点击显示 收起
1 From the Departments of Family and Preventive Medicine (EF), Nutrition and Health Sciences (EHW), Epidemiology (GB), and Biostatistics (LKE), Emory University, Atlanta, and the Division of Nutrition and Physical Activity, Centers for Disease Control and Prevention, Atlanta (MKS).
2 Supported by the American Medical Association's Education and Research Foundation, the American Heart Association, an NIH (NHLBI) Institutional National Research Service Award (5T32-HL-07034) and an NCI award, the Emory Medical Care Foundation, and the Ulrich and Ruth Frank Foundation for International Health. 3 Address reprint requests to E Frank, Emory University, 69 Butler Street, Atlanta, GA 30306. E-mail: efrank{at}fpm.eushc.org.
ABSTRACT
Background: The extent to which female physicians personally and clinically adhere to dietary recommendations is unknown and has implications for patients.
Objectives: We aimed to identify US female physicians' personal and professional nutrition- and weight-related habits and to identify which, if any, of their personal habits predicted their clinical practices.
Design: Our sample included the 4501 respondents to the Women Physicians' Health Study, a large, cross-sectional, questionnaire-based study of the health behaviors and counseling practices of US female physicians.
Results: Forty-three percent of physicians performed nutrition counseling, and 50% performed weight counseling with patients at least yearly. Forty-six percent thought that discussing nutrition was highly relevant to their practices, 47% thought the same about discussing weight, and 21% stated that they had received extensive related training. Primary care physicians, obstetricians-gynecologists, pediatricians, vegetarians, and those with a personal history of obesity were more likely to provide nutrition and weight counseling to patients. Female physicians report regularly performing more nutrition and weight counseling than they do most other types of prevention-related counseling. Female physicians report relatively healthy diet-related habits, and these personal habits are related to their likelihood to counsel their patients about nutrition and weight.
Conclusions: Nutrition and weight-related issues are important to female physicians in both their personal and professional lives, and these 2 spheres influence each other.
Key Words: Physicians • women • diet • nutrition • body weight • obesity • counseling • eating disorders • Women Physicians' Health Study • female physicians
INTRODUCTION
The importance of a nutritious diet in maintaining health and preventing disease is well recognized. To promote a balanced diet, the US Preventive Services Task Force (USPSTF) recommends patient nutrition counseling during regular office visits, and both the USPSTF and the National Heart, Lung, and Blood Institute clinical guidelines on weight control recommend that physicians and other health care providers include weight counseling and referral for patients during regular office visits (1, 2). Likewise, increased frequency of nutrition counseling is a goal of not only the USPSTF, but also the US Public Health Service's Healthy People 2010 (3).
Patients look on physicians as good and credible sources of health information (4), and they expect dietary advice and guidance that will help them avoid risk factors and prevent disease (5, 6). Office visits may be good opportunities for health care providers to help patients choose balanced, nutritious diets. Because public awareness of the role of diet in health has led to only modestly improved eating habits and because levels of obesity are rising (7, 8), the need for nutrition counseling is even more urgent.
Although many physicians believe in the importance of nutrition in preventing and treating disease and recognize their role in providing regular nutrition counseling (9), the previously reported prevalences of such counseling are low. Estimates of the percentage of patient consultations that include nutrition or weight counseling range from 14% to 50% (9–13).
This paper compares demographic and practice characteristics among physicians who provide frequent nutrition and weight counseling and those who do not. The participants constitute a large, representative sample of female medical doctors in the United States. The sample includes 4501 respondents to the Women Physicians' Health Study (WPHS), a large, cross-sectional, questionnaire-based study of the health behaviors and counseling practices of US female physicians. In addition, female physicians' personal dietary practices are briefly examined; such practices may be of interest for several reasons. Although past research indicates that physicians' personal health habits are important predictors of their patient counseling practices (14–18), female physicians' health habits have not previously been studied in large enough randomly selected cohorts to allow many meaningful conclusions. In addition, this study examines a wider array of predictor variables than has yet been reported. Furthermore, because there is a positive correlation between high socioeconomic status and many positive health-related behaviors (2, 19, 20), the personal dietary practices of physicians, a group with a very high socioeconomic status, might be considered a benchmark of the diets in the general population. We hypothesize that physicians who themselves engage in dietary and weight-related health-promoting behaviors are more likely to advise their patients on those topics (18).
SUBJECTS AND METHODS
The design of the WPHS and the basic demographics of the WPHS population are more fully described elsewhere (21, 22). The WPHS surveyed a stratified random sample of US female medical doctors. The sampling frame was based on the American Medical Association's (AMA's) Physician Masterfile, a database intended to record all medical doctors residing in the United States and its possessions. A sampling scheme stratified by decade of graduation from medical school was used to randomly select 2500 women from graduating classes in each of 4 decades (1950 through 1989). With this scheme, older female physicians, a population that would otherwise be underrepresented because of the recent increase in numbers of female physicians, were oversampled. The sample included active, part-time, professionally inactive, and retired medical doctors aged 30–70 y who were not in residency training programs in September 1993 when the sampling frame was constructed. Starting in that month, the first of 4 mailings was sent out. Each mailing contained a cover letter and a self-administered, 4-page questionnaire. Responses were accepted until October 1994.
Of the potential respondents, an estimated 23% were ineligible to participate because their addresses were wrong or they were men, deceased, living out of the country, or interns or residents. The WPHS response rate was 59% of the physicians eligible to participate, with a final sample population of 4501. Respondents were compared with nonrespondents in 3 ways: 1) by conducting a phone survey to compare a random sample of 200 nonrespondents with all respondents, 2) by using data from the AMA Physician Masterfile to compare all eligible participants, and 3) by comparing respondents across all 4 survey mailings. These investigations showed that nonrespondents were less likely than respondents to be board certified. However, respondents and nonrespondents did not consistently or substantively differ on other variables including age, ethnicity, marital status, number of children, alcohol consumption, fat intake, exercise, smoking status, hours worked per week, likelihood of being a primary care practitioner, personal income, and whether actively practicing medicine.
On the basis of these findings, the weighting strategy was determined by decade of graduation to adjust for the stratified sampling scheme and by decade-specific response rate and board certification status to adjust for the identified response bias. Weighting the data made possible inferences about the entire population of female physicians who graduated from medical school between 1950 and 1989.
Professional variables were the primary outcome variables, including specialty and patient counseling frequency, relevance to clinical practice, self-confidence in counseling, and training on nutrition and weight counseling (separately queried). In the context of 12 other prevention-related questions, physicians were asked, "Considering your typical patient: How often do you discuss or perform screening?" This question was followed by a list that included "nutrition" and "weight." Response options on frequency included every visit, every <1 y, every >1–2 y, every >2–3 y, every >3–5 y, only at initial visit, only if clinically indicated, and never. The outcome variable for each individual counseling item was dichotomized. Those physicians who counseled at every visit or every <1 y were considered high counselors and are described as "counseling at least once a year" in this paper.
Physicians were asked to note the relevance to their practice of nutrition and weight counseling and their self-confidence in counseling about nutrition and weight. We compared a response of "highly" with the responses "somewhat," "not very," or "not at all." Physicians were also queried about their amount of training in counseling for these dependent variables. We compared a response of "extensive" with the responses "some," "little," or "none." Practitioners in family medicine, general practice, and general internal medicine were grouped as primary care specialists. Obstetricians-gynecologists and pediatricians were examined as 2 separate types of primary care specialist, and practictioners of all other specialties were considered non–primary care specialists. Non–primary care specialists included anesthesiologists, dermatologists, emergency physicians, neurologists, ophthalmologists, psychiatrists, and surgeons. Because of their relatively minimal clinical contact, pathologists, radiologists, and anyone spending <5 h/wk in clinical practice were excluded from counseling analyses. Altogether, 912 practitioners were excluded from the 4501 respondents, leaving a final sample of 3589.
Personal variables queried included alcohol drinking habits, self-perception of being overweight, current priority of changing eating habits and losing weight, personal and family history of obesity (>30% over ideal weight) or eating disorders, and dietary practices. A dietary screener queried, "How many medium-sized servings per day, week, or month do you consume of the following items?" The list of items included 4 categories of fruit and vegetables (fruit juice, other fruit, green salad, and other vegetables). The daily consumption of the 4 was summed as an estimate of fruit and vegetable consumption and the sample median determined.
A WPHS fat score was calculated by using a sum of the number of daily servings times the grams of fat in each serving for 12 items, covering beef and pork consumption (hot dogs, cold cuts, cheeseburger or hamburger or meatloaf, pork, and other beef), dairy products (whole milk, cheese, and butter), margarine, egg yolks, French fries, and sweet breads (doughnuts, cookies, cakes, and pastry). This WPHS fat score is comparable to the 1989 Block fat score (23) except that the WPHS score excludes salad dressing and bread, whereas the Block score includes those 2 items. Physicians were also asked, "Do you consider yourself to be vegetarian?" Nutrition-related habits were classified as good if respondents ate 5 servings of fruit and vegetables/d and less than the median amount of fat.
SUDAAN (24), a program that computes statistics in accordance with the sample design, was used to perform chi-squre, F, and t tests to determine whether counseling was related to personal and professional characteristics. To adjust for multiple comparisons, the conservative cutoff of P < 0.01 was chosen for discussing significant univariate results. The category "unknown" was added to any variable having >2% missing information. This step was taken primarily to avoid losing hundreds of observations during modeling. Logistic regression in SUDAAN was used to model counseling practices as a function of several personal and professional characteristics. Backward selection for logistic regression was used, with goodness-of-fit tests performed on the final models through a modification of the Hosmer and Lemeshow technique (25).
Initial variables offered to the models were personal health status, region of residence, practice type (eg, solo), practice site (eg, hospital), perceived control of work environment, career satisfaction, self-stated vegetarian status, personal and family history of obesity, personal and family history of eating disorders, age (30–49 compared with 50–70 y), ethnicity, specialty, perceived relevance of counseling, self-confidence, training, and hours of continuing medical education in the past month. For the nutrition model, we added current priority of changing eating habits and a variable that measured compliance with eating a diet high in fruit and vegetables combined with a low fat score. For the weight model, we added priority of trying to reduce weight and a variable that measured compliance with exercising more than the median amount combined with a low fat score. To determine which variables remained in the final logistic model, we used as an inclusion criterion a P < 0.10 for the Wald F test (to permit inclusion of possible confounders that might be marginally significant). Among those variables included, we present 95% CIs on the odds ratios for each category (as compared with the referent category) for those that were significant (P < 0.05) by the Wald F test.
RESULTS
The female physicians' average fat score was 24.1, and they ate 3.5 servings of fruit and vegetables per day (median = 3.0; Table 1). About 25% abstained from alcohol. Of the 29.7% who considered themselves overweight, most were trying to lose weight and were doing so by trying to change their exercise and eating habits. Personal histories of eating disorders and obesity were uncommon, as were family histories of eating disorders. Of those reporting a personal history of eating disorders, 29% reported a family history and 87% were trying to change their exercise habits (data not shown). Of the 59.7% of female physicians trying to lose weight, 94% were trying to change their exercise habits and 82% were trying to change their eating habits. Forty-six percent of the respondents thought that discussing nutrition was highly relevant to their practices, 47% thought that discussing weight was highly relevant to their practices; 43% and 46% were highly confident of their abilities to counsel in these respective areas, 21% had received extensive related training, and 39% had received little to no training in weight and nutrition counseling (data not shown).
View this table:
TABLE 1 . US female physicians' personal dietary practices and diet-related history1
Weight
Among these female physicians, 50.2% reported discussing weight with typical patients at least once a year (Table 2). One-third (33.6%) counseled on weight at every visit, 16.6% every <1 y, 3.9% every >1–2 y, 0.2% every >2–3 y, 0.1% every >3–5 y, 1.2% only at the initial visit, and 37.9% only if clinically indicated; 6.6% never counseled on weight (data not shown). Significant univariate predictors of weight counseling (P 0.01) were nonwhite ethnicity, primary care specialty, vegetarian diet, personal history of obesity, belief in the relevance of weight counseling in their practice, self-confidence about counseling, more training, and residence in the US territories. Marginally significant predictors (0.10 > P > 0.01) were being 50 y of age, feeling overweight, trying to lose weight, and trying to change eating habits. In multivariate analyses (Table 3), characteristics associated with frequent weight counseling at significance levels of P < 0.05 were primary care specialty; nonwhite ethnicity; residence in the US territories, East Coast, or east central regions; vegetarian diet; personal history of obesity; more training; self-confidence about counseling; and belief in the relevance of weight counseling. Primary care specialists did not differ significantly from one another in their weight-counseling habits.
View this table:
TABLE 2 . Significant and marginally significant (P < 0.10) correlates of weight or nutrition counseling by US female physicians1
View this table:
TABLE 3 . Predictors of frequent weight counseling among US female physicians, selected through logistic regression modeling1
Nutrition
Among female physicians, 43.4% reported discussing nutrition at least once a year (Table 2). Roughly 23.5% counseled on nutrition at every visit, 20.0% every <1 y, 5.3% every >1–2 y, 0.3% every >2–3 y, 0.2% every >3–5 y, 1.7% only at the initial visit, and 39.9% only if clinically indicated; 9.1% never counseled on nutrition (data not shown). Significant univariate predictors of nutrition counseling (P 0.01) were age 50 y, nonwhite ethnicity, primary care specialty (especially pediatrics), efforts to lose weight, vegetarian diet, personal history of obesity, belief in the relevance of nutrition counseling to their practice, self-confidence, more training, and residence in the US territories. Marginally significant predictors (0.10 > P > 0.01) were attempts to change eating habits and practice site. In multivariate analyses (Table 4), characteristics associated with greater odds of frequent nutrition counseling (P < 0.05) were primary care specialty, age 50 y, residence in the US territories or East Coast, vegetarian diet, more training, self-confidence, and belief in the relevance of weight counseling. After finding that primary care physicians were more frequent counselors, we compared nutrition counseling among the different groups of primary care physicians. We found that pediatricians were significantly more likely to counsel (P < 0.0002) than were the other groups of primary care physicians, but obstetricians-gynecologists were not significantly different (P = 0.29) from the family practice–general practice–internal medicine group (data not shown).
View this table:
TABLE 4 . Predictors of frequent nutrition counseling among US female physicians, selected through logistic regression modeling1
Nutrition and weight
Of those physicians who counseled frequently on nutrition, 90% also counseled frequently on weight; of those who counseled frequently on weight, 79% also counseled frequently on nutrition (data not shown). Nearly all variables that were significant, or marginally significant, for one outcome were also significantly associated with the other. Conversely, many traits affected neither nutrition nor weight counseling: family or personal histories of eating disorders, family history of obesity, and (data not shown) personal health status, amount of perceived work control, career satisfaction, and practice location (rural, suburban, or urban).
DISCUSSION
Of the 12 prevention-related counseling variables queried in the WPHS, nutrition and weight counseling ranked in the top one-half for frequent counseling (43% and 50%, respectively, compared with a median of 39% for all 12). Among these female physicians, about one-half thought that discussing nutrition and weight was highly relevant, and about one-half were self-confident in their abilities to counsel in these 2 areas. Our findings confirm previous diet counseling–related research among physicians (5, 7, 9, 11, 18, 26–28). In fact, they may represent an upper bound for counseling frequency among US physicians; others found that female physicians typically provide more dietary counseling than do men (29). The reasons nutrition and weight counseling are not even higher than 50% may include inadequate training (4 of 10 respondents received little or none), knowledge, teaching materials, and reimbursement, as well as patient noncompliance, low perceived relevance to practice, and low self-confidence or self-efficacy (5, 9, 27, 28, 30).
Primary care physicians, including obstetricians-gynecologists and pediatricians, were significantly more likely to counsel patients regarding both nutrition and weight; non–primary care physicians may have fewer opportunities or may be less interested or less well trained in counseling patients than are primary care physicians.
Physicians who consider themselves overweight reported weight loss or improving eating habits to be a high personal priority. Overweight, weight loss, and changing eating habits were all predictors of nutrition and weight counseling. Similarly, vegetarians were more likely to provide nutrition and weight counseling to patients. These findings suggest that physicians who have intentionally altered their diets, and thus may have a higher personal awareness of diet, are more likely to counsel patients about nutrition and weight. Lewis and colleagues (12, 17, 31) and others (30) reported similar findings. Others also found that physicians who had a greater knowledge of risk factors (32), were better trained in counseling (9), believed in its importance (16), and believed in their personal effectiveness in changing patients' behavior (17, 32) were more likely to counsel and screen their patients. Also associated with counseling for both nutrition and weight were nonwhite ethnicity and region of practice, factors that may be associated with the prevalence of poor diets and obesity among patient populations.
As we reported previously, US female physicians aged 30–70 y ate somewhat less fat and more fruit and vegetables than did 30–70-y-old women not classified as having a high socioeconomic status in the United States, and they ate somewhat fewer fruit and vegetables and modestly more fat than did women of high socioeconomic status (22, 33). Other dietary habits have been explored in previous WPHS analyses (22, 34, 35).
Of the 59.7% of female physicians trying to lose weight, 94% were trying to change their exercise habits, and 82% were trying to change their eating habits, compared with a similar 44% of US adult women trying to lose weight, 82% of whom also reported "eating less," but only 60% of whom reported increasing physical activity (36). A personal history of an eating disorder was reported by 6% of female physicians; prevalence estimates among normal-weight adult women range from 2% to 9%, with higher numbers among overweight women (37, 38).
This study has several limitations. First, the data on personal dietary habits and nutrition counseling practices of female physicians were based on self-reporting. Given the social desirabilities of eating a balanced diet and regularly counseling patients about weight control and nutrition, the responses about counseling may have been overestimates. Future researchers may wish to use alternate methods of data collection that offer fewer threats to internal validity. For example, use of actors portraying standardized patients, chart audits, or direct observation by a third party would objectively verify counseling on behavior modification.
Second, the counseling questions on the WPHS questionnaire did not provide details on what type of counseling was provided, nor did they address counseling on specific diseases or specific nutritional needs, such as diabetes, high cholesterol, or high blood pressure. One cannot ascertain whether the counseling was specified for disease prevention or treatment. Third, the study did not evaluate the referral of patients to nutritionists, registered dietitians, health educators, nurses, or other providers with nutritional expertise who may provide nutrition services in some clinical settings. Some might consider that physicians making such referrals have provided nutritional counseling to such patients.
A fourth limitation is that the findings of this study cannot be generalized to all physicians because the sample population contained only female medical doctors. A fifth limitation is that the WPHS fat score is slightly different from the Block fat score. Finally, this cross-sectional study was designed to be descriptive, and thus no causal inferences can be made.
Future research should evaluate the approaches used by physicians when counseling about diet. Also, it is important to evaluate the circumstances under which nutrition counseling occurs. Most counseling about diet and weight initiated by physicians is disease specific and given in the context of a treatment regimen (27). Physicians should be taught and encouraged to provide general information about eating a balanced and nutritious diet during routine visits and particularly during educable moments. Non–primary care physicians can take advantage of educable moments with patients to reinforce health messages delivered by primary care physicians. Finally, it is important to evaluate the content of the health messages delivered by physicians. Although physicians may have good theoretical understandings of the role of diet in optimizing health (28), many do not know how to translate nutrition information into practical suggestions on choosing foods to promote a balanced and healthful diet (7). Besides the traditional methods of completing a face-to-face consultation or disseminating instructional fact sheets or brochures, innovations such as cooking classes, mock supermarket tours, telephone counseling, e-mail, and interactive CD-ROMs are viable supplements for some physicians and support staff. These nontraditional instructional methods provide an opportunity to reinforce critical health messages.
REFERENCES