Clinical decision-making skills are the touchstone of quality medical education and practice. Gathering all pertinent data and integrating this information into a reasoned diagnostic and treatment strategy challenges all astute clinicians.
Many factors influence the clinical decision-making process in medicine. While most would agree that training environment, clinical exposure, and competent preceptors play an important role in the acquisition of clinical decision-making skills, there is also reason to consider the physician’s own personality characteristics and associated tendencies to experience negative moods as another source of significant influence on the decision-making process.
Research has shown, for example, that negative emotions, specifically anger and embarrassment, accompanied by arousal are associated with risk-taking decisions, presumably via effects of higher arousal negative mood states to cause an impairment in self-regulation whereby the individual fails to consider rational cost-benefit calculations related to their actions (1, 2). The implications of these findings for clinical decision-making in medicine are highlighted by findings in a pilot study by Thomas et al. (3), in which physicians’ psychological characteristics were correlated with their responses to a series of clinical vignettes. Physicians with increased levels of hostility and anger were more likely to handle difficult patients through unnecessary medication and/or by refusing to order additional tests in ambiguous clinical situations. This refusal to order additional tests in these vignettes was considered to be a less prudent decision, illustrating how negative emotion may increase the risk-taking behaviors of physicians in the clinical decision-making process, favoring decisions which might place patients at increased health risk.
The findings of Leith and Baumeister (1), that angry persons instructed to think about their decision making showed a marked reduction in risk-taking, suggest that the decision-making process is amenable to improvement by persons expressing these characteristics when they adopt a more introspective perspective. In this case, intervention might be feasible through a process that would identify such predilections on the part of student physicians and provide the necessary psycho-educational experiences to improve decision-making skills.
One such intervention—the LifeSkills Workshop (4)—has been shown in carefully conducted, randomized, controlled trials to reduce hostility, anger, depression and anxiety in patients with coronary heart disease, and to increase positive affect and satisfaction with social support (5, 6). A video designed to deliver the same training modules on a self-administered basis was shown in a randomized controlled trial to reduce perceived stress and anxiety in normal community volunteers who reported high stress levels (7).
Based on the foregoing review, the purpose of this study was twofold: to replicate and confirm the earlier finding (3) that physicians’ personality characteristics associated with hostility, cynicism, anger, and aggression exert an adverse influence on their clinical decision-making skills; and to determine in a real world observational study of second-year medical students whether a cognitive behavior skills training program designed to identify and reduce anger, hostility and cynicism, and improve interpersonal skills will reduce these characteristics, thereby helping student physicians overcome their adverse influence on clinical decision-making skills.
Second year medical students (N=150) at the University of Miami Miller School of Medicine were invited to participate in this study as part of the Behavioral Aspects of Special Populations module. Participation in the research study was voluntary and all questionnaires were completed anonymously. Participants chose an “identifier” whereby the pre- and postquestionnaires could be matched for purposes of data analysis. Participants were included on an analysis-by-analysis basis where all participants who had complete data for each analysis were included in that analysis. The study was approved by the University of Miami Human Subjects Research Office.
Study participants completed the first set of questionnaires (see Assessment Instruments) immediately prior to participating in the LifeSkills program, which was a mandatory component of the curriculum. They completed a second set of questionnaires following their participation in the program. Sixty-nine students completed both sets of questionnaires and were included in the analysis.
The 2-day LifeSkills Program consisted of a group lecture followed by small-group workshops each day. Participants received a videotape covering 10 “LifeSkills” as homework on day 1. Lectures addressed the effects of psychosocial variables on physical health, possible gene/environment mechanisms, and on strategies to reduce health risk. The 10 skills included: separating evidence from interpretations and increasing awareness of feelings and thoughts; evaluating negative thoughts and feelings to decide between deflection and action; deflection skills; action skills, including active problem solving; assertion; saying “no”; and interpersonal skills including speaking effectively, listening well, empathy, and reframing thoughts and feelings (emphasizing the positive). Ninety-minute small group sessions followed the lectures and were led by faculty facilitators who had been trained by Drs. R. Williams and V. Williams (8). During the first small group session, each student reported a specific situation occasioning negative thoughts and feelings. Students then evaluated their thoughts and feelings in that situation, to decide between deflection of the negative thoughts and feelings or action to change the stressful situation. Deflection, action, and communication skills were also addressed on day 2, with continuing emphasis on each student getting hands-on practice in using the skill around a situation that had occurred in his or her own life. During the week following the LifeSkills training, students completed the second set of questionnaires.
As it was not possible to have a true randomized “control” condition during the experimental phase of this study, the same questionnaire protocol was employed the following year with 150 members of the second-year class of Miller School of Medicine. No intervention took place with these students, as it was not offered as part of the curriculum. Students volunteering to participate filled out both sets of questionnaires with a 3-day interlude between sessions. Twenty-three students completed both sets of questionnaires.
Although a concomitant randomized control condition would have been preferable to the 1 year separating the experimental and comparison assessments, this was not feasible due to the structure of the second-year curriculum. However, it must be kept in mind that the principal purpose of this comparison condition was to account for any “test-retest” bias (e.g., practice effect) in the analysis of the pre/posttest results from the experimental condition. Under these circumstances, even the limited number of students who volunteered to take the assessment instruments twice within 1 week would be sufficient to control for such potential bias.
The first set of questionnaires included a demographics questionnaire, clinical vignettes, the Hostility Questionnaire (9), the LifeSkills Questionnaire (10), the Lester Fear of Death Scale (11) and the Center for Epidemiologic Studies Depression Scale (CES-D) (12). The second set of questionnaires—completed within 1 week following LifeSkills training—included the Clinical Vignettes and Hostility Questionnaire.
The clinical vignettes, created by a team of physicians and psychologists, were designed to measure reactions toward dealing with difficult patients, communication regarding life and death issues, and reactions toward patients using nontraditional medical treatment. Ten vignettes identical to those used in a previous study of graduate physicians (3), depicting how physicians handled these difficult clinical cases, were presented to the participants. The students rated their agreement with the physician on a five-point scale from “strongly disagree” to “strongly agree.” The 10 items were then entered into an exploratory factor analysis to determine which items represented a common construct.
The Hostility Questionnaire is a 46-item scale subdivided into cynicism, anger, and aggression subscales. The Cronbach alpha for the overall hostility was moderate [α=0.66]. A total hostility score is derived from the three subscale scores. The LifeSkills Questionnaire is a 30-item scale assessing individuals’ self-rating of ability to use the 10 skills taught in the LifeSkills Program. The Fear of Death Scale is a 15-item scale which measures anxiety related to death and dying. The CES-D is a 20-item scale with four subscales measuring depressed affect, unhappiness, somatic and retardation symptoms, and interpersonal problems. A total depression score is derived from the four subscale scores.
The analysis was conducted using the responses to the Clinical Vignettes of 90 second-year medical students. To ensure that items were significantly correlated, we conducted Bartlett’s test of sphericity, and we used the Kaiser-Meyer-Olkin Measure of Sampling Adequacy to evaluate whether items shared sufficient variance to justify factor extraction. Principal-axis factoring was selected as the method of factor extraction to reveal factors based only on the shared item variance, excluding unique and error variance (13, 14). We used an oblique method of rotation (e.g., oblimin), because we expected any factors to be related. Factors were selected on the basis of eigenvalues (i.e., >1.00), scree plot, and interpretability. Items were considered to load on a factor if the rotated factor loading was greater than or equal to 0.40 on one factor, and were distinguished from another factor by more than 0.10. Examination of sampling adequacy indices indicated an acceptable ratio of interitem correlations to partial correlations (Measure of Sampling Adequacy=0.62), and that enough correlations were available to conduct a factor analysis (Bartlett’s test of sphericity=p<0.01). Four vignettes were not included in the analysis due to communalities less than 0.10.
A two-factor solution was selected on the basis of eigenvalues (both>1.00), the scree plot, and interpretability. The first factor was composed of three vignettes (see Table 1
) and accounted for 34% of the total item variance. A panel of investigators reviewed the vignettes and determined that the three vignettes represent situations where the physician avoided human interaction with the patients. The second factor was composed of two vignettes and accounted for an additional 23% of variance. The panel determined that these vignettes represented placating frustrating patients with unnecessary medications. The responses to each vignette were summed to create composite factor scores. The correlation between these factors was significant (r = 0.30, p<0.01). The baseline descriptive statistics of these factors are presented in Table 1
; higher scores indicate more agreement with the construct.
+
Clinical Judgment Correlates
To determine the correlates of the clinical judgment factors, we assessed their correlations with the Hostility Questionnaire, the Fear of Death scale, the CES-D, and the LifeSkills Questionnaire. The only significant correlations with the avoidance factor were the cynicism scale (r=0.31, p<0.01), the aggression scale (r=0.22, p<0.03), and the total hostility score (r=0.23, p<0.03). All other correlations were not significant (p>0.05). The placating factor was also correlated with cynicism (r=0.22, p<0.04), but not hostility (r=0.04, p<0.73). These findings agree with those of Thomas et al. (3), in which higher levels of hostility and anger were associated with physicians’ preference to handle complaints from difficult patients by prescribing unnecessary medications and refusal to order additional tests.
To examine the impact of the LifeSkills training upon hostility in second-year medical students we compared the mean scores of the Hostility Questionnaire, pre- and postintervention. There were no significant differences at baseline between the LifeSkills and comparison groups on the total hostility score or any of the subscales. Paired t tests comparing the total hostility score and the cynicism, anger and aggression subscales in the LifeSkills group revealed significant decreases in cynicism (t=5.66, df=1, 68, p<0.0001), anger (t=7.56, df=1, 68, p<0.0001), aggression (t=2.55, df=1, 68, p<0.013), and overall hostility (t=7.61, df=1, 68, p<0.0001) (Table 2
). Among the comparison group, there was a decrease only in overall hostility (t=2.09, df=1, 22, p<0.019) (Table 3
).
A repeated measures analysis of variance (ANOVA) was then conducted to compare the change over time for each of the hostility scales between the LifeSkills and comparison conditions. The group by time interaction was significant for the cynicism (F=5.40, df=1, 91, p<0.022), anger (F=6.76, df=1, 91, p<0.011), and total hostility (F=5.01, df=1, 91, p<0.028) scales—indicating larger decreases in the LifeSkills condition. Both groups showed parallel decreases in aggression (group by time interaction (F=0.274, df=1, 91, p<0.602).
There was no difference between the experimental and comparison groups on the mean avoidance or placating score at baseline (avoidance: F=0.063, df=1, 109, p ns; placating: F=0.002, df=1, 109, p ns) (Table 4
). A paired t test was conducted using pre- and postintervention scores for each factor. Among students who received LifeSkills training, there were highly reliable decreases in scores on the avoidance factor (t=8.70, df=1, 81, p<0.0001) and the placating factor (t=5.79, df=1, 81, p<0.0001), while students in the comparison group showed nonsignificant increases in both factors.
A repeated measures ANOVA was conducted to compare change over time for each clinical judgment scale between LifeSkills and comparison groups. The group by time interaction was significant for both the avoidance (F=20.19, df=1, 99, p<0.001), and the placating (F=8.72, df=1, 99, p<0.004) factors. As shown in Table 4
, avoidance and placating factor scores decreased in the students who received LifeSkills training but increased slightly in the comparison group.
Change in hostility was then examined as a potential mediator of the change in clinical judgment following the guidelines of Baron and Kenny (15). Change scores for each of the hostility scales and both clinical judgment scales were calculated by subtracting the baseline score from the post score. These change scores were then correlated with each other. Significant correlations found between change in cynicism and change in avoidance (r=0.36, p<0.01) and between change in total hostility and change in avoidance (r=0.28, p<0.01) were tested for possible mediation effects of LifeSkills training.
The first mediator was change in cynicism with change in avoidance as the dependent variable. There was no full or partial mediation of the relationship between LifeSkills versus comparison condition and change in avoidance when change in cynicism was added as a mediator (Sobel: −1.07, p<0.281). The second mediator was change in overall hostility with change in avoidance as the dependant variable. There was no full or partial mediation of the relationship between LifeSkills versus comparison condition and change in avoidance when change in overall hostility was added as a mediator (Sobel: −0.504, p<0.614).
These findings provide support for both hypotheses advanced earlier. First, higher levels of hostility, particularly its cynicism component, were associated with higher levels of medical student agreement with the avoidance and placating factors derived from the clinical vignettes. Our confidence in the validity of these findings is increased by the fact that they essentially replicate the earlier findings of Thomas et al. (3) with practicing physicians. The findings of both studies are consistent with the notion that personality characteristics related to hostility, cynicism, and anger exert adverse effects on clinical decisions made by medical students and physicians in the context of dealing with “difficult” patients. Such effects have the potential to impair patient well-being and result in less than optimal clinical outcomes.
Communication skills training for medical students has demonstrated significant improvement in patient assessment, communication competence, negotiation skills and shared decision-making, all considered important to positive patient outcomes (16, 17).
The LifeSkills intervention, successful with coronary heart disease patients in randomized controlled trials (5, 6), demonstrated efficacy on similar dimensions with second-year medical students over a 2-day period in the real world setting of a mandatory class inserted into the second-year medical curriculum. Those students who received the LifeSkills training exhibited lower hostility scores following the training as well as lower levels of agreement with the avoidance and placating factors derived from responses to the clinical vignettes. In contrast, students who were tested on two occasions without the LifeSkills training intervention showed no change in avoidance and placating factors and a smaller decrease in hostility. These findings support our second hypothesis that it is possible to adapt a cognitive behavior intervention to ameliorate the effects of negative mood and interpersonal personality traits on physicians’ behavior in dealing with difficult patients. Mediation analysis did not indicate, however, that the decrease in avoidance was mediated by the decrease in hostility or cynicism, raising the possibility that it was the LifeSkills training in interpersonal skills that accounted for the reduced avoidant and placating responses to the clinical vignettes.
This study has several limitations that must be noted. Although we did obtain the same assessments in a comparison group of second-year students who were tested 1 year following the LifeSkills group training, they were not randomly assigned to the LifeSkills and comparison groups. A stronger design, and one that should be evaluated in further research, would be to randomly assign students to receive either LifeSkills training or a credible control condition (e.g., health promotion lectures). It has been shown, however, that well-designed observational studies with either a cohort or case-control design produce estimates of treatment effects that are comparable to those in randomized controlled trials on the same topic (18). It is also likely that, even though participants in observational trials are self-selected, the same can be said for those who agree to participate in randomized, controlled trials and, given the inclusion/exclusion criteria often imposed in randomized, controlled trials, participants in observational trials are often more representative of the real world (19). Given that the LifeSkills program has been shown to produce robust decreases in negative moods in rigorously conducted randomized controlled trials, a case can be made that the present findings of similar benefits in an observational trial among second year medical students makes a case for the real world effectiveness of LifeSkills training in the context of the medical school curriculum.
Another limitation of the current study is that we only evaluated the students at a single time point, immediately after the completion of training. It will be important in future research evaluating this sort of training in medical students to assess additional indicators of clinical performance—e.g., performance in their clinical clerkships—and over longer follow-up periods. Lastly, there was some attrition between the preintervention and follow-up assessments in the LifeSkills condition. This may suggest that those students who completed the follow-up were more highly motivated to change or found the LifeSkills training more useful.
Despite these limitations, the current results are encouraging in suggesting that the deleterious effects of personality traits related to hostility, anger and cynicism on clinical decision making by physicians may be amenable to change by skills training designed to help one cope more effectively with situations that arouse anger and to relate in more positive and supportive ways to others. Besides the benefits that could accrue to patients from such changes, such training could also result in improvements in mood and interpersonal relations that would be beneficial in the personal lives of the physicians themselves.
The current findings make a strong case for undertaking further research, perhaps with a randomized design and certainly with more extensive follow-up. If future research confirms the validity of the current findings, then cognitive behavior skill training such as that offered in the LifeSkills program should be considered for inclusion in the standard medical school curriculum.