, 65% were white, 35% were African American, and two-thirds had post-high school education (TABLE 1). Mean patient age was 59.8 years; only 4% of the patients were new to their physicians.

TABLE 1
Patient characteristics (N=461)

Frequency of advice
Physicians gave some type of weight-related advice in 63% of the encounters. They combined types of advice in 34% of all conversations, provided physical activity advice only in 13%, nutrition advice only in 8%, nonspecific advice in 5%, and weight loss advice only in 3%. Many times when physicians gave advice, it was centered on self (eg, “I need you to do X” or “What will it take for me to get you to do Y?”).

Nutrition advice most commonly pertained to specific food items from multiple categories (27% of conversations). Physicians also advised patients to reduce sugar/carbohydrates, control calories and portions, add other micronutrients, eat more fruits/vegetables, and eat meals more frequently.

Walking was the physical activity topic discussed most frequently, followed by exercise duration, exercise for comorbidities, aerobic activities, exercise intensity, and anaerobic exercise. The most common specific weight loss topic was weight loss behavioral advice, followed by weight loss for comorbid conditions. Physicians rarely provided referrals to weight-loss programs.

Effect of type of advice on fat and fiber diet behavior score
Receipt of nutrition advice only was not associated with reduction in fat intake (P=.43, TABLE 2). However, those who received combined types of advice exhibited a significantly greater reduction of fat intake compared with those who received no advice (Fat- and Fiber-Related Diet Behavior Questionnaire score reduction of 0.15 vs 0.05; P=.02).

TABLE 2
How types of physician advice affected dietary fat intake, physical activity, and weight

Effect of type of advice on Framingham Physical Activity score
No type of advice, including physical activity advice, led to a change in Framingham Physical Activity scores at the 3-month visit (overall P=.76; TABLE 2).

Effect of type of advice on weight loss
Patients who received physical activity advice gained significantly more weight than patients who received no advice (1.41 kg gained vs 0.18 kg lost; P=.02). Patients who received combined advice lost more weight than patients who received no advice, but the difference did not reach statistical significance (0.55 kg lost vs 0.18 kg lost; P=.08).

Discussion

Physicians typically took an “all or nothing” approach to weight-related issues, giving no advice (37%) or a combination of nutrition, physical activity, and weight loss advice (34%). It seems when physicians do give advice, most of them follow the USPSTF guidelines by addressing nutrition and physical activity together.¹⁵

Providing advice alone did not predict a change in patient behavior. For instance, we found no significant association between dietary fat reduction and having received only nutrition advice. Possible explanations include the following:

• Although physicians advised patients to reduce fat/cholesterol intake in 28% of conversations, they did so mostly in combination with other types of advice. Nutrition-only advice occurred in only 8% of conversations. Thus, there may have been insufficient power to detect the impact of this specific type of advice.
• With nutrition-only advice, the most common recommendation was to reduce carbohydrates/sweets, which should not affect fat intake.

Advising patients solely on physical activity led to unintended weight gain overall. Other data have shown that exercise without dietary changes, though beneficial in many ways, is not substantially effective for weight loss.¹⁵ People may eat more when they exercise, either to reward themselves or to satiate increased appetite from increased energy expenditure. Or, if physicians recommend the standard goal of 150 minutes of intensive physical activity per week, normally sedentary patients may see that as unattainable and become too discouraged to try.^1,16,17

Combining types of advice seemed to help patients reduce their fat intake. Overall, however, simple, brief advice from a physician may not be enough to promote healthy lifestyle changes.

Also notable was that physicians rarely provided referrals, even though this is a strong recommendation from the National Institutes of Health, the American Diabetes Association, and the USPSTF.^1,16,17 It could be that many physicians believe referrals are not covered by insurance. Yet, the low frequency of referrals may suggest an important missing component of weight loss therapy, especially given that physician advice alone seems an inadequate intervention.

Avoid physician-centered appeals. Advice was often given in a physician-centered way. There are 3 possible explanations for such phrasing:

1. In the absence of clear evidence about how to deliver weight loss advice, physicians may be formulating advice based on their personal or clinical experiences.
2. Physicians either assume or sense that patients lack internal motivation to make lifestyle changes for themselves and instead request that patients make changes for the doctor-patient relationship.
3. Physicians might be trying to invoke authority in the hope that patients will respond accordingly.

Whatever the reason, the literature on self-centered physician talk indicates that patients are less satisfied when physicians make the visit more about themselves than about patients.¹⁸ A better strategy might be to use Motivational Interviewing¹⁹ that supports patient autonomy and attempts to elicit and build on internal motivation.

The take-away message is that behavior change is complex and that knowledge is a necessary but insufficient agent for change. Following the tenets of Social Cognitive Theory,²⁰ physicians might also need to address patient motivation, confidence, outcome expectations, and skills to help promote behavior change.

Strengths and limitations of this study
We recorded conversations rather than relying on physician or patient recall. Additionally, these primary care patients were not enrolled in a weight-loss trial and, therefore, were not self-selected to be highly motivated to lose weight. Because of this, and the large and ethnically diverse sample, our results should be generalizable to many clinical settings.

One limitation is that few younger, lower-income patients were included in the sample, which limits generalizability to those populations. Also, the study was observational. Although we adjusted for a broad set of patient, physician, and visit covariates, unmeasured confounding variables may still account for at least part of the observed associations. The analysis is limited by the use of self-reported dietary fat intake and physical activity measures. A food diary and accelerometer would have been more accurate; however, such involved measures could invoke changes in behavior, which would have confounded our ability to assess the effect of physician advice on weight loss.

Acknowledgements

The authors thank all of the physicians and patients who participated in this study, the study project managers Gretchen Yonish and Iguehi Esoimeme, and research assistant Justin Manusov.

CORRESPONDENCE
Stewart C. Alexander, PhD, Department of Medicine, Duke University School of Medicine, P.O. Box 3140 Medical Center, Durham, NC 27710; [email protected]

The US Preventive Services Task Force (USPSTF) recommends that physicians screen patients for obesity and offer intensive counseling and behavioral interventions to promote sustained weight loss.¹ Evidence suggests that physician counseling, including advice, can help patients to lose weight, increase physical activity, and improve diet.^2-9 However, little is known about what specific types of weight loss advice physicians give to patients, and whether some types are more effective than others at influencing behavior change.

We analyzed physician weight loss advice delivered in primary care visits and measured changes in patients’ dietary intake, physical activity, and body weight. We examined both the type of weight loss advice delivered and the impact of type of advice on weight and behavior change.

Methods

This study analyzed audio recordings from Project CHAT – Communicating Health: Analyzing Talk. The project was approved by the Duke University Medical Center Institutional Review Board.

Recruitment Physicians. We obtained consent from 40 primary care physicians in community-based practices and told them the study would examine communication around preventive health topics, not weight specifically.

Patients. We identified potential participants by reviewing scheduled appointments 3 weeks in advance. Eligible participants were at least 18 years of age, English-speaking, overweight or obese (body mass index [BMI] ≥25 kg/m²), cognitively competent, and not pregnant. After we obtained consent, a remotely located research assistant started a digital audio recorder as the patient entered the exam room. Immediately after the encounter, the research assistant administered a post-encounter survey to the patient and recorded the patient’s vital signs (N=461). Three months later, the research assistant met with the participant to record vital signs and administer a survey assessing changes in dietary fat intake and exercise (N=426).

Data coding
We coded advice into 4 broad categories: (1) nutrition advice, (2) physical activity advice, (3) specific weight loss advice, and (4) nonspecific weight loss/weight-related advice. We transcribed each piece of advice verbatim.

Nutrition advice consisted of 9 sub-categories: calorie/portion control, meal timing/planning, commercial diet plans, negative diet plans, increase fruits/vegetables, reduce sugar/carbohydrates, reduce fat/cholesterol, other micronutrient recommendations, and specific food items from multiple categories.

Physical activity advice consisted of 6 subcategories: walking, aerobic exercise, anaerobic exercise, exercise intensity, exercise duration, and exercise for comorbid conditions.

Specific weight loss advice consisted of 3 categories: weight loss behavior, weight loss for comorbid conditions, and referrals.

Nonspecific weight loss advice also consisted of 3 subcategories in which physicians provided no details about the general topics of nutrition, physical activity, or weight loss.

Two independent coders (CBT and MEC) assessed each piece of advice and double coded 20% of conversations for reliability. Cohen’s kappa was used to calculate inter-rater reliability for each code using Landis and Koch’s classification (0.21-0.40=fair agreement; 0.41-0.60=moderate agreement; 0.61-0.80=substantial agreement; 0.81-1.0=near-perfect agreement).¹⁰ Three advice categories achieved near perfect agreement: nutrition (kappa= 0.94; 95% confidence interval [CI] 0.82-1.0; 99.2% agreement), physical activity (kappa=0.91; 95% CI, 0.84-0.99; 98.6% agreement), and weight loss (kappa=0.95; 95% CI, 0.82-1.0; 99.7% agreement). The nonspecific weight loss advice category had slightly lower agreement but still achieved near-perfect agreement (kappa=0.82; 95% CI, 0.62-1.0; 99.2% agreement).

After all advice was coded, we placed conversations into 1 of 6 categories: (1) no advice given; (2) nonspecific advice only; (3) nutrition only; (4) physical activity only; (5) weight loss only; or (6) combination of nutrition, physical activity, and/or weight loss.

Measures

Dietary fat and fiber intake. We assessed dietary fat intake at baseline and at 3 months using the 22-item Fat- and Fiber-Related Diet Behavior Questionnaire.^11,12 Questions about frequency of food selections included, “When you ate dessert, how often did you eat only fruit?” and “When you ate chicken, how often did you take off the skin?” We averaged responses into a total score wherein 1 reflected higher fiber, lower fat food choices; a score of 4 reflected lower fiber, higher fat choices (α=0.74 at baseline and α=0.77 at 3-month follow-up).

Physical activity. We measured physical activity (baseline, 3 months) using the Framingham Physical Activity Index.¹³ Participants recalled the average number of hours spent engaged in various daily activities (sleeping, working, leisure) and the level of activity for each (sedentary, slight, moderate, or heavy). The composite score accounts for activity duration and intensity.

Anthropometrics. We measured patient weight (baseline, 3 months) and height (baseline only) using a calibrated scale and portable stadiometer. Patients removed shoes, outerwear, and belongings from their pockets before being weighed.

Analysis
We analyzed data using SAS (SAS Institute, Inc., Cary, NC). We assessed the association between type of advice and weight loss, improvement in dietary fat intake behaviors, and increase in physical activity between baseline and the 3-month follow-up visit. We used PROC MIXED to fit general linear models; we incorporated responses into these models from all participants who provided measurements for at least one time point. This modeling framework yields unbiased estimates when missing data are unrelated to the observed variable.¹⁴

Primary predictors: (1) type of advice (none, nonspecific, nutrition, physical activity, weight loss, and combination), (2) time since baseline visit, and (3) time by type of advice interaction. All models included a priori defined patient, physician, and visit-related covariates that were theoretically or empirically related to changes in the outcomes (weight, physical activity, or dietary fat in-take). The 14 patient covariates were sex; age; race; high school education; economic security (enough money to pay monthly bills); over-weight (BMI, 25-29.9 kg/m²) or obese (BMI ≥30 kg/m²); actively trying to lose weight (yes/ no); motivated to lose weight (Likert scale 1-7); comfortable discussing weight (Likert scale 1-5); confident about losing weight (Likert scale 1-5); and patient-reported comorbid conditions of diabetes, hypertension, arthritis, and hyperlipidemia.

The 9 physician covariates were sex; race; years since medical school graduation; specialty (family vs internal medicine); self-efficacy (Likert scale 1-5); barriers for weight counseling (Likert scale 1-5); comfort discussing weight (Likert scale 1-5); insurance reimbursement concerns (Likert scale 1-5); and prior training in behavioral counseling (yes/no). Finally, 2 visit-level covariates were included: minutes spent addressing weight issues and visit type (preventive vs chronic).

Results

Sample characteristics
Of the 40 physicians, 19 were family physicians and 21 were internists. More than half of the physicians were female (60%), and 85% were white. Mean age was 47.2 years and mean BMI was 24.9 kg/m². Of the 461 patients, 66% were female, 65% were white, 35% were African American, and two-thirds had post-high school education (TABLE 1). Mean patient age was 59.8 years; only 4% of the patients were new to their physicians.

TABLE 1
Patient characteristics (N=461)

Frequency of advice
Physicians gave some type of weight-related advice in 63% of the encounters. They combined types of advice in 34% of all conversations, provided physical activity advice only in 13%, nutrition advice only in 8%, nonspecific advice in 5%, and weight loss advice only in 3%. Many times when physicians gave advice, it was centered on self (eg, “I need you to do X” or “What will it take for me to get you to do Y?”).

Nutrition advice most commonly pertained to specific food items from multiple categories (27% of conversations). Physicians also advised patients to reduce sugar/carbohydrates, control calories and portions, add other micronutrients, eat more fruits/vegetables, and eat meals more frequently.

Walking was the physical activity topic discussed most frequently, followed by exercise duration, exercise for comorbidities, aerobic activities, exercise intensity, and anaerobic exercise. The most common specific weight loss topic was weight loss behavioral advice, followed by weight loss for comorbid conditions. Physicians rarely provided referrals to weight-loss programs.

Effect of type of advice on fat and fiber diet behavior score
Receipt of nutrition advice only was not associated with reduction in fat intake (P=.43, TABLE 2). However, those who received combined types of advice exhibited a significantly greater reduction of fat intake compared with those who received no advice (Fat- and Fiber-Related Diet Behavior Questionnaire score reduction of 0.15 vs 0.05; P=.02).

TABLE 2
How types of physician advice affected dietary fat intake, physical activity, and weight

Effect of type of advice on Framingham Physical Activity score
No type of advice, including physical activity advice, led to a change in Framingham Physical Activity scores at the 3-month visit (overall P=.76; TABLE 2).

Effect of type of advice on weight loss
Patients who received physical activity advice gained significantly more weight than patients who received no advice (1.41 kg gained vs 0.18 kg lost; P=.02). Patients who received combined advice lost more weight than patients who received no advice, but the difference did not reach statistical significance (0.55 kg lost vs 0.18 kg lost; P=.08).

Discussion

Physicians typically took an “all or nothing” approach to weight-related issues, giving no advice (37%) or a combination of nutrition, physical activity, and weight loss advice (34%). It seems when physicians do give advice, most of them follow the USPSTF guidelines by addressing nutrition and physical activity together.¹⁵

Providing advice alone did not predict a change in patient behavior. For instance, we found no significant association between dietary fat reduction and having received only nutrition advice. Possible explanations include the following:

• Although physicians advised patients to reduce fat/cholesterol intake in 28% of conversations, they did so mostly in combination with other types of advice. Nutrition-only advice occurred in only 8% of conversations. Thus, there may have been insufficient power to detect the impact of this specific type of advice.
• With nutrition-only advice, the most common recommendation was to reduce carbohydrates/sweets, which should not affect fat intake.

Advising patients solely on physical activity led to unintended weight gain overall. Other data have shown that exercise without dietary changes, though beneficial in many ways, is not substantially effective for weight loss.¹⁵ People may eat more when they exercise, either to reward themselves or to satiate increased appetite from increased energy expenditure. Or, if physicians recommend the standard goal of 150 minutes of intensive physical activity per week, normally sedentary patients may see that as unattainable and become too discouraged to try.^1,16,17

Combining types of advice seemed to help patients reduce their fat intake. Overall, however, simple, brief advice from a physician may not be enough to promote healthy lifestyle changes.

Also notable was that physicians rarely provided referrals, even though this is a strong recommendation from the National Institutes of Health, the American Diabetes Association, and the USPSTF.^1,16,17 It could be that many physicians believe referrals are not covered by insurance. Yet, the low frequency of referrals may suggest an important missing component of weight loss therapy, especially given that physician advice alone seems an inadequate intervention.

Avoid physician-centered appeals. Advice was often given in a physician-centered way. There are 3 possible explanations for such phrasing:

1. In the absence of clear evidence about how to deliver weight loss advice, physicians may be formulating advice based on their personal or clinical experiences.
2. Physicians either assume or sense that patients lack internal motivation to make lifestyle changes for themselves and instead request that patients make changes for the doctor-patient relationship.
3. Physicians might be trying to invoke authority in the hope that patients will respond accordingly.

Whatever the reason, the literature on self-centered physician talk indicates that patients are less satisfied when physicians make the visit more about themselves than about patients.¹⁸ A better strategy might be to use Motivational Interviewing¹⁹ that supports patient autonomy and attempts to elicit and build on internal motivation.

The take-away message is that behavior change is complex and that knowledge is a necessary but insufficient agent for change. Following the tenets of Social Cognitive Theory,²⁰ physicians might also need to address patient motivation, confidence, outcome expectations, and skills to help promote behavior change.

Strengths and limitations of this study
We recorded conversations rather than relying on physician or patient recall. Additionally, these primary care patients were not enrolled in a weight-loss trial and, therefore, were not self-selected to be highly motivated to lose weight. Because of this, and the large and ethnically diverse sample, our results should be generalizable to many clinical settings.

One limitation is that few younger, lower-income patients were included in the sample, which limits generalizability to those populations. Also, the study was observational. Although we adjusted for a broad set of patient, physician, and visit covariates, unmeasured confounding variables may still account for at least part of the observed associations. The analysis is limited by the use of self-reported dietary fat intake and physical activity measures. A food diary and accelerometer would have been more accurate; however, such involved measures could invoke changes in behavior, which would have confounded our ability to assess the effect of physician advice on weight loss.

Acknowledgements

The authors thank all of the physicians and patients who participated in this study, the study project managers Gretchen Yonish and Iguehi Esoimeme, and research assistant Justin Manusov.

CORRESPONDENCE
Stewart C. Alexander, PhD, Department of Medicine, Duke University School of Medicine, P.O. Box 3140 Medical Center, Durham, NC 27710; [email protected]

Research has shown that when physicians advise overweight patients to lose weight, improve their diet, or increase their physical activity, patients are more likely to report attempting to do so.^1-3 In a study of 433 primary care patients, 46% reported trying to lose weight after their physician counseled them about nutrition, compared with 37% who were not counseled.¹

The reality, though, is that physicians are not very likely to address weight loss. Data from the Behavioral Risk Factor Surveillance System indicate that patients report their providers address weight loss in fewer than 20% of their examinations.⁴ These low rates are concerning; when physicians do not advise patients to lose weight, patients may believe their weight is not a problem.⁵ Even more worrisome: Physicians are rarely trained on how to counsel patients about weight loss. So, when physicians do counsel patients, it may not be effective.

Using Motivational Interviewing

One effective style of counseling is Motivational Interviewing (MI). MI is a patient-centered, directive counseling style used to help patients explore and resolve their ambivalence related to a particular behavior change (see What is Motivational Interviewing?).^6,7 Researchers have studied the use of MI by counselors and case managers (in handling smoking cessation),^8-11 but not by physicians. Further, no one has examined whether physicians instinctively use MI techniques when discussing weight loss with their patients, or whether MI counseling results in patients trying to lose weight.

The primary aim of this study was to assess how overweight and obese female patients discuss weight loss with their physicians. We also wanted to explore the role that physicians’ way of discussing weight loss—and the use of MI in particular—might play in their patients’ motivation to lose weight.

Methods

Setting and recruitment

All data were collected in a family practice clinic within Duke University Medical Center. We approached 9 physicians in the practice to participate, and all consented. Only 7 physicians had visits with overweight or obese patients and were included in this report. We reviewed their electronic patient appointment schedules twice a week to identify female patients meeting the following criteria: English-speaking, overweight or obese (body mass index [BMI] ≥25 kg/m²), 40 years of age or older, and with health maintenance or chronic care appointments scheduled at least 7 days later. We sent these patients a letter describing the study, and allowed them 7 days to call a toll-free number if they didn’t want to participate.

We took several steps to avoid priming physicians and patients about the purpose of the study. First, both physicians and patients were told the study was about how doctors and patients discuss preventive health topics—they were not told the study was about examining discussions of weight. Second, we surveyed physicians 1 month prior to audio-recording visits, and patients 1 week prior to their visit. Third, we included measures for other preventive health topics (eg, smoking and alcohol) to detract attention from weight.

Gathering data

1. Phone survey before patient visit. We telephoned those patients who did not refuse participation and conducted a baseline survey. We asked about date of birth, race, marital status, level of education, income, weight, height, history of weight loss attempts, and whether this was their first visit with that physician. We categorized women with a BMI ≥25 but <30 as overweight, and those with a BMI ≥30 as obese.¹²

We also assessed each patient’s

• self-efficacy—that is, confidence in their ability to lose weight. We asked: “How confident are you that you can lose weight?” (1=not at all confident, 5=extremely confident).
• readiness to lose weight. We asked: “Are you seriously considering trying to losing weight within the next 6 months?” and, if yes, “Are you planning to try to lose weight in the next 30 days?”¹³ Those not considering trying to lose weight were staged as precontemplation; those who were considering trying but not planning to try in the next 30 days were staged as contemplation; and those who were planning to try to lose weight in the next 30 days were staged as preparation.

2. Office visit. When patients came in for their appointments, the research assistant gave them consent forms to sign. The assistant then escorted the patient to the examination rooms and started the digital audio recorder. The exams typically took 27 minutes.

Immediately following the exam, the research assistant surveyed the patients. The assistant asked 2 questions we’d asked at baseline: “How confident are you that you can lose weight?” and “Are you seriously considering trying to losing weight within the next 6 months?” (If yes, “Are you planning to try to lose weight in the next 30 days?”) She also made an appointment to conduct a 1-month follow-up telephone survey.

3. One-month follow-up survey. During a follow-up phone survey, we asked patients whether they had attempted to lose weight by changing their diet, exercise patterns, or both. Subsequent to this call, we sent the study participants a $25 check.

Analyzing the patient-physician discussion

Content. Two authors coded 9 topics that physicians and patients discussed that were “weight-related.” Topics included: physical activity, diet, BMI, psychosocial issues, referral to a nutritionist, weight loss surgery, goal setting, weight loss medications, and health care avoidance. We also coded who first brought up the topic.

Time spent. We calculated time spent discussing weight-related topics and also the total time of the patient’s visit.

Motivational Interviewing. Two coders assessed MI. To assess fidelity to MI principles, we used sections of the Motivational Interviewing Treatment Integrity scale (MITI)¹⁴ to rate patient interactions on a scale of 1 (low) to 7 (high) in 2 categories: empathy and MI spirit.

• Empathy is when physicians convey understanding of patients’ perspective.
• MI spirit includes evocation, collaboration, and autonomy. Evocation is when physicians draw out patients’ own reasons for change. Collaboration is when physicians act as partners, supporting and exploring patients’ concerns. Autonomy is when physicians convey that decisions to change lie completely with patients. Inter-rater reliability for the Empathy and MI Spirit was adequate (ICC=.94 and .97, respectively).
• MI-adherent behaviors were those where the physician asked permission to do things, affirmed statements, offered words of support, and emphasized patient control. For instance, the physician might say, “It’s great that you have stopped drinking sweetened tea” or “Whether you lose weight is up to you.”
• MI-nonadherent behaviors were those where the physician advised without asking permission. For example, the physician might say, “Let me tell you what you need to do to make this work…” or “Well, if you want to continue on the way you are, you know your diabetes is only going to get worse.”

These were combined to create a ratio of percentage MI-adherent behaviors by dividing MI-adherent by MI-nonadherent. There was an excellent level of agreement between coders for MI-nonadherent (kappa=.80) and a moderate level of agreement for MI-adherent (kappa=.52) behaviors.

Data analysis

We used Spearman correlations to assess the relationship between our predictors, quantity (time spent and whether weight was addressed) and quality (MI techniques), and mediators of behavior change (readiness to lose weight and self-efficacy to lose weight) and behavior change (attempts to lose weight, change in diet, and change in exercise patterns). We used SAS 9.1 (SAS Institute, Inc, Cary, NC) for all analyses. The study was approved by the Duke University Medical Center Institutional Review Board.

Results

We identified 202 eligible female patients. Of those, 96 had appointments that passed before we could contact them; 11 called the 800 number to refuse. Of the remaining 95 women, we reached 94 by phone. Of those, 19 refused to participate, 46 were ineligible because we had reached the targeted number of women in their weight category, and 4 skipped their appointments. Thus, we audio-recorded 25 encounters (for 14 obese and 11 overweight patients). Of these 25 patients, 24 completed the 1-month follow-up.

Patient demographics. Patients had a mean age of 59 years (standard deviation [SD]=11). Half were white; 42% were college-educated. Forty-two percent reported being in poor to fair health (TABLE 1).

The typical participant was moderately confident and ready to lose weight both before and after their visit. One month after their visit, 63% reported attempting to lose weight. More than half attempted to change their diet (67%); slightly more than half changed their exercise patterns (58%) (TABLE 2).

Physician demographics. Physicians had a mean age of 43 years (SD=10). About half were white; about half were female. No physicians were overweight.

TABLE 1
Characteristics of patients and physicians

TABLE 2
Feeling about weight loss before and after the visit

Patients were more likely to raise the weight issue

Weight-related topics were addressed in 19 of the 25 encounters (11 out of 12 preventive health visits, 8 out of 13 chronic care visits). The mean time spent discussing weight-related topics was 6.9 minutes out of an mean total of 27.0 minutes, or 26% of the total patient-physician time. Weight was more likely to be addressed with obese patients (86%) than with overweight patients (63%).

Patients were more likely than physicians to initiate discussions on weight. Physicians raised weight-related topics 37% of the time. Obese patients were slightly more likely to raise weight-related topics (8 out of 12 times [67%]) than overweight patients (4 out of 7 times [57%]).

The weight-related topics addressed were, in order from most to least frequent: physical activity, diet, BMI, psychosocial issues (eg, motivation to lose weight, triggers for unhealthful eating [such as family cookouts], negative talk [such as telling oneself that losing weight is too hard]), referral to a nutritionist, weight loss surgery, goal setting, health care avoidance, and weight loss medication. When comparing those who attempted to lose weight (n=15) with those who did not (n=9), there was no significant difference in whether or how often a topic was addressed.

Physicians’ empathy scores are moderate

Physicians had a moderate score for Empathy (mean=3.8, standard deviation [SD]=1.5, on 7-point scale), a low score for MI Spirit (mean=2.4, standard deviation [SD]=1.4, on 7-point scale), and displayed fewer MI-adherent behaviors than MI-nonadherent behaviors (mean=0.4, SD=0.3). These means did not differ significantly based on the patients’ weight.

Weight loss conversations linked to patients’ readiness

The discussion of weight-related topics, and the time spent doing so, were related to patients’ readiness to lose weight after their initial examination, when patients’ baseline readiness to lose weight was controlled. The more ready patients were to lose weight after their visit, the more likely they had discussed weight (Spearman’s rank correlation coefficient [r]=.52, P=.01) and spent more time discussing weight (r=.42, P=.05). No other associations were statistically significant (TABLE W1).

Several of the Motivational Interviewing scores predicted patients’ outcomes. When physicians showed more empathy, patients were more likely to report changing their exercise patterns 1 month after the visit (r=.50, P=.02). When physicians displayed more of an MI Spirit, patients were more likely to be ready to lose weight (r=.63, P=.005) and change their exercise patterns (r=.47, P=.04). Further, when physicians used more MI-adherent techniques, patients were more likely to attempt to lose weight (r=.42, P=.08).

Discussion: Good quality discussions lead to change

While more discussion about weight loss led to a greater readiness to lose weight, it was the quality of the discussions that actually led to behavior changes. Most patients had virtually the same levels of readiness to lose weight before and after the visit. It is likely that patients who were ready to lose weight discussed their weight with their physicians—and spent more time discussing it than those patients who were not ready to lose weight.

How patients and physicians discussed weight influenced behavior change. When physicians were more empathic and used techniques consistent with Motivational Interviewing, patients were more likely to report changing their exercise routine and attempting to lose weight.

To date, no one has examined the effect of physicians’ MI techniques on weight-related behavior change in a large study. The low adherence to MI techniques suggests that physicians can improve their counseling skills.

Patients aren’t afraid to talk about their weight

Unexpectedly, patients were more likely than physicians to initiate weight-related discussions. Only one third of the time did physicians raise the topic. Patients appear to be “empowered” to initiate discussions about weight loss. We expected physicians and patients to both be somewhat apprehensive about raising this sensitive topic. However, these findings suggest that overweight and obese patients will initiate the discussion most of the time.

Limitations and strengths of this study

The small sample size limited the analyses. Nonetheless, we found strong correlations in this sample that suggest true relationships that were unlikely to have occurred by chance. Also, we were unable to conduct nested analyses to account for the clustering of patients seen by the same physicians. The results may not generalize to settings outside of academic medical centers and practices in which physicians have less time to spend with patients.

The physicians in this study were not overweight, which could limit the generalizability of the results. Patients may be less likely to raise the topic of weight with physicians who were themselves overweight. In addition, while we assessed single-item outcomes, more objective and extensive standard measures of diet, physical activity, and weight loss would have been optimal.

Some notable strengths of this study were that we used a comprehensive multimodal measurement in assessing both content and style of conversations in addition to patient self-report. We also examined Motivational Interviewing techniques among physicians with little or no MI training; most studies have examined MI among trained counselors only.

How to talk about weight loss: More study is needed

The most commonly addressed weight-related topics were diet and physical activity. However, when looking at the topics that were discussed, we found no patterns between those who attempted to lose weight and those who didn’t. This may mean that because weight loss is such a complex behavior, mention of any aspect of it—be it physical activity, diet, psychosocial issues, and so on—helps patients in their efforts. It also could be that the physician and patient discussed some other aspects in a previous visit; therefore, it was the cumulative effect of many conversations that influenced the patient to change.

These results need to be explored in a larger study to understand whether discussing certain topics is more influential than discussing others in promoting weight loss.

Acknowledgments

We thank Miranda West, Laura Fish, and Mary Sochaki for their work on this project. We are also grateful to the physicians and patients who agreed to have their encounters audio recorded.

Funding

This work was supported by National Cancer Institute grant 2P50 CA68438-06A2. The authors were supported in part by National Cancer Institute grants R01CA089053, R01CA100387, and National Institute of Diabetes and Digestive and Kidney Disorders grant R01DK64986.

CorrespondenceKathryn I. Pollak, PhD, Duke Comprehensive Cancer Center, Cancer Prevention, Detection and Control Research Program, 2424 Erwin Road, Room 6029, Hock Plaza I, Suite 602, Durham, NC 27705; [email protected].

Research has shown that when physicians advise overweight patients to lose weight, improve their diet, or increase their physical activity, patients are more likely to report attempting to do so.^1-3 In a study of 433 primary care patients, 46% reported trying to lose weight after their physician counseled them about nutrition, compared with 37% who were not counseled.¹

The reality, though, is that physicians are not very likely to address weight loss. Data from the Behavioral Risk Factor Surveillance System indicate that patients report their providers address weight loss in fewer than 20% of their examinations.⁴ These low rates are concerning; when physicians do not advise patients to lose weight, patients may believe their weight is not a problem.⁵ Even more worrisome: Physicians are rarely trained on how to counsel patients about weight loss. So, when physicians do counsel patients, it may not be effective.

Using Motivational Interviewing

One effective style of counseling is Motivational Interviewing (MI). MI is a patient-centered, directive counseling style used to help patients explore and resolve their ambivalence related to a particular behavior change (see What is Motivational Interviewing?).^6,7 Researchers have studied the use of MI by counselors and case managers (in handling smoking cessation),^8-11 but not by physicians. Further, no one has examined whether physicians instinctively use MI techniques when discussing weight loss with their patients, or whether MI counseling results in patients trying to lose weight.

The primary aim of this study was to assess how overweight and obese female patients discuss weight loss with their physicians. We also wanted to explore the role that physicians’ way of discussing weight loss—and the use of MI in particular—might play in their patients’ motivation to lose weight.

Methods

Setting and recruitment

All data were collected in a family practice clinic within Duke University Medical Center. We approached 9 physicians in the practice to participate, and all consented. Only 7 physicians had visits with overweight or obese patients and were included in this report. We reviewed their electronic patient appointment schedules twice a week to identify female patients meeting the following criteria: English-speaking, overweight or obese (body mass index [BMI] ≥25 kg/m²), 40 years of age or older, and with health maintenance or chronic care appointments scheduled at least 7 days later. We sent these patients a letter describing the study, and allowed them 7 days to call a toll-free number if they didn’t want to participate.

We took several steps to avoid priming physicians and patients about the purpose of the study. First, both physicians and patients were told the study was about how doctors and patients discuss preventive health topics—they were not told the study was about examining discussions of weight. Second, we surveyed physicians 1 month prior to audio-recording visits, and patients 1 week prior to their visit. Third, we included measures for other preventive health topics (eg, smoking and alcohol) to detract attention from weight.

Gathering data

1. Phone survey before patient visit. We telephoned those patients who did not refuse participation and conducted a baseline survey. We asked about date of birth, race, marital status, level of education, income, weight, height, history of weight loss attempts, and whether this was their first visit with that physician. We categorized women with a BMI ≥25 but <30 as overweight, and those with a BMI ≥30 as obese.¹²

We also assessed each patient’s

• self-efficacy—that is, confidence in their ability to lose weight. We asked: “How confident are you that you can lose weight?” (1=not at all confident, 5=extremely confident).
• readiness to lose weight. We asked: “Are you seriously considering trying to losing weight within the next 6 months?” and, if yes, “Are you planning to try to lose weight in the next 30 days?”¹³ Those not considering trying to lose weight were staged as precontemplation; those who were considering trying but not planning to try in the next 30 days were staged as contemplation; and those who were planning to try to lose weight in the next 30 days were staged as preparation.

2. Office visit. When patients came in for their appointments, the research assistant gave them consent forms to sign. The assistant then escorted the patient to the examination rooms and started the digital audio recorder. The exams typically took 27 minutes.

Immediately following the exam, the research assistant surveyed the patients. The assistant asked 2 questions we’d asked at baseline: “How confident are you that you can lose weight?” and “Are you seriously considering trying to losing weight within the next 6 months?” (If yes, “Are you planning to try to lose weight in the next 30 days?”) She also made an appointment to conduct a 1-month follow-up telephone survey.

3. One-month follow-up survey. During a follow-up phone survey, we asked patients whether they had attempted to lose weight by changing their diet, exercise patterns, or both. Subsequent to this call, we sent the study participants a $25 check.

Analyzing the patient-physician discussion

Content. Two authors coded 9 topics that physicians and patients discussed that were “weight-related.” Topics included: physical activity, diet, BMI, psychosocial issues, referral to a nutritionist, weight loss surgery, goal setting, weight loss medications, and health care avoidance. We also coded who first brought up the topic.

Time spent. We calculated time spent discussing weight-related topics and also the total time of the patient’s visit.

Motivational Interviewing. Two coders assessed MI. To assess fidelity to MI principles, we used sections of the Motivational Interviewing Treatment Integrity scale (MITI)¹⁴ to rate patient interactions on a scale of 1 (low) to 7 (high) in 2 categories: empathy and MI spirit.

• Empathy is when physicians convey understanding of patients’ perspective.
• MI spirit includes evocation, collaboration, and autonomy. Evocation is when physicians draw out patients’ own reasons for change. Collaboration is when physicians act as partners, supporting and exploring patients’ concerns. Autonomy is when physicians convey that decisions to change lie completely with patients. Inter-rater reliability for the Empathy and MI Spirit was adequate (ICC=.94 and .97, respectively).
• MI-adherent behaviors were those where the physician asked permission to do things, affirmed statements, offered words of support, and emphasized patient control. For instance, the physician might say, “It’s great that you have stopped drinking sweetened tea” or “Whether you lose weight is up to you.”
• MI-nonadherent behaviors were those where the physician advised without asking permission. For example, the physician might say, “Let me tell you what you need to do to make this work…” or “Well, if you want to continue on the way you are, you know your diabetes is only going to get worse.”

These were combined to create a ratio of percentage MI-adherent behaviors by dividing MI-adherent by MI-nonadherent. There was an excellent level of agreement between coders for MI-nonadherent (kappa=.80) and a moderate level of agreement for MI-adherent (kappa=.52) behaviors.

Data analysis

We used Spearman correlations to assess the relationship between our predictors, quantity (time spent and whether weight was addressed) and quality (MI techniques), and mediators of behavior change (readiness to lose weight and self-efficacy to lose weight) and behavior change (attempts to lose weight, change in diet, and change in exercise patterns). We used SAS 9.1 (SAS Institute, Inc, Cary, NC) for all analyses. The study was approved by the Duke University Medical Center Institutional Review Board.

Results

We identified 202 eligible female patients. Of those, 96 had appointments that passed before we could contact them; 11 called the 800 number to refuse. Of the remaining 95 women, we reached 94 by phone. Of those, 19 refused to participate, 46 were ineligible because we had reached the targeted number of women in their weight category, and 4 skipped their appointments. Thus, we audio-recorded 25 encounters (for 14 obese and 11 overweight patients). Of these 25 patients, 24 completed the 1-month follow-up.

Patient demographics. Patients had a mean age of 59 years (standard deviation [SD]=11). Half were white; 42% were college-educated. Forty-two percent reported being in poor to fair health (TABLE 1).

The typical participant was moderately confident and ready to lose weight both before and after their visit. One month after their visit, 63% reported attempting to lose weight. More than half attempted to change their diet (67%); slightly more than half changed their exercise patterns (58%) (TABLE 2).

Physician demographics. Physicians had a mean age of 43 years (SD=10). About half were white; about half were female. No physicians were overweight.

TABLE 1
Characteristics of patients and physicians

TABLE 2
Feeling about weight loss before and after the visit

Patients were more likely to raise the weight issue

Weight-related topics were addressed in 19 of the 25 encounters (11 out of 12 preventive health visits, 8 out of 13 chronic care visits). The mean time spent discussing weight-related topics was 6.9 minutes out of an mean total of 27.0 minutes, or 26% of the total patient-physician time. Weight was more likely to be addressed with obese patients (86%) than with overweight patients (63%).

Patients were more likely than physicians to initiate discussions on weight. Physicians raised weight-related topics 37% of the time. Obese patients were slightly more likely to raise weight-related topics (8 out of 12 times [67%]) than overweight patients (4 out of 7 times [57%]).

The weight-related topics addressed were, in order from most to least frequent: physical activity, diet, BMI, psychosocial issues (eg, motivation to lose weight, triggers for unhealthful eating [such as family cookouts], negative talk [such as telling oneself that losing weight is too hard]), referral to a nutritionist, weight loss surgery, goal setting, health care avoidance, and weight loss medication. When comparing those who attempted to lose weight (n=15) with those who did not (n=9), there was no significant difference in whether or how often a topic was addressed.

Physicians’ empathy scores are moderate

Physicians had a moderate score for Empathy (mean=3.8, standard deviation [SD]=1.5, on 7-point scale), a low score for MI Spirit (mean=2.4, standard deviation [SD]=1.4, on 7-point scale), and displayed fewer MI-adherent behaviors than MI-nonadherent behaviors (mean=0.4, SD=0.3). These means did not differ significantly based on the patients’ weight.

Weight loss conversations linked to patients’ readiness

The discussion of weight-related topics, and the time spent doing so, were related to patients’ readiness to lose weight after their initial examination, when patients’ baseline readiness to lose weight was controlled. The more ready patients were to lose weight after their visit, the more likely they had discussed weight (Spearman’s rank correlation coefficient [r]=.52, P=.01) and spent more time discussing weight (r=.42, P=.05). No other associations were statistically significant (TABLE W1).

Several of the Motivational Interviewing scores predicted patients’ outcomes. When physicians showed more empathy, patients were more likely to report changing their exercise patterns 1 month after the visit (r=.50, P=.02). When physicians displayed more of an MI Spirit, patients were more likely to be ready to lose weight (r=.63, P=.005) and change their exercise patterns (r=.47, P=.04). Further, when physicians used more MI-adherent techniques, patients were more likely to attempt to lose weight (r=.42, P=.08).

Discussion: Good quality discussions lead to change

While more discussion about weight loss led to a greater readiness to lose weight, it was the quality of the discussions that actually led to behavior changes. Most patients had virtually the same levels of readiness to lose weight before and after the visit. It is likely that patients who were ready to lose weight discussed their weight with their physicians—and spent more time discussing it than those patients who were not ready to lose weight.

How patients and physicians discussed weight influenced behavior change. When physicians were more empathic and used techniques consistent with Motivational Interviewing, patients were more likely to report changing their exercise routine and attempting to lose weight.

To date, no one has examined the effect of physicians’ MI techniques on weight-related behavior change in a large study. The low adherence to MI techniques suggests that physicians can improve their counseling skills.

Patients aren’t afraid to talk about their weight

Unexpectedly, patients were more likely than physicians to initiate weight-related discussions. Only one third of the time did physicians raise the topic. Patients appear to be “empowered” to initiate discussions about weight loss. We expected physicians and patients to both be somewhat apprehensive about raising this sensitive topic. However, these findings suggest that overweight and obese patients will initiate the discussion most of the time.

Limitations and strengths of this study

The small sample size limited the analyses. Nonetheless, we found strong correlations in this sample that suggest true relationships that were unlikely to have occurred by chance. Also, we were unable to conduct nested analyses to account for the clustering of patients seen by the same physicians. The results may not generalize to settings outside of academic medical centers and practices in which physicians have less time to spend with patients.

The physicians in this study were not overweight, which could limit the generalizability of the results. Patients may be less likely to raise the topic of weight with physicians who were themselves overweight. In addition, while we assessed single-item outcomes, more objective and extensive standard measures of diet, physical activity, and weight loss would have been optimal.

Some notable strengths of this study were that we used a comprehensive multimodal measurement in assessing both content and style of conversations in addition to patient self-report. We also examined Motivational Interviewing techniques among physicians with little or no MI training; most studies have examined MI among trained counselors only.

How to talk about weight loss: More study is needed

The most commonly addressed weight-related topics were diet and physical activity. However, when looking at the topics that were discussed, we found no patterns between those who attempted to lose weight and those who didn’t. This may mean that because weight loss is such a complex behavior, mention of any aspect of it—be it physical activity, diet, psychosocial issues, and so on—helps patients in their efforts. It also could be that the physician and patient discussed some other aspects in a previous visit; therefore, it was the cumulative effect of many conversations that influenced the patient to change.

These results need to be explored in a larger study to understand whether discussing certain topics is more influential than discussing others in promoting weight loss.

Acknowledgments

We thank Miranda West, Laura Fish, and Mary Sochaki for their work on this project. We are also grateful to the physicians and patients who agreed to have their encounters audio recorded.

Funding

This work was supported by National Cancer Institute grant 2P50 CA68438-06A2. The authors were supported in part by National Cancer Institute grants R01CA089053, R01CA100387, and National Institute of Diabetes and Digestive and Kidney Disorders grant R01DK64986.

CorrespondenceKathryn I. Pollak, PhD, Duke Comprehensive Cancer Center, Cancer Prevention, Detection and Control Research Program, 2424 Erwin Road, Room 6029, Hock Plaza I, Suite 602, Durham, NC 27705; [email protected].

Research has shown that when physicians advise overweight patients to lose weight, improve their diet, or increase their physical activity, patients are more likely to report attempting to do so.^1-3 In a study of 433 primary care patients, 46% reported trying to lose weight after their physician counseled them about nutrition, compared with 37% who were not counseled.¹

The reality, though, is that physicians are not very likely to address weight loss. Data from the Behavioral Risk Factor Surveillance System indicate that patients report their providers address weight loss in fewer than 20% of their examinations.⁴ These low rates are concerning; when physicians do not advise patients to lose weight, patients may believe their weight is not a problem.⁵ Even more worrisome: Physicians are rarely trained on how to counsel patients about weight loss. So, when physicians do counsel patients, it may not be effective.

Using Motivational Interviewing

One effective style of counseling is Motivational Interviewing (MI). MI is a patient-centered, directive counseling style used to help patients explore and resolve their ambivalence related to a particular behavior change (see What is Motivational Interviewing?).^6,7 Researchers have studied the use of MI by counselors and case managers (in handling smoking cessation),^8-11 but not by physicians. Further, no one has examined whether physicians instinctively use MI techniques when discussing weight loss with their patients, or whether MI counseling results in patients trying to lose weight.

The primary aim of this study was to assess how overweight and obese female patients discuss weight loss with their physicians. We also wanted to explore the role that physicians’ way of discussing weight loss—and the use of MI in particular—might play in their patients’ motivation to lose weight.

Methods

Setting and recruitment

All data were collected in a family practice clinic within Duke University Medical Center. We approached 9 physicians in the practice to participate, and all consented. Only 7 physicians had visits with overweight or obese patients and were included in this report. We reviewed their electronic patient appointment schedules twice a week to identify female patients meeting the following criteria: English-speaking, overweight or obese (body mass index [BMI] ≥25 kg/m²), 40 years of age or older, and with health maintenance or chronic care appointments scheduled at least 7 days later. We sent these patients a letter describing the study, and allowed them 7 days to call a toll-free number if they didn’t want to participate.

We took several steps to avoid priming physicians and patients about the purpose of the study. First, both physicians and patients were told the study was about how doctors and patients discuss preventive health topics—they were not told the study was about examining discussions of weight. Second, we surveyed physicians 1 month prior to audio-recording visits, and patients 1 week prior to their visit. Third, we included measures for other preventive health topics (eg, smoking and alcohol) to detract attention from weight.

Gathering data

1. Phone survey before patient visit. We telephoned those patients who did not refuse participation and conducted a baseline survey. We asked about date of birth, race, marital status, level of education, income, weight, height, history of weight loss attempts, and whether this was their first visit with that physician. We categorized women with a BMI ≥25 but <30 as overweight, and those with a BMI ≥30 as obese.¹²

We also assessed each patient’s

• self-efficacy—that is, confidence in their ability to lose weight. We asked: “How confident are you that you can lose weight?” (1=not at all confident, 5=extremely confident).
• readiness to lose weight. We asked: “Are you seriously considering trying to losing weight within the next 6 months?” and, if yes, “Are you planning to try to lose weight in the next 30 days?”¹³ Those not considering trying to lose weight were staged as precontemplation; those who were considering trying but not planning to try in the next 30 days were staged as contemplation; and those who were planning to try to lose weight in the next 30 days were staged as preparation.

2. Office visit. When patients came in for their appointments, the research assistant gave them consent forms to sign. The assistant then escorted the patient to the examination rooms and started the digital audio recorder. The exams typically took 27 minutes.

Immediately following the exam, the research assistant surveyed the patients. The assistant asked 2 questions we’d asked at baseline: “How confident are you that you can lose weight?” and “Are you seriously considering trying to losing weight within the next 6 months?” (If yes, “Are you planning to try to lose weight in the next 30 days?”) She also made an appointment to conduct a 1-month follow-up telephone survey.

3. One-month follow-up survey. During a follow-up phone survey, we asked patients whether they had attempted to lose weight by changing their diet, exercise patterns, or both. Subsequent to this call, we sent the study participants a $25 check.

Analyzing the patient-physician discussion

Content. Two authors coded 9 topics that physicians and patients discussed that were “weight-related.” Topics included: physical activity, diet, BMI, psychosocial issues, referral to a nutritionist, weight loss surgery, goal setting, weight loss medications, and health care avoidance. We also coded who first brought up the topic.

Time spent. We calculated time spent discussing weight-related topics and also the total time of the patient’s visit.

Motivational Interviewing. Two coders assessed MI. To assess fidelity to MI principles, we used sections of the Motivational Interviewing Treatment Integrity scale (MITI)¹⁴ to rate patient interactions on a scale of 1 (low) to 7 (high) in 2 categories: empathy and MI spirit.

• Empathy is when physicians convey understanding of patients’ perspective.
• MI spirit includes evocation, collaboration, and autonomy. Evocation is when physicians draw out patients’ own reasons for change. Collaboration is when physicians act as partners, supporting and exploring patients’ concerns. Autonomy is when physicians convey that decisions to change lie completely with patients. Inter-rater reliability for the Empathy and MI Spirit was adequate (ICC=.94 and .97, respectively).
• MI-adherent behaviors were those where the physician asked permission to do things, affirmed statements, offered words of support, and emphasized patient control. For instance, the physician might say, “It’s great that you have stopped drinking sweetened tea” or “Whether you lose weight is up to you.”
• MI-nonadherent behaviors were those where the physician advised without asking permission. For example, the physician might say, “Let me tell you what you need to do to make this work…” or “Well, if you want to continue on the way you are, you know your diabetes is only going to get worse.”

These were combined to create a ratio of percentage MI-adherent behaviors by dividing MI-adherent by MI-nonadherent. There was an excellent level of agreement between coders for MI-nonadherent (kappa=.80) and a moderate level of agreement for MI-adherent (kappa=.52) behaviors.

Data analysis

We used Spearman correlations to assess the relationship between our predictors, quantity (time spent and whether weight was addressed) and quality (MI techniques), and mediators of behavior change (readiness to lose weight and self-efficacy to lose weight) and behavior change (attempts to lose weight, change in diet, and change in exercise patterns). We used SAS 9.1 (SAS Institute, Inc, Cary, NC) for all analyses. The study was approved by the Duke University Medical Center Institutional Review Board.

Results

We identified 202 eligible female patients. Of those, 96 had appointments that passed before we could contact them; 11 called the 800 number to refuse. Of the remaining 95 women, we reached 94 by phone. Of those, 19 refused to participate, 46 were ineligible because we had reached the targeted number of women in their weight category, and 4 skipped their appointments. Thus, we audio-recorded 25 encounters (for 14 obese and 11 overweight patients). Of these 25 patients, 24 completed the 1-month follow-up.

Patient demographics. Patients had a mean age of 59 years (standard deviation [SD]=11). Half were white; 42% were college-educated. Forty-two percent reported being in poor to fair health (TABLE 1).

The typical participant was moderately confident and ready to lose weight both before and after their visit. One month after their visit, 63% reported attempting to lose weight. More than half attempted to change their diet (67%); slightly more than half changed their exercise patterns (58%) (TABLE 2).

Physician demographics. Physicians had a mean age of 43 years (SD=10). About half were white; about half were female. No physicians were overweight.

TABLE 1
Characteristics of patients and physicians

TABLE 2
Feeling about weight loss before and after the visit

Patients were more likely to raise the weight issue

Weight-related topics were addressed in 19 of the 25 encounters (11 out of 12 preventive health visits, 8 out of 13 chronic care visits). The mean time spent discussing weight-related topics was 6.9 minutes out of an mean total of 27.0 minutes, or 26% of the total patient-physician time. Weight was more likely to be addressed with obese patients (86%) than with overweight patients (63%).

Patients were more likely than physicians to initiate discussions on weight. Physicians raised weight-related topics 37% of the time. Obese patients were slightly more likely to raise weight-related topics (8 out of 12 times [67%]) than overweight patients (4 out of 7 times [57%]).

The weight-related topics addressed were, in order from most to least frequent: physical activity, diet, BMI, psychosocial issues (eg, motivation to lose weight, triggers for unhealthful eating [such as family cookouts], negative talk [such as telling oneself that losing weight is too hard]), referral to a nutritionist, weight loss surgery, goal setting, health care avoidance, and weight loss medication. When comparing those who attempted to lose weight (n=15) with those who did not (n=9), there was no significant difference in whether or how often a topic was addressed.

Physicians’ empathy scores are moderate

Physicians had a moderate score for Empathy (mean=3.8, standard deviation [SD]=1.5, on 7-point scale), a low score for MI Spirit (mean=2.4, standard deviation [SD]=1.4, on 7-point scale), and displayed fewer MI-adherent behaviors than MI-nonadherent behaviors (mean=0.4, SD=0.3). These means did not differ significantly based on the patients’ weight.

Weight loss conversations linked to patients’ readiness

The discussion of weight-related topics, and the time spent doing so, were related to patients’ readiness to lose weight after their initial examination, when patients’ baseline readiness to lose weight was controlled. The more ready patients were to lose weight after their visit, the more likely they had discussed weight (Spearman’s rank correlation coefficient [r]=.52, P=.01) and spent more time discussing weight (r=.42, P=.05). No other associations were statistically significant (TABLE W1).

Several of the Motivational Interviewing scores predicted patients’ outcomes. When physicians showed more empathy, patients were more likely to report changing their exercise patterns 1 month after the visit (r=.50, P=.02). When physicians displayed more of an MI Spirit, patients were more likely to be ready to lose weight (r=.63, P=.005) and change their exercise patterns (r=.47, P=.04). Further, when physicians used more MI-adherent techniques, patients were more likely to attempt to lose weight (r=.42, P=.08).

Discussion: Good quality discussions lead to change

While more discussion about weight loss led to a greater readiness to lose weight, it was the quality of the discussions that actually led to behavior changes. Most patients had virtually the same levels of readiness to lose weight before and after the visit. It is likely that patients who were ready to lose weight discussed their weight with their physicians—and spent more time discussing it than those patients who were not ready to lose weight.

How patients and physicians discussed weight influenced behavior change. When physicians were more empathic and used techniques consistent with Motivational Interviewing, patients were more likely to report changing their exercise routine and attempting to lose weight.

To date, no one has examined the effect of physicians’ MI techniques on weight-related behavior change in a large study. The low adherence to MI techniques suggests that physicians can improve their counseling skills.

Patients aren’t afraid to talk about their weight

Unexpectedly, patients were more likely than physicians to initiate weight-related discussions. Only one third of the time did physicians raise the topic. Patients appear to be “empowered” to initiate discussions about weight loss. We expected physicians and patients to both be somewhat apprehensive about raising this sensitive topic. However, these findings suggest that overweight and obese patients will initiate the discussion most of the time.

Limitations and strengths of this study

The small sample size limited the analyses. Nonetheless, we found strong correlations in this sample that suggest true relationships that were unlikely to have occurred by chance. Also, we were unable to conduct nested analyses to account for the clustering of patients seen by the same physicians. The results may not generalize to settings outside of academic medical centers and practices in which physicians have less time to spend with patients.

The physicians in this study were not overweight, which could limit the generalizability of the results. Patients may be less likely to raise the topic of weight with physicians who were themselves overweight. In addition, while we assessed single-item outcomes, more objective and extensive standard measures of diet, physical activity, and weight loss would have been optimal.

Some notable strengths of this study were that we used a comprehensive multimodal measurement in assessing both content and style of conversations in addition to patient self-report. We also examined Motivational Interviewing techniques among physicians with little or no MI training; most studies have examined MI among trained counselors only.

How to talk about weight loss: More study is needed

The most commonly addressed weight-related topics were diet and physical activity. However, when looking at the topics that were discussed, we found no patterns between those who attempted to lose weight and those who didn’t. This may mean that because weight loss is such a complex behavior, mention of any aspect of it—be it physical activity, diet, psychosocial issues, and so on—helps patients in their efforts. It also could be that the physician and patient discussed some other aspects in a previous visit; therefore, it was the cumulative effect of many conversations that influenced the patient to change.

These results need to be explored in a larger study to understand whether discussing certain topics is more influential than discussing others in promoting weight loss.

Acknowledgments

We thank Miranda West, Laura Fish, and Mary Sochaki for their work on this project. We are also grateful to the physicians and patients who agreed to have their encounters audio recorded.

Funding

This work was supported by National Cancer Institute grant 2P50 CA68438-06A2. The authors were supported in part by National Cancer Institute grants R01CA089053, R01CA100387, and National Institute of Diabetes and Digestive and Kidney Disorders grant R01DK64986.

CorrespondenceKathryn I. Pollak, PhD, Duke Comprehensive Cancer Center, Cancer Prevention, Detection and Control Research Program, 2424 Erwin Road, Room 6029, Hock Plaza I, Suite 602, Durham, NC 27705; [email protected].