A Quantitative Assessment of a 4-year Intervention That Improved Patient Counseling Through Improving Medical Student Health

Abstract

Objective

Despite efforts to produce healthier physicians and patients, there are no published experiments where health promotion interventions throughout medical school have been compared with a control group regarding the school environment, students' personal health practices, and students' patient counseling practices.

Design

Using the Class of 2002 as controls, we performed a 4-year pilot study of a personal health promotion intervention on the Class of 2003 at Emory University School of Medicine (EUSM). We focused on improving the actual and perceived healthfulness of the educational milieu, and on improving their personal and clinical practices about diet, tobacco, exercise, and alcohol use. Data were collected at freshman and ward orientations and during a senior rotation (n_controls = 110, 109, 100 and n_treatment = 114, 104, 106; all response rates greater than 90%).

Results

Students receiving the intervention perceived EUSM as a healthier environment than did control students. By senior year, control males reported twice the tobacco use reported by males in the intervention (43% vs 22%, P = .02), although they had previously reported very similar levels (31% vs 29%, P = .8). Diet, exercise, and tobacco counseling practices were positively related to the intervention; alcohol was inversely related to the intervention.

Conclusions

In this pilot, compared with controls, the intervention positively affected medical students' perceptions of their school health promotion environment, reduced tobacco use among male students and, to some extent, improved their patient counseling practices. Such a medical school-based health promotion intervention shows promise and should be studied in a broader setting.

Introduction

In 1999, we began a 4-year intervention with curricular and extracurricular activities to improve medical students' prevention counseling by improving their personal health practices. This intervention and its qualitative assessment are described elsewhere.[1] This article describes the quantitative outcomes of this program, using survey-based information collected during the treatment and control classes' undergraduate medical educations.

The study's foundation was a scientific literature that supports intuition: healthcare providers preach what they practice. Analyses of physician behavior have shown that physicians who have healthy personal habits are more likely to encourage patients to adopt such habits.[2,3] However, despite the clear possibility that promoting medical student health should therefore be an efficient and effective way to improve patient counseling, no one had previously developed a coordinated curricular and extracurricular program testing that possibility with a control group throughout an undergraduate medical education.

Methods

Study design

We developed and implemented a 4-year intervention to promote healthy behaviors among students in the Class of 2003 at Emory University School of Medicine. This IRB-approved study was a treatment/control group design: Emory's Class of 2003 was the treatment group, and the Class of 2002 was the control group. Data were collected at freshman orientation, ward orientation (beginning of third year), and during senior year in their internal medicine subinternship (n_controls = 110,109,100 and n_treatment = 114, 104, 106). Response rates at all time points were greater than 90%.

The intervention was comprised of curricular and extracurricular components on diet and exercise (more strongly), and also on alcohol and tobacco. The intervention has been more fully described, along with some focus group results of this study, in an earlier article.[4] Examples of curricular components included having students use their de-identified health data to practice biostatistical techniques freshman year, having their sophomore lecture on the pathophysiology of hepatic disease include a new emphasis on students' personal alcohol practices, and having every clinical rotation provide a personal health promotion component. For example, Family Medicine asked students to keep a fruit/vegetable/exercise log for a week, and had them practice counseling each other on their health behaviors, Ethics had students complete an advance directive, Beck Depression Inventory, and CAGE alcohol screener on themselves (with an offer of follow-up, as needed), and Dermatology provided skin cancer prevention information and tubes of sunscreen for students' use. Extracurricular activities included our providing healthy breakfasts before examinations, wine tastings (to demonstrate the use of alcohol for purposes other than intoxication), and healthy, quick cooking classes for students and significant others.

In addition to personal health interventions, we believed that students needed to understand the scientific basis of prevention to be motivated to and have the skills to improve their personal and clinical health habits. Because provision of this basic prevention curriculum was not part of our study question, we gave both treatment and control groups an expanded, more coordinated prevention curriculum than had been provided to prior classes at Emory. In addition to their traditional 12 hour sophomore nutrition course, we helped coordinate 10-15 hours for both control and intervention students of additional lectures and some nondidactic exposures about nutrition, exercise, tobacco, alcohol, heart disease, hypertension, cancer screening, patient counseling, and other topics in prevention, with materials often presented by national leaders in these fields.

Predictors and outcomes

Our primary predictor was the intervention, and the primary outcomes we were concerned with were personal health behaviors (exercise, fruit and vegetable consumption, tobacco use, and immoderate alcohol use), and the frequency of students' clinical prevention practices on 4 topics (diet, exercise, tobacco, and alcohol).

Self-reported personal health attitudes and behaviors were measured using a scannable questionnaire administered and collected at freshman orientation, ward orientation, and at the end of their senior year internal medicine subinternship. Wording of queries concerning tobacco use and immoderate alcohol came from the Centers for Disease Control's (CDC's) Behavioral Risk Factor Surveillance System questionnaire.[5] Exercise was queried as the frequency and duration of moderate and strenuous activities[6]; a dichotomous measure of whether or not a subject met the CDC exercise recommendation was then created (vigorous exercise for at least 20 minutes 3 days a week or at least 30 minutes of moderate exercise 5 days a week).[7] Fruit and vegetable consumption was estimated with a screener that was extensively validated in this population.[8]

Clinical practices were recorded in 4 standardized patient examinations administered during the control and intervention groups' senior year internal medicine rotation. These are fully described in an earlier article.[9] We categorized each encounter's counseling as none, minimal, or more extensive, based on the whether the topic was mentioned (minimal) and if suggestions for improved behavior were discussed (more extensive).

To assess the extent of the intervention's reception, we asked 15 Likert-scaled questions regarding the school's health promotion environment. This was queried at 2 time points (at ward orientation and during senior year). See Table 3 for question wording.

Table 3.

Reported Perceptions Regarding School's Attitudes Toward Alcohol Use, at Orientation to Wards and During Senior Year

Perception	Ward Orientation %	Chi-square P Value	Senior Year %	Chi-square P Value
Reported perception of school's attitude toward alcohol use		.02		.05
Control
No obvious attitude	14		14
Don't drink	1		4
Drinking in moderation is ok	75		76
Drinking is a good release	10		5
Intervention
No obvious attitude	13		21
Don't drink	3		4
Drinking in moderation is ok	83		60
Drinking is a good release	1		14

Statistical analyses

First, we compared control and intervention groups on baseline characteristics, and then on personal health habits and perceptions of school health promotion environment over the course of medical school. Intervention and control groups started medical school with somewhat different habits and opinions; therefore, our main interest was whether exposure to the intervention was associated with differential change. Logistic models using general estimating equations to account for repeated measures were used to assess such change by examining the interaction of group and time. When group differences at a particular time point were of interest, the chi square test was used. All analyses were conducted under the intention-to-treat principle; 13 students that started with the controls but had delayed graduation, therefore receiving some of the intervention, were analyzed as control students.

Examining the intervention's effect on seniors' observed counseling habits during 4 standardized patient encounters, general estimating equations were again used to account for repeated measures on an individual. Logistic models included the extent of observed counseling as the dependent variable (more extensive vs minimal or none) and intervention as the independent variable. Variables that could confound the relationship between group and counseling habit were included in the models: gender, intended specialty during senior year, and a variable measured at freshman orientation that indicated a student's general attitude toward prevention before our study started (agreement with the statement “physicians have a responsibility to promote prevention with their patients”). We chose this statement because it was significantly associated with increased self-reported prevention counseling in our concurrent 16-school observational study.[10]

Results

At baseline, the control and treatment groups were similar in age, gender, and ethnic distribution (Table 1). The baseline control group had significantly fewer single students, was more likely to intend to practice in a primary care specialty, somewhat more likely to agree with the statement that physicians have a responsibility to promote prevention with their patients, and less likely to strongly agree that a physician must personally adhere to a healthy lifestyle to effectively counsel patients to do so.

Table 1.

Baseline Characteristics of the Control and Intervention Groups

Characteristic	Control	Intervention	Chi-square P Value
No. of respondents (% of class)	110 (99%)	114 (99%)
Female, %	38	46	.3
Age in years, mean (range)	23.2 (21, 35)	23.2 (21, 40)	1.0
Ethnicity, %			.6
Asian	9	9
Black	11	9
Hispanic	5	2
White	73	76
Other	2	4
Marital status			.03
Single	81	93
Married	8	3
Unmarried couple	11	4
Intended specialty			.03
Primary care	49	32
Nonprimary care	31	43
Undecided	19	25
“Physicians have a responsibility to promote prevention with their patients,” %			.04
Strongly agree	35	31
Agree	64	61
Neutral/disagree/strongly disagree	1	8
“In order to effectively encourage patient adherence to a healthy lifestyle, a physician must adhere to one him/herself,” %			.009
Strongly agree	21	35
Agree	67	47
Neutral/disagree/strongly disagree	12	19

The treatment group reported substantially and significantly more agreement with nearly all statements related to Emory's health promotion environment, both at ward orientation and during senior year (Table 2), the 2 points at which they were queried on this item. This differential was true for questions concerning school-controlled items such as curricular encouragement of health, emphasis on preventive medicine, and provision of extracurricular activities and healthy food options, as well as classmate encouragement of healthy eating. Classmate encouragement to exercise was similarly high in both groups. In general, agreement with statements regarding school health promotion environment was reported less frequently during senior year among both groups. Treatment and control group students did not significantly differ in their perceptions of Emory's efforts to reduce stress (and this was not 1 of our 4 major interventions). Table 3 shows that control group students became more likely and treatment students less likely to feel that Emory promoted drinking as a good release.

Table 2.

Percent of Medical Students Agreeing With Statements About School Environment at Ward Orientation and During Senior Year, Control vs Intervention Group

Statement	Ward Orientation % agreeing			Senior Year % agreeing
"Our school ...."	Control	Intervention	Chi-square P Value	Control	Intervention	Chi-square P Value
Encourages a healthy life	59	75	.02	35	61	.0002
Curriculum encourages health	42	61	.006	25	36	.1
Curriculum emphasized Preventive medicine	55	69	.03	42	59	.02
Encourages extracurricular activities	25	62	<.0001	29	46	.01
Tries to minimize stress	23	22	.9	12	20	.1
Helps students cope with stress	22	21	.9	18	18	1.0
Events have healthy food options	35	50	.03	25	40	.02
Encourages students' healthy eating	29	53	.0003	13	36	.0001
Classmates encourage each other to eat healthily	24	64	<.0001	26	41	.02
Encourages students to exercise	25	55	<.0001	25	42	.01
Classmates encourage each other to exercise	70	78	.1	67	64	.7
Faculty are good role models for a healthy lifestyle	8	31	<.0001	16	30	.01
Faculty cares about the students	77	77	.9	73	66	.3
“I have a good mentor”	30	51	.002	65	57	.2
“I am glad I chose to train to become a physician”	83	93	.02	83	83	1.0

Original response categories were collapsed: ‘Agreement’ includes Strongly agree and Agree; ‘Does not express agreement’ includes Neither agree nor disagree, Disagree, Strongly disagree

Personal health habits targeted by our intervention are compared in Table 4 . Freshman orientation data for control group students are unavailable for tobacco use, exercise, and fruit and vegetable consumption, so comparisons for these items are limited to entry to wards and senior year. Tobacco use and binge drinking habits differed strongly by gender; we therefore present these habits stratified by gender.

Table 4.

Comparing Control and Intervention Groups on Personal Health Habits Over the Course of Medical School

Personal Health Habit	Freshman Orientation	Entry to Wards	Senior Year	Group Effect P Value	Group-time Interaction P Value
Using Tobacco,1 %
Females				.3	.8
Control	–2	28	23
Intervention	23	18	17
Males				–3	.05
Control	–2	31	43
Intervention	36	29	22
Binge drinking in the past month, %				.6	.5
Females
Control	21	33	23
Intervention	44	38	36
Males
Control	50	55	51
Intervention	43	58	46
Complying with CDC exercise recommendations,4 %				.2	.2
Control	–2	64	50
Intervention	76	71	66
Eating > 4 servings of fruit and vegetables a day, % (median servings/day)				.003	.6
Control	–2	27 (3.0/day)	21 (2.6/day)
Intervention	39 (3.4/day)	42 (3.3/day)	39 (3.6/day)
“In order to effectively encourage patient adherence to a healthy lifestyle, a physician must adhere to one him/herself.”					.2
Control
Strongly agree, %	21	21	11
Agree	67	61	66
Neutral/disagree/strongly disagree	12	18	23
Intervention
Strongly agree, %	35	16	9
Agree	47	54	56
Neutral/disagree/strongly disagree	19	30	35

¹Includes use of cigarettes, cigars, pipe, chewing tobacco, dip, or snuff;

²Comparable data are unavailable for the control group at freshman orientation;

³Significant interaction with time;

⁴Exercising strenuously 3 or more times/week for at least 20 minutes each time, and/or exercising at moderate intensity at least 30 minutes, 5 or more times/week

Control group males increased their use of tobacco products between ward orientation and their senior year, while intervention group males decreased their use (P_group*time = .05). By senior year, control males reported twice the tobacco use reported by males in the intervention (43% vs 22%, P = .02), although at the midpoint they had reported very similar levels (31% vs 29%, P = .8). Females' tobacco habits did not change significantly over the 4 years.

Throughout the study, the prevalence of alcohol binging was not significantly different between the 2 groups (P = .6), nor was there a significant change over time (P = .5). Nonbinge drinking patterns were also similar between groups (data not shown).

The proportion of control group students exercising according to Centers for Disease Control (CDC) guidelines fell from 64% to 50% (ward orientation to senior year), but changed little in the intervention group, from 71% to 66%. However, these trends were not statistically significantly different (P_group*time = .2).

The intervention did not increase students' fruit and vegetable intake during their last 2 years of training, although the intervention group did consistently have higher consumption (P = .003) during their last 2 years of training than did controls. During senior year, the median consumption was 1 serving per day greater in the intervention group, compared to 0.3 servings greater at entry to wards. However, this time-group effect was not statistically significant. The reduced fruit and vegetable consumption over time in the control group of 0.4 servings is the same as that seen in our national natural history study of 16 medical schools.[11]

The intervention group entered with more agreement that physicians' habits affect patient adherence to a healthy lifestyle, but by ward orientation and senior year they were more likely to state disagreement with this than was the control group (P_group-time = .02). The 2 groups did not significantly differ over time on their attitudes about physicians' responsibility to promote prevention.

Students in the intervention group had approximately 50% greater odds of providing extensive counseling on diet (P = .04) and exercise (P = .03) during their standardized patient encounters than did the control students (Table 5).

Table 5.

The Effect of the Intervention on the Frequency of Counseling¹ During Standardized Patient Examinations

Topic	Crude OR (95% CI)	Adjusted OR (95% CI)2	Score test P Value From the Adjusted Model
Diet	1.49 (1.04, 2.12)	1.49 (1.01, 2.19)	.04
Exercise	1.65 (1.16, 2.35)	1.56 (1.04, 2.34)	.03
Tobacco	1.05 (0.54, 1.42)	1.10 (0.77, 1.56)	.6
Alcohol	0.52 (0.32, 0.85)	0.47 (0.27, 0.82)	.006

¹Results are presented as the odds of extensive vs minimal or no counseling on the topic in the intervention group compared to the odds in the control group;

²Adjusted for gender, intended specialty, and agreement (at baseline) with the statement that physicians have a responsibility to promote prevention

The odds of extensively counseling on alcohol in the intervention group was about half that in the control group. The groups did not significantly differ on tobacco counseling.

Conclusions

Students receiving the intervention generally perceived Emory as a healthier environment than did control students, and their prevention-related attitudes, our four target personal health behaviors, and (all but one of) their related patient counseling practices were positively influenced by the intervention (though the associations were not always statistically strong). The two target areas with the most interventions (diet and exercise) showed the most change, and a fifth area that we did not focus on for intervention (stress) did not differ between treatment and control groups.

To our knowledge, this is the first study to use a control group to evaluate the effect of a comprehensive, 4-year health promotion experience on medical students' personal and professional prevention-related attitudes and behaviors. While others have performed interventions on students' personal health practices,[12–14] these interventions have typically been brief (eg, conducted at freshman orientation), have not been compared with a control group, and have not had their implications for longer-term health behaviors or patient consequences studied. Others have studied medical students' health and some of the clinical implications of this, but have not performed interventions.[15,16]

Our most distal outcome, patient counseling, showed a meaningful and statistically significant improvement for diet and exercise (although a small improvement in tobacco counseling, and a significant negative effect for alcohol counseling). Such findings are noteworthy, given our small sample size, the relatively small size of the intervention dose when compared with the entirety of their medical school experience,[4] and the potential (and realized) perils of conducting such a novel intervention.[4] Furthermore, prior studies have shown that senior medical students generally do not consider themselves ready to counsel patients about health issues.[17,18] This is not surprising, as the American Association of Medical Colleges found that 21% to 23% of graduating medical students in 2003-2005 believed that they had received inadequate instruction in risk assessment and counseling.[19] Students' motivation to learn counseling skills may also be low, as counseling skills are typically not a major basis for student assessments.[20] Thus, while we increased the training of both treatment and control groups in patient counseling, a greater dose of such training may need to be delivered to achieve a greater effect. We believe that the counterproductive alcohol counseling and physician role modeling findings were likely the result of the intervention group's resentment (and/or defensiveness), as they told us that they were annoyed because they thought we were pressuring them to abstain entirely from alcohol (this finding is discussed more fully in our article about the medical student focus groups).[4]

Although our use of validated instruments and standardized patients were strengths of our study, it was limited by its small size (and therefore its power to detect change), by potential contamination of the control group by the intervention, and by not having students randomized to treatment (with the controls at baseline having greater interest in prevention-oriented primary care specialties, and difficult-to-interpret differences in their perceptions of the clinical importance of physician lifestyle choices). These 3 factors likely all drove significance levels toward the null. The analysis was hindered by the lack of identical questions at freshman orientation for some of the personal behaviors, and for perceived counseling topic relevance. Lastly, subjects were not blinded to the intervention; this may have influenced reporting of habits and attitudes.

In conclusion, we have shown that a medical student-focused health promotion intervention can improve the reported perception of the healthfulness of the medical school environment, and of both self-reported and objectively measured personal health behaviors in medical students. Corroborating our generally positive results regarding students' directly observed patient-related practices, we have shown previously in a larger, natural history study that students reporting a healthier medical school environment also reported highly significantly better patient counseling practices.[10] This suggests that further refinements and broader implementation of some of the types of interventions we have tried in this experiment could be a novel, efficient, and beneficent method to improve the health of health professionals and the enormous populations of patients they serve.