点击显示 收起
1 From the School of Medicine (EHS and EF) and the School of Public Health, Biostatistics (LKE and VSH), Emory University, Atlanta, GA, and the Centers for Disease Control and Prevention, Division of Nutrition and Physical Activity, Atlanta, GA (DAG and MKS)
2 The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the CDC. 3 Supported by the American Cancer Society. 4 Address reprint requests to E Frank, Department of Health and Epidemiology, University of British Columbia, 5804 Fairview Avenue, Vancouver, BC, Canada V6T 1Z4. E-mail: efrank{at}emory.edu.
ABSTRACT
Background: Nutrition counseling by physicians can improve patients' dietary behaviors and is affected by physicians' nutrition practices and attitudes, such as the perceived relevance of nutrition counseling.
Objective: The objective was to provide data on medical students' perceived relevance of nutrition counseling, reported frequency of nutrition counseling, and frequency of fruit and vegetable intakes.
Design: Students (n = 2316) at 16 US medical schools were surveyed and tracked at freshmen orientation, at the time of orientation to wards, and in their senior year.
Results: Freshmen students were more likely (72%) to find nutrition counseling highly relevant than were students at the time of ward orientation (61%) or during their senior year (46%; P for trend = 0.0003). Those intending to subspecialize had lower and declining perceptions of counseling relevance (P for trend = 0.0009), whereas the perceived relevance of counseling by primary care specialists remained high (P for trend = 0.5). Students were significantly more likely to find nutrition counseling highly relevant if they were female, consumed more fruit and vegetables, believed in primary prevention, had personal physicians who encouraged disease prevention, or intended to specialize in primary care. Only 19% of students believed that they had been extensively trained in nutrition counseling, and 17% of seniors reported that they frequently counseled their patients about nutrition. Students who consumed more fruit and vegetables, believed that they would be more credible if they ate a healthy diet, were not Asian or white, or intended to specialize in primary care counseled patients about nutrition more frequently. Medical students consumed an average of 3.0 fruit and vegetable servings/d, which declined over time.
Conclusions: The perceived relevance of nutrition counseling by US medical students declined throughout medical school, and students infrequently counseled their patients about nutrition. Interventions may be warranted to improve the professional nutritional practices of medical students.
Key Words: Medical students • diet • fruit and vegetable intakes • nutrition counseling • counseling correlates
INTRODUCTION
Although Americans suffer from increased rates of obesity, hyperlipidemia, and diabetes (1–3), some studies have shown that if physicians advise their patients about nutrition, the incidence of these diseases will decline (4, 5). Despite the potential for counseling to improve dietary practices, <50% of primary care physicians include nutrition or dietary counseling in their patient visits (6–11).
Although many factors affect counseling rates, one of the least explored factors is the observation that physicians' healthy personal practices are positively associated with their clinical prevention-related practices (9, 12–15). Specifically, physicians' healthy dietary practices positively correlate with their clinical nutrition counseling attitudes (16) and practices (5, 10, 14, 17). Some small studies have shown that training interventions may improve both medical students' personal dietary behaviors (18) and their prevention counseling attitudes (18–20). Building on this preliminary association, we implemented the "Healthy Doc-Healthy Patient" (HD-HP) study to describe not only medical students' attitudes and behaviors regarding personal and clinical prevention but also the relation between their personal and clinical practices (21).
Our primary objective was to describe the characteristics of medical students associated with more frequent nutrition counseling and a higher perceived relevance of such counseling. The secondary objective was to describe temporal trends in medical students' perceived relevance of, reported confidence in, and reported training in nutrition counseling. To achieve our primary objective, we assessed the relation between counseling frequency and its predictors in the students' senior year and examined how the relations between the perceived relevance of nutrition counseling and its predictors change during the students' training. Novel predictors of interest include whether the health-promotion efforts (health-promotion score) of medical schools and the fruit and vegetable intakes of medical students are associated with the nutrition counseling attitudes and behaviors of the students. Fruit and vegetable servings per day are the principal dietary component of interest because of their importance in public health recommendations (22, 23).
METHODS
Study design
All medical students in the class of 2003 at 17 participating schools were eligible for participation at each of 3 HD-HP questionnaire sessions: at freshman orientation, at the time of orientation to wards, and in the senior year. Our sample was designed to be similar to all US medical schools in terms of age, region, school size, National Institutes of Health research ranking, private–public school balance, underrepresented minorities, and sex (24–27). Our study received approval from the Institutional Review Board at Emory University.
The HD-HP questionnaires were usually administered after semimandatory activities to encourage participation, but they were anonymous and participation was voluntary. One school was excluded because of protocol nonadherence and a response rate of <55% on the first 2 questionnaires. The final sample comprised 16 schools (Appendix A). Our total response rate, including all respondents from any of the 3 survey administrations and any of the 16 schools, was 80.3%.
View this table:
Appendix A. Characteristics of participating US medical schools
Students were tracked across 3 time points throughout 4 y of medical school by using an anonymous unique identifier that used a student's mother's initials at her birth and father's first 2 initials. Of the 2316 students who provided responses at some time point, 72% (n = 1658) did so at more than one time point, and 42% (n = 970) were tracked across all 3 survey time points.
Variables
Within the HD-HP questionnaire, queries on outcomes of self-reported relevance and frequency of talking to patients about nutrition were asked along with 11 other counseling and 9 screening practices. The questionnaire administered while the students were freshmen contained only one nutrition counseling question: "How relevant do you think talking to patients about nutrition will be in your intended practice?" The response options were "not at all," "somewhat," and "highly." All subsequent questionnaires also contained questions on confidence and training in nutrition counseling. The questionnaire administered to seniors also included the following question: "With a typical general medicine patient, how often do you actually talk to your patients about nutrition?" The response options were "never/rarely," "sometimes," or "usually/always." Because of the relatively small cell sizes for subjects reporting "not at all" relevant and "never/rarely" concerning counseling (<17% for all categories), the less than "highly" relevant and less than "usually/always" counseling responses were collapsed into one category.
Independent predictors were a priori choices based on past literature concerning medical student or physician nutrition counseling behaviors. Both perceived relevance and reported frequency of nutrition counseling were cross-tabulated with the following medical student variables: dietary practices (servings per day of fruit and vegetables, change in fruit and vegetable servings over time, and vegetarianism), demographic characteristics (sex and race-ethnicity), physical health (body mass index, attempted weight loss, and perceived general health), and clinical characteristics (intended specialty and a variable measuring a student's assessment of his or her school's health-promotion score). Counseling outcomes were also cross-tabulated with the level of agreement with several attitudinal questions plus one mentoring question. These questions were as follows: "Primary prevention is the best way to eradicate premature cardiovascular disease (CVD)," "Physicians have a responsibility to promote prevention with their patients," "Patients will adopt a healthier lifestyle if counseled to do so," "I will be able to provide more credible and effective counseling if I eat a healthy diet," and "How much emphasis has your personal physician placed on preventing disease?". The counseling relevance model additionally contained a time variable (freshman orientation, ward orientation, or senior year).
Queries on predictors included 8 fruit and vegetable items (French fries, other potatoes, fruit juice, fruit, vegetable juice, green salad, and vegetable soup, and other vegetables) from a 43-item food-frequency questionnaire within the HD-HP questionnaire (28). Agreement with attitudes was measured on a Likert scale of responses: "strongly agree," "agree," "neither agree nor disagree," "disagree," or "strongly disagree." To evaluate a student's exposure to health promotion and prevention at his or her school, we asked 16 questions about the school's encouragement of a minimization of stress and about both the school's and classmates' encouragement of healthy eating, regular exercise, responsible alcohol use, and discouragement of smoking. We weighted each topic equally when summing the 16 responses to create an individual school health-promotion or "preventive-dose" score, as assessed by one student; this individual student's "preventive dose" assessment of a particular school was the variable of interest. Higher values for this variable indicate a stronger perception of a more preventive, health-promoting school environment (29). At each time point, intended specialties were collapsed into "primary care" (family medicine, general internal medicine, obstetrics/gynecology, pediatrics, and preventive medicine/public health), "subspecialists" (anesthesia/pathology/radiology, emergency medicine, medical subspecialist, surgery, psychiatry, and urology), and "undecided."
Statistics
All analyses were conducted with the use of SUDAAN (30)—a program that accounts for nonindependent observations arising from the clustering of students into schools and the correlated responses from each student over time. The counseling outcomes across time are reported in Table 1, only for the subset of students responding at all 3 time points. The cross-tabulated associations of relevance and frequency with categorical predictors were determined (Table 2) by using the chi-square test; we used a P value 0.01 to test for statistical significance because of the large number of bivariate associations being tested. To help preserve statistical power, we included all observations on the predictors and outcome at all time points.
View this table:
TABLE 1. US medical students' self-reported nutrition counseling behaviors and attitudes throughout medical school: perceived relevance to intended practice, training, confidence, and frequency1
View this table:
TABLE 2. Characteristics of US medical students associated with their perceived relevance and self-reported frequency of clinical nutrition counseling: bivariate associations
Starting with multivariate logistic models that included all the potential predictors listed above, final models were selected (Tables 3 and 4) via backward elimination and stepwise regression methods, leaving only covariates with a significance of P < 0.05. Because clustering limited our models to 14 maximal df, multiple response levels of some variables were collapsed. Meaningful interaction terms were also evaluated (time x intended specialty, fruit and vegetable consumption x time, and fruit and vegetable consumption x preventive dose). At P < 0.01, only time x intended specialty was bivariately significant and hence offered into the multivariate relevance model. Models were examined to confirm modeling assumptions and to assess fit.
View this table:
TABLE 3. Multivariate predictors for US medical students who perceived the relevance of nutrition counseling to be "high" throughout medical school
View this table:
TABLE 4. Multivariate predictors of senior US medical students who self-reported frequent nutrition counseling of typical patients
Multiple imputation was used to overcome the large quantity of information lost as a result of incomplete data on at least one variable (31). Although the median item nonresponse rate was 3% (range: 0–15%), the fruit and vegetable variable was missing for up to 14% of participants, because this summary variable was computed as missing if any of its 8 constituent components were missing. The selected multivariate models were then analyzed with 5 imputed datasets. Variables with missing rates over 4% were imputed by using a nonnormal Bayesian imputation procedure (32) or SAS' PROC MI (33); variance estimates were adjusted for imputation use by the MIANALYZE procedure of SAS. Imputed results were found to be consistent with those from the unimputed models. The results of the imputation procedures were selected to represent the final results of our multivariate analyses.
RESULTS
Our response rates were 87%, 78%, and 75% on the 3 questionnaire administrations. The median age at freshman orientation was 23 y (range: 17–45 y); age was not associated with any of our outcomes (data not shown). Most of the students reported fruit and vegetable intakes of 3 servings/d (median: = 2.7 servings per day), and these intakes declined over time (P = 0.008). The women's consumption decreased from 2.8 to 2.4 servings/d, and the men's consumption decreased from 2.6 to 2.2 servings/d; the women reported higher fruit and vegetable intakes than did the men (P = 0.03) (data not shown).
Overall, 60% of all 970 students responding to all 3 surveys perceived nutrition counseling to be highly relevant in their intended practices (data not shown). Freshman-year students were more likely (72%) to find nutrition counseling highly relevant than were students at the time of ward orientation (61%) or in their senior year (46; overall P for trend = 0.0003) (Table 1). Over time, all students (P = 0.0003) as well as those intending to subspecialize (P = 0.0009) had declining perceptions of counseling relevance, whereas intended primary care specialists' perceived relevance remained high and did not decline (P = 0.5). Only 22% of the students believed that they had been extensively trained in nutrition counseling in their senior year. Although the percentage of students who felt extensively trained increased over time (P = 0.02), the percentage who felt highly confident did not change over time (P = 0.8). A similar percentage of seniors reported "usually/always" (17%) and "never/rarely" (16%) providing nutrition counseling to typical general medicine patients.
With the exception of intended specialty, predictors of counseling relevance were consistent across time and, hence, the bivariate relations are reported overall (Table 2). There was a dose-response for higher perceived nutrition counseling relevance by increasing quintile of fruit and vegetable consumption (P for trend < 0.0001). Students intending to specialize in primary care were much more likely to find nutrition counseling highly relevant (79%) than were those intending to subspecialize (45%; P < 0.0001) and were more likely to counsel (P = 0.006). A school's preventive dose was a significant predictor (P = 0.005) of nutrition counseling frequency only. Both perceived nutrition counseling relevance and frequency were predicted by sex, ethnicity, fruit and vegetable intake, BMI, belief in the efficacy of CVD prevention, belief in increased credibility if a healthy diet was consumed, and intended specialty.
Results from the multivariate model showed that students were more likely to find nutrition counseling highly relevant if they were female (P = 0.005), consumed more fruit and vegetables (P = 0.002), believed that primary prevention was effective against premature CVD (P 0.02), or had personal physicians who had encouraged disease prevention (P = 0.02) (Table 3). The effect of intended specialty was modified by time. For intended subspecialists, perceived counseling relevance was lower than that for intended primary care physicians (regardless of when queried about) and decreased only for subspecialists (odds ratio: 0.26) over time.
Students were more likely to report frequent nutrition counseling if they were black, Hispanic, or other (compared with Asian or white) (P = 0.001), consumed more fruit and vegetables (P = 0.004), believed strongly that they would be more credible if they ate a healthy diet themselves (P = 0.003), or intended to specialize in primary care specialties (P = 0.0007) (Table 4).
DISCUSSION
By senior year, 46% of the students (compared with 72% of freshmen) perceived nutrition counseling to be highly relevant in their intended practices. The substantial decline in perceived nutrition counseling relevance over time was fueled only by those intending to go into a subspecialty. Moreover, an intended primary care specialty consistently predicted both higher perceived nutrition counseling relevance and more frequent counseling. These findings are similar to the findings of 2 studies in which primary care practitioners, relative to subspecialists, counseled their patients more frequently about nutrition (13, 15). However, another physician study did not find this specialty-based difference (34). In our study, medical school training did not maintain or increase the relatively lower percentage of subspecialists that perceived nutrition counseling to be highly relevant. We found that a school's preventive, health promotive emphasis was a weak predictor of nutrition counseling frequency in the bivariate analysis, but not in the multivariate analysis.
In their senior year, <25% of the students believed that they had been extensively trained in nutrition, <50% were highly confident about their nutrition counseling, and <20% usually or always counseled their typical general medicine patients. Regarding training, a 2003 Association of American Medical Colleges survey of students from all US medical schools (n = 13 764) similarly reported that 46% believed that "appropriate time" had been "devoted to nutrition instruction"; 3% believed that the time devoted was "excessive" (35). It is an interesting conundrum that, despite the improvement in the medical students' perception of feeling adequately trained to provide nutrition counseling (P = 0.02), the students' confidence in counseling patients did not improve (P = 0.8).
A broad review of the literature suggests that the nutrition training of many medical students is inadequate; our data suggest that the training the medical students in our study received did not adequately address the requisite skills for real patient encounters. This finding was also suggested in 2 previous surveys. Similar to our findings, a 2002 survey (36) of 290 first-, second-, and third-year medical students and a 1986 survey (37) of 139 third-year medical students reported that some students were lacking in knowledge about dietary recommendations, healthy BMI, CVD risk factors (36), and the ability to deliver nutrition counseling and education (37). This lack of confidence may have important consequences; according to a popular behavioral theory (38), medical students' feelings of self-efficacy would be important in achieving higher rates of nutrition counseling (39). Indeed, the literature suggests that self-efficacy is associated with medical students' personal health successes (40) and with the amount of effort physicians spend on health-promotion strategies (41).
In models, higher fruit and vegetable intakes, at least one positive opinion on the importance of disease prevention, sex, and ethnicity were consistent predictors of both frequent counseling and a high perceived relevance of such counseling. These findings are supported by the literature on physician nutrition counseling. Several studies have reported associations between physicians' healthier diets and increased nutrition counseling of patients (5, 10, 14, 17). Our study is only the third study (10, 14) to report this association between healthier dietary intakes that correspond to national dietary guidelines and the first study to report this association in medical students. As in our study, increased clinical nutrition counseling was previously shown to be associated with physicians having more positive prevention attitudes (42, 43), being female (42, 44), or being black, Hispanic, or other (10). In our study, black physicians counseled their patients about nutrition more frequently than did other ethnic groups, regardless of specialty, and a higher percentage of blacks and Hispanics than of other ethnic groups intended to specialize in primary care. These 2 phenomena were responsible for the increased rates of counseling by blacks and Hispanics.
Although national guidelines at the time of this survey recommended consuming 5 fruit and vegetable servings/d (23), only 11.4% of medical students reported consuming this amount. Furthermore, fruit and vegetable consumption decreased during medical school, perhaps because of normal dietary changes during those years (45) or because of a declining interest in or prioritization of personal prevention as training progressed.
A limitation of this study was that our sample of schools was not randomly selected. This may have caused our conclusions to be less generalizable than those derived from a random sample. Although our self-reported data could have introduced some bias toward overreporting, we validated the frequency of nutrition counseling in our study population via extensive standardized patient (46, 47) testing and found a strong relation (odds ratio: 1.93) between self-report and objective measures (48). Frank et al (29) also found strong correlations between deans' and students' perceptions of their schools' health-promotion environments. Loss to follow-up is a limitation common to longitudinal studies of this type. Another limitation was that we were confined to qualitative descriptions of "talking to patients about nutrition." Therefore, we were unable to describe the quantity, quality, or content of the counseling of the students and could not account for the counseling of patients not perceived as "typical general medicine" patients.
One strength of our study was that we collected data on both personal and professional nutrition behaviors from a sample of 16 medical schools across the United States. Previous studies of medical students' dietary intakes have been limited by sample size (only 1 of 49 was a survey of >300 students) and location (only 1 of 50 was at more than one school). Regarding professional nutrition practices, previous reports have not assessed how the emphasis of various medical schools on preventive medicine affects the counseling behaviors of medical students differently; we are aware of only one study that evaluated the counseling and preventive-nutrition attitudes of medical students at many schools (49). Our study was unique in that it examined the effect of the general promotion of preventive nutrition by numerous medical schools on the association between students' personal and clinical nutritional attitudes and behaviors.
The principal strength of our study was that it was the first, to our knowledge, to provide a natural history of the entire medical school experience. We examined temporal trends in 3 nutrition counseling variables and many potential correlates, including how changes over time in key correlates affected counseling frequency and a change in perceived relevance of counseling. Other than for subspecialist-focused students, there were no different effects by time for predicting perceived nutrition counseling relevance or its observed decrease over time. One conclusion based on this relative absence of a time effect was that most students' nutrition counseling attitudes correlated more strongly with their endogenous attitudes than with the experience of medical school.
Interventions to improve the professional nutrition practices of students can be built on this study's foundation. During medical education training, students were progressively less likely to find nutrition counseling highly relevant in their intended practices. Training interventions in nutrition counseling are warranted. These interventions could be targeted at students interested in subspecialties to enhance their perceptions of the relevance of nutrition counseling in their practices. Because students interested in primary care also expressed modest enthusiasm for clinical nutrition counseling, it may be beneficial to rethink the paradigm of nutrition education, making nutrition more relevant to all disciplines. All students can benefit from more practical experiences with standardized patients and with efforts to bolster their confidence in talking to their patients about nutrition. Future researchers may attempt to quantify the nutrition counseling behaviors of medical students in an effort to improve their nutrition training and thereby the quality of their clinical counseling practices.
ACKNOWLEDGMENTS
EHS conducted the analyses, interpreted the data, and wrote and edited the manuscript. EF designed the study, helped interpret the data, and edited the manuscript. LKE and VSH helped with the statistical analyses. DAG and MKS helped interpret the data and edited the manuscript extensively. None of the authors had any potential conflicts of interest.
REFERENCES