tasaht

octsicwgc

tnpcnAcc

phtcpa

v

m

2

Pd

Does Physician Weight Affect Perception of Health Advice?1

Robert B. Hash, M.D.,*,2 Rana K. Munna, M.D.,† Robert L. Vogel, Ph.D.,* and James J. Bason, Ph.D.‡

*Department of Family Medicine, Mercer University School of Medicine, 1550 College Street, Macon, Georgia 31207; †Department of

reventive Medicine 36, 41–44 (2003)oi:10.1006/pmed.2002.1124

Geo, At

hotispCmt

sbtdvutefp

idafsoorpapn

University of Georgia

Background. Obesity is considered a growing healthhreat in the United States. Although physicians haven important role in counseling their patients for obe-ity prevention and treatment, physicians themselvesre often overweight. There are few data regardingow physician body weight might affect patient recep-iveness to obesity counseling.

Methods. A 43-item survey instrument was devel-ped that consisted of three scales related to physicianharacteristics, health locus of control, and percep-ions on receiving health advice from overweight phy-icians. The survey was administered to 226 patientsn five physician offices. Two of the physicians werelassified as obese using BMI calculations, and threeere nonobese. The responses from the surveys wererouped into those from obese and nonobese physi-ians.Results. Significant differences were found for pa-

ient receptiveness to counseling for treatment of ill-ess (P � 0.038) and health advice (P � 0.049), with theatients of nonobese physicians indicating greateronfidence scores. The difference for weight and fit-ess counseling did not reach significance (P � 0.075).nalysis revealed that patient BMI was not a signifi-ant covariate nor were items related to physicianharacteristics in general or health locus of control.Conclusions. Patients seeking care from nonobese

hysicians indicated greater confidence in generalealth counseling and treatment of illness than pa-ients seeing obese physicians. It is not known if thisan be translated into increased success in obesityrevention and treatment. © 2002 American Health Foundation

nd Elsevier Science (USA)

Key Words: obesity; body weight; counseling; pre-ention; health behavior.

fisa

41

hens, Georgia 30601

INTRODUCTION

Despite identification of obesity as a significantealth threat in the United States, the prevalence ofbesity and excess body weight in the general popula-ion continues to increase [1–3]. Physicians are assum-ng an increasingly important role in identifying obe-ity and risk factors for obesity, as well as counselingatients on risk factor reduction and prevention [4,5].ounseling obese and overweight patients on lifestyleodifications is considered the mainstay of obesity

reatment and prevention [6–8].There are few data regarding the prevalence of obe-

ity and overweight status in the physician population,ut the available data indicate it does not differ fromhe general population [9–11]. Likewise, there are fewata regarding how patients perceive health care ad-ice from physicians who appear to be overweight ornhealthy [12]. We have undertaken this study to de-ermine if patients perceive health care advice in gen-ral, and specifically weight management advice, dif-erently when received from obese and nonobesehysicians.

METHODS

A 43-item survey instrument was developed for usen this study. Items were included from previouslyeveloped scales and previous research into beliefsbout the influence of personal behaviors or outsideactors on one’s own health [13–15]. Using past re-earch as a guide was felt to increase the reliability ofur instrument by using measures that have been dem-nstrated to have high internal consistency. Questionsegarding physical characteristics of the group anderceptions of advice from overweight physicians werelso developed. Prior to administration of the survey, ailot study was conducted among 26 adults who wereot participants in the study in order to increase con-

Internal Medicine, Atlanta Medical Center, Atlanta,

1 This study was funded in part by a grant from the U.S. Depart-ent of Health and Human Resources (Grant 1045PE50190-01).2 To whom reprint requests should be addressed. Fax: (478) 301-

045. E-mail: hash_rb@mercer.edu.

rgia 30312; and ‡Institute for Behavioral Research,

dence that the measures utilized showed high con-truct validity. The questionnaire items were factornalyzed to determine whether distinct constructs

0091-7435/03 $35.00© 2002 American Health Foundation and Elsevier Science (USA)

All rights reserved.

emerged that matched key concepts to be measured inthe study. Factor analysis of the pilot data confirmedexpectations that appearance/reputation, physical char-acteristics, and health locus of control items formed dis-tinct factors among the items tested. Two additional con-structs, weight and fitness and treatment for illness, alsowere apparent. By utilizing the above procedures, pre-liminary evidence of survey instrument validity was es-tablished [16].

The final questionnaire consisted of three mainscales of 10 questions each. Scale 1 related to generalphysician characteristics. Scale 2 addressed health lo-cus of control. Scale 3 consisted of 10 items addressingperceptions of advice from overweight physicians. Allitems on the three scales were graded on an ordinalscale from 1 to 7. The remaining 13 items on the surveyaddressed demographic information about the patientand his or her health habits.

Each of the three scales was divided into subscalesthat deal with particular issues. Scale 1 was dividedinto three subscales relating to physician general ap-pearance and reputation (1a), physician physical char-acteristics (1b), and physician likeableness (1c).

Scale 2 was also divided into three subscales. Sub-scale 2a pertains to chance health locus of control.Subscale 2b pertains to an internal health locus ofcontrol. Subscale 2c, a personal health locus of control,comprises items pertaining to physician-oriented is-sues.

Scale 3 was divided into two subscales. Subscale 3areflected patient perception of advice for weight andfitness, and Subscale 3b represented perception on ad-vice for treatment of illness.

Five family physicians in four nonacademic commu-nity office practice settings agreed to participate in thestudy by permitting one author (R.M.) to interviewpatients in the office setting. The physicians agreed toprovide personal height, weight, and body mass indexmeasurements. The physicians also agreed to beblinded to the data and analysis until the completion ofthe study. The physicians were selected on the basis oftheir appearance related to body mass. The physicians’offices were randomly visited during 4 weeks in Janu-

ary 2001. Each office was visited an equal number oftimes.

Patients were approached in the physician’s waitingroom and asked if they would participate in the study.If patients agreed, they were asked to read and sign theinformed consent approved by the Mercer UniversityIRB. Patients were given the option of having assis-tance in completing the questionnaire.

The five physicians were classified as either obese(BMI �30) or nonobese (BMI �30).

Two groups of patients were then determined on thebasis of whether they were visiting an obese or a non-obese physician. Each scale and subscale was exam-ined for reliability with each group with Cronbach’s �.Differences between scale scores and subscale scoreswere analyzed by two-sample t test or Mann–Whitneytest when normality was questionable. Analysis of co-variance was employed to test for differences betweenscale scores with the patient’s BMI used as a covariate.Patient characteristics were analyzed by two-sample ttest, the Mann–Whitney test, �2 test, and Fisher’s ex-act test when appropriate.

RESULTS

The five physicians were divided into two groupsbased on body mass index. The major characteristics ofthe physicians are listed in Table 1A. The obese phy-sicians were older and male.

A total of 238 patients were asked to participate inthe study and 226 completed the survey. The charac-teristics of the patients for the two groups of physiciansare given in Table 1B. Obese physicians’ patientstended to be older (P � 0.014), to be male (P � 0.001),to consume more alcohol (P � 0.0008), and to have ahigher level of education (P � 0.017).

The results of the analysis of covariance failed toindicate that patient BMI was a significant covariatecomparing scale scores for the two physician groups. Asa result, patient BMI was dropped from the model andthe scale scores for the two patient groups were ana-lyzed via two-sample t tests. The results of the analysisare found in Table 2. No significant differences were

TABLE 1A

Characteristics of Physicians in the Study

Physician

A B C D E

Physician ht (m) 1.83 1.75 1.78 1.73 1.73Physician wt (kg) 125.0 102.3 73.6 62.7 63.6Physician BMI (kg/m2) 37 33 23 21 21Physician age (years) 43 38 31 33 31No. of patients surveyed 57 51 42 47 20Physician classification Obese Obese Nonobese Nonobese NonobeseGender M M M F F

42 HASH ET AL.

found between the patients of obese and those of nono-bese physicians for Scales 1 and 2, as well as thecorresponding subscales. Small but significant differ-ences were found on Scale 3 and Subscale 3b, in whichthe nonobese physician group had higher scores (P �0.038 and P � 0.049, respectively.)

Cronbach’s � for the survey subscales as determinedin the validation phase of the study ranged from 0.31 to0.97 (Table 3). The three major scales were examinedfor reliability within the two physician obesity groups.In addition, all subscales were reexamined for reliabil-ity within physician obesity groups. Cronbach’s �ranged between 0.38 and 0.94 for the obese group. Forthe nonobese group, Cronbach’s � ranged between 0.40and 0.95 (Table 3). In general, the Cronbach’s � scoreswere above 0.70 for the study population.

DISCUSSION

Our data support those from Frank et al., who foundthat physician disclosure of healthy habits improvedphysician ability to motivate patients toward healthybehaviors [12]. Our study indicates that there is asmall but significant difference in the way in whichsome patients perceive health advice from obese versusnonobese physicians. This finding is independent ofperceptions of general physician characteristics (Scale1) or health locus of control (Scale 2).Of interest, ourdata also show that the patients in our study havehigher confidence scores for receiving advice fromnonobese physicians (versus obese physicians) for dis-ease treatment (Subscale 3b) than for general weightcontrol and fitness. Although the reason for this is notclear, it may be the result of obtaining the data duringphysician visits rather than from a neutral survey site.

Our study has several limitations. Our resourceslimited the study to a small number of patients andphysicians. The two physicians in the obese group areolder and male, which may confound the results. Thephysicians were from a narrow age range and singleprimary care specialty, which likely introducedpractice-style biases. The gender difference betweenthe two sets of physicians and their patients may havebeen a source of bias. Anderson et al. demonstratedthat physicians may treat weight issues differentlyamong women [17]. Conversely, differences have beenfound in the way that patients choose, interact, andrespond to male vs female physicians [18–20]. Con-founding between age, physician BMI, and gendercould introduce bias. The data were collected from pa-tients in a small geographic area of Georgia, and it isunknown if these findings can be applied to a broaderpopulation. Georgia has been shown to have a highprevalence of obesity and high rate of increase in obe-sity [21], as well as a low self-reported health status[22]. It is possible that patients in Georgia considerobesity in general, and in physicians in particular, as

TABLE 2

Covariate Analysis of Survey Results

Scale

X (SD)Patientsof obese

physicians

X (SD)Patients

of nonobesephysicians P value

Scale 1 39.6 (8.7) 39.8 (10.1) 0.736Subscale 1a 9.4 (3.9) 9.6 (5.2) 0.772Subscale 1b 12.6 (2.6) 12.8 (2.5) 0.536Subscale 1c 12.1 (3.5) 12.7 (3.4) 0.148

Scale 2 40.9 (10.9) 38.1 (9.4) 0.072Subscale 2a 19.3 (5.1) 18.6 (4.7) 0.283Subscale 2b 8.4 (4.0) 8.1 (3.5) 0.554Subscale 2c 13.3 (5.7) 12.5 (5.5) 0.259

Scale 3 26.3 (13.5) 31.7 (16.9) 0.038Subscale 3a 15.7 (7.8) 17.8 (9.2) 0.075Subscale 3b 12.2 (6.9) 14.4 (8.4) 0.049

TABLE 1B

Characteristics of Study Patients Grouped by ObesityClassification of Their Physician

Patientsof obese

physicians

Patientsof nonobesephysicians P value

Total patients 112 114Male 49/110 27/113 0.0011Caucasian 85/111 88/113 0.8166African American 24/111 21/113 0.5704Other 2/111 4/113 0.4235a

Mean weight (kg) 84.6 81.8 0.326Did not complete high schoolb 18/111 16/112 0.6884Completed high schoolb 30/111 45/112 0.034Attend college/tech schoolb 36/111 39/112 0.7057College graduateb 17/111 10/112 0.1423Postgraduateb 10/111 2/112 0.0170a

Considered self overweight 70/111 71/110 0.8186Smoker 29/111 34/113 0.5090Mean height (m) 1.71 1.67 0.004EtOH consumer 36/107 15/114 0.004Age (years) 46.1 40.8 0.014BMI (kg/m2) 29.05 29.24 0.852

a Fisher Method.b Highest level of education completed.

TABLE 3

Survey Validation Scores (�)

Pilot

Patientsof obese

physicians

Patientsof nonobesephysicians

Scale 1 0.74 0.74 0.73Subscale 1a 0.84 0.60 0.67Subscale 1b 0.96 0.77 0.82Subscale 1c 0.72 0.38 0.34

Scale 2 0.58 0.74 0.66Subscale 2a 0.31 0.54 0.40Subscale 2b 0.75 0.65 0.51Subscale 2c 0.68 0.82 0.76

Scale 3 0.96 0.94 0.95Subscale 3a 0.97 0.90 0.93Subscale 3b 0.94 0.90 0.90

43PERCEPTION OF HEALTH ADVICE

an acceptable norm. Although there were no differ-ences in the scales pertaining to physician character-istics in general or mean patient BMI between thegroups, the individual patient’s relationship with orcharacteristics of the physician may be confoundingfactors.

Should physicians be concerned about their personalweight issues when discussing weight managementissues with patients? Loomis et al. found that 93% ofthe military family physicians they surveyed felt theyshould be role models for obesity prevention [23]. It isunclear if this sentiment transfers to the civilian phy-sician workforce. Woolf’s text on disease preventionstates that physicians engaging in weight manage-ment counseling should consider their own weight asan example for their patients [24].

Advice from health care professionals to lose weightand improve diet has been shown to increase the rate ofweight loss effort and diet improvement [25,26]. It hasalso been shown that physicians who practice a healthbehavior are more likely to counsel their patients onthat health behavior [10,27,28] and that physicianstrying to actively improve their personal health habitshad higher rates of patient counseling about healthhabits in general [27]. Frank et al. found that physi-cians’ personal health practices are second only to spe-cialty type as a predictor of preventive counseling [28].

Thus, there is evidence to suggest that personalweight management practices by physicians does re-sult in both higher rates of weight counseling andweight loss efforts. The data from this study suggestthat physician body weight may also be a factor forsome patients. It is not known if this translates intoeffective weight management or decreased prevalenceof obesity in the patient populations.

ACKNOWLEDGMENTS

The authors thank the University of Georgia’s Survey ResearchCenter for their assistance in developing and validating the researchsurvey. The authors thank Jennifer K. Rayhill and Lillian Rice fortheir assistance in the preparation of the manuscript.

REFERENCES

1. Mokdad AH, Serdula MK, Dietz WH, et al. The continuing epi-demic of obesity in the United States. JAMA 2000;284:1650–1.

2. U.S. Department of Health and Human Services. Healthy People2000: national health promotion and disease prevention objec-tives. Washington, DC: Department of Health and Human Ser-vices, 1990.

3. U.S. Department of Health and Human Services. Healthy People2010: understanding and improving health. Washington, DC:Department of Health and Human Services, 1998.

4. Albright CL, Cohen S, Gibbons L, et al. Incorporating physicalactivity advice into primary care. Am J Prev Med 2000;18:225–34.

5. Fontanarosa PB. Patients, physicians, and weight control.JAMA 1999;282:1581–2.

6. Krauss RM, Eckel RH. Howard B, et al. AHA scientific state-ment: AHA dietary guidelines: revision 2000: a statement forhealthcare professionals from the nutrition committee of theAmerican Heart Association. J Nutr 2001;131:132–46.

7. Poston WSC, Foreyt JP. Successful management of the obesepatient. Am Fam Physician 2000;61:3615–22.

8. Anderson D, Wadden TA. Treating the obese patient. Arch FamMed 1999;8:156–67.

9. Bortz WM. Health behaviors and experiences of physicians: re-sults of a survey of Palo Alto medical clinic physicians. WesternJ Med 1992;156:50–1.

10. Wells KB, Lewis CE, Leake B, Ware JE. Do physicians preachwhat they practice? A study of physicians’ health habits andcounseling practices. JAMA 1984;252:2846–8.

11. Glanz K, Fiel SB, Walker LR, Levy MR. Preventive health be-havior of physicians. J Med Educ 1982;57:637–9.

12. Frank E, Breyan J, Elon L. Physician disclosure of healthypersonal behaviors improves credibility and ability to motivate.Arch Fam Med 2000;9:287–90.

13. Wallston KA, Wallston BS, DeVellis R. Development of the mul-tidimensional health locus of control (MHLC) scales. HealthEduc Monogr 1978;6:160–70.

14. Saltzer EB. The weight loss of control (WLOC) scale: a specificmeasure for obesity research. J Pers Assess 1982;46:620–8.

15. Smith MS, Wallston KA, Smith CA. The development and vali-dation of the perceived health competence scale. Health EducRes 1995;10:51–64.

16. Carmines EG, Zeller RA. Reliability and validity assessment. In:Sage university paper series on quantitative applications in thesocial sciences. Newbury Park (CA): Sage, 1979.

17. Anderson C, Peterson CB, Fletcher L, et al. Weight loss andgender: an examination of physician attitudes. Obes Res 2001;9:257–63.

18. Roter DL, Hall JA. How physician gender shapes the communi-cation and evaluation of medical care. Mayo Clin Proc 2001;76:673–6.

19. Hall JA, Roter DL. Medical communication and gender: a sum-mary of research. J Gender Specific Med 1998;1:39–42.

20. Schmittdiel J, Grumbach K, Selby JV, Quesenberry CP. Effect ofphysician and patient gender concordance on patient satisfactionand preventive care services. J Gen Intern Med 2000;15:761–9.

21. Mokdad AH, Serula MK, Dietz WM, et al. The spread of theobesity epidemic in the United States, 1991–1998. JAMA 1999;282:1519–22.

22. Dever GEA, Choi ST, Caldwell MA, Trawick L. Health statusand socio-economic status in Georgia: a quality of life perspec-tive. J Med Assoc GA 2000;Early spring:30–5.

23. Loomis GA, Connolly KP, Clinch CR, Djuric DA. Attitudes andpractices of military family physicians regarding obesity. Mili-tary Med 2001;166:21–5.

24. In: Woolf SH, Jonas S, Lawrence RS, editors. Weight manage-ment: health promotion and disease prevention in clinical prac-tice. Baltimore: Williams & Wilkins, 1996:217–31.

25. Galuska DA, Will JC, Serdula MK, Ford ES. Are health careprofessionals advising obese patients to lose weight? JAMA1999;282:1576–8.

26. Nawaz H, Adams ML, Katz DL. Physician–patient interactionsregarding diet, exercise, and smoking. Prev Med 2000;31:52–7.

27. Lewis CE, Wells KB, Ware J. A model for predicting the coun-seling practices of physicians. J Gen Intern Med 1986;1:14–9.

28. Frank E, Rothenberg R, Lewis C, Belodoff BF. Correlates ofphysicians’ prevention-related practices: findings from thewomen physicians’ health study. Arch Fam Med 2000;9:359–67.

44 HASH ET AL.