Do Studentsʼ Attitudes during Preclinical Years Predict Their Humanism as Clerkship Students?

There is an increased awareness of the importance of humanism in the medical school curriculum.1 Most of the early work concerned teaching methods2,3,4,5,6 and measurement of humanism.7,8,9,10,11,12,13,14 Contemporary work reinforces the critical role of empathy in humanism15 but also broadens the concept to include other values, qualities, and behaviors: authenticity, compassion, fidelity, integrity, respect, spirituality, and virtue.16,17,18,19,20,21 To distinguish it from humanism, professionalism is characterized as accountability, altruism, commitment to excellence, duty and commitment to service, and honor and respect for others.18 The project we report here began over eight years ago, so compassion, empathy, respect, and considerate biopsychosocial interactions are the conceptual cornerstones of the operational definition we used in this work; we therefore defined the humanistic physician as one who respects the patient's viewpoints and considers his or her opinions when determining health care decisions, attends to the psychological well-being of the patient, regards the patient as a unique individual, treats the patient in the context of his or her family and social and physical environment, possesses good communication and listening skills, engenders trust and confidence, demonstrates warmth and compassion, and is empathetic.22 Despite the considerable attention to humanism in medical education, little is known about predictors of humanism in students. Knowledge about predictors of humanism could foster the evaluation and design of curricular innovations by identifying attitudes that may be affected by educational interventions. The objective of our work was to identify potentially modifiable attitudes that are associated with students' humanistic performance during clinical clerkships. These predictors may be important outcome measures for the curricular interventions, and may help assess the need for additional innovations. Methods Between 1992 and 1995, we had students complete attitude questionnaires during a second-year required preclinical course and again during a required third-year clerkship in family and community medicine. The third-year students also completed a clinical performance examination (CPX) where standardized patients (SPs) rated the students' humanism. We administered four previously developed attitude questionnaires: (1) Physician Belief Scale,23 (2) Physician Reactions to Uncertainty,24 (3) Risk in Clinical Practice,25 and (4) Decision Making Style.26 We had students complete these instruments to provide students with feedback on their attitudes that might affect their clinical behaviors. We gave each student a confidential report with his or her score for each scale, the class's average score and range for each scale, and the normative averages and ranges from the instrument-development samples. The third-year students' report included their second-year scores, so each student could reflect on any changes in attitude scores after experience in clinical rotations. We considered the concepts measured by the instruments to be important to general clinical performance, and not specifically or solely to be predictors of humanism. Similarly, we chose an available measure of humanism that could be completed by SPs to give students feedback on this particular aspect of clinical performance. The Physician Belief Scale is a 32-item questionnaire about ways physicians who adopt a biopsychosocial approach to patient care differ from physicians who do not. Responses are recorded on a five-point Likert-type scale ranging from 1 = disagree to 5 = agree. The scale scores range from 32 (maximum degree of psychosocial orientation) to 160 (minimum psychosocial orientation). The authors of the instrument determined the internal consistency of the scale by Kuder-Richardson formula 20, which measures the extent to which the items reflect a single underlying construct. This unidimensional scale is highly internally consistent, with r = 0.88. The mean score of the 180 family physicians, psychiatrists, and internists in the development sample was 74.3 (SD = 13.7), and that of the 99 family physicians, psychiatrists, internists, and pediatricians in the validation sample was 72.1 (SD = 13.0).23 Eight faculty family physicians in the Department of Family and Community Medicine had a mean score of 58.4 (SD = 10.1). The Physician Reactions to Uncertainty scale measures physicians' affective reactions to uncertainty, which “seem to be a significant, yet overlooked, dimension of patient care decisions and variations in practice patterns.”24 This scale consists of two subscales derived from factor analysis: (1) Stress from Uncertainty subscale (13 items) and (2) Reluctance to Disclose Uncertainty to Others subscale (nine items). Both subscales use a six-point Likert response scale from 1 = strongly agree to 6 = strongly disagree, with many items reverse-scored to prevent response-set bias. The Stress from Uncertainty subscale ranges from 13 to 78 (the higher the score the greater the stress) and has a Cronbach alpha of.90, indicating excellent internal consistency. The mean score for 428 family physicians, general practitioners, general internists, medical subspecialists, and surgeons in the development sample was 44 (SD = 11). The Reluctance to Disclose Uncertainty to Others subscale ranges from 9 to 39 (the higher the score the greater the reluctance to disclose uncertainty) and has a Cronbach alpha of.75, indicating acceptable internal consistency. The mean score for the 428 physicians was 23 (SD = 6).24 The Risk in Clinical Practice questionnaire measures physicians' general self-perceptions of levels of risk aversion/risk seeking and risk attitudes in financial, physical well-being, social, and ethical domains. One nine-point Likert scale item is used for each domain (1 = avoid risk/danger to 9 = seek risk/danger). In a study of laboratory usage, 12 family physicians completed the questionnaire, producing a mean score for each item: gambling for money (3.6, SD = 1.8), physical danger (4.6, SD = 1.4), hurting people's feelings (reverse-scored, 5.3, SD = 1.9), professional norms (reverse-scored, 6.2, SD = 1.7), risks to self (4.7, SD = 1.4), and risks to patients (3.8, SD = 1.6). Significant positive rank-order correlations were observed between: gamble and norms, risk self and gamble, risk self and danger, and risk self and norms.25 In the present study, a student's aggregate risk score was the proportion of the maximum possible score. A measure of internal consistency for this scale is not available. The Decision-making Style is a 32-item questionnaire that measures the extent to which a person is intuitive or analytic in making decisions. Each item presents a forced choice between two alternatives. Scores range from 0 to 32, with low scores indicating analytic approaches and high scores indicating intuitive approaches to decisions and problems.26 A measure of internal consistency for this scale is not available. Standardized patients (SPs) rated students' performances during a CPX on interview style, history taking, physical examination, management negotiation, and patient education. The SPs completed an encounter checklist and a separate humanism questionnaire for each student in each encounter. Interstation exercises assessed the students' differential diagnoses, management plans, and identification of ethical principles. The CPXs were conducted during the second and fifth weeks of the six-week rotation, and each CPX had a minimum of five cases. Students completed between ten and 13 patient cases in the CPX, with ten minutes for the SP activity and five minutes for the interstation activity. The CPX interstation reliability alpha coefficients ranged between 0.43 and 0.56. These moderate scores are comparable to the reliabilities reported in the literature for CPXs that used similar numbers of cases and tested similarly wide varieties of clinical skills, such as those encountered in family medicine.27,28 A student had to complete at least one full CPX with a minimum of five cases to be included in this study, in order to have multiple SP ratings of humanism and hopefully a stable measure for each student. For humanism, the SPs rated each student at each CPX station using the eight-item, abbreviated Humanism Scale developed by Hauck et al.22 The full Humanism Scale is a 24-item questionnaire assessing whether a physician has a “sensitive, non-humiliating, and empathetic way of helping [a patient] deal with some problem or need” and correlates highly with patients' satisfaction with physician-related aspects of care.22 The full scale includes the eight components listed in the first paragraph of this article, with three items each. The eight-item scale has the following items (one per component): This doctor seems to take a personal interest in me. Even when my problem is small, this doctor is concerned. I have confidence in this doctor's decisions. This doctor respects my beliefs. I would talk to this doctor if something were troubling me. This doctor takes an interest in my home life. This doctor is easy to talk to. This doctor seems to know what I am going through when I tell him/her about a problem. In the development study, responses were recorded as an “x” on a line between strongly disagree and strongly agree. The response point was measured with a ruler and converted to a percentage of the total line (1 to 99 for each item). The scale score was the mean for the 24 items. The development sample of 185 patients produced humanism scores ranging from 16 to 99 (1 to 99 is the possible range), with a mean of 75. Cronbach alpha (reliability coefficient) for the 24 items was.95; the eight-item scale with one item per component had a coefficient of.93. We used a seven-point Likert response scale (1 = strongly disagree, 7 = strongly agree) for the eight-item scale, which had a Cronbach alpha of.96. To adjust for variability among SPs on these ratings, we normalized each SP's humanism scores to the average for all SPs. We used SPSS for Windows® for data analysis to produce Pearson correlation coefficients. Results The students' scores for the attitude scales were quite similar to those of the development samples of experienced clinicians for the Physician Belief Scale and the Risk in Clinical Practice Scale. The students appeared to have more stress from uncertainty and reluctance to disclose uncertainty to others than the experienced clinicians in the development sample (Table 1). The standardized patients' ratings of the students' humanism also were quite similar to those of the development sample; when both scale scores are converted to a proportion of the maximum possible score (mean/maximum): development sample 75/99 =.76 and students 42/56 =.75.TABLE 1: Medical Students' Attitudes about Psychosocial Aspects of Care, Uncertainty, Risk, and Decision-making Style—Comparison of Students' Attitude Scores during their Preclinical and Clinical Years with Attitude Scores of Physicians in Instrument-development Samples and Correlation of Students' Attitude Scores with Humanism Ratings by Standardized Patients during a Third-year Clerkship Clinical Performance Examination, Baylor College of Medicine, 1992–1995*The students' preclinical Physician Belief Scale scores were significantly inversely correlated with clerkship humanism (Table 1). Students who rated the psychosocial aspects of medicine lower (higher Physician Belief Scale score) had lower humanism scores. This relationship persisted for students' psychosocial beliefs during the clerkship. Students' preclinical decision-making style was directly related to humanism, with more intuitive students having higher scores on the humanism scales, but this relationship was not stable into the clinical year. The students' preclinical Reluctance to Disclose Uncertainty to Others scores were not significantly related to humanism, but their clinical ratings were inversely related, indicating higher levels of humanism in students less reluctant to disclose their uncertainty. The students' Stress from Uncertainty and their Risk in Clinical Practice scores were not related to humanism. Discussion These students' attitudes about biopsychosocial aspects of care and risk in clinical practice appear similar to those of experienced clinicians, but students may be more stressed by uncertainty and reluctant to disclose it to others than experienced clinicians. The standardized patients' ratings of the students' humanism were similar to the ratings the patients gave practicing physicians. These results lend some support to the legitimacy of using these instruments with students. The students' preclinical attitudes toward the biopsychosocial aspects of medical care are a potential predictor for their humanism on clinical rotations. The Physician Belief Scale was developed as a self-report instrument for practicing physicians, but may be useful as a predictive tool for students. Attitudes about uncertainty, risk, and decision making do not appear to be consistently related to humanism. These results are sensible considering the eight items that compose the Humanism Scale. The consistent relationship with the Physician Belief Scale indicates that students who do not value highly the biopsychosocial aspects of care are not be able to display the interest, concern, and respect necessary for high ratings of humanism by standardized patients. The inconsistent relationship for the Reluctance to Disclose Uncertainty to Others and Decision-making Style suggests that students comfortable with sharing their uncertainty or those preferring an intuitive decision style may be perceived by standardized patients as open about decisions, easy to talk to, or knowing what the standardized patients are going through. The concepts of Stress from Uncertainty and Risk in Clinical Practice aren't as obviously related to the components of humanism and were not associated with humanism in this study. The concepts of uncertainty, risk, and decision making may still be important for clinical performance in general but not for humanism in particular. On the other hand, the concepts inherent in biopsychosocial attitudes seem to be related to clinical performance of the concepts of humanism we measured in this study: compassion, respect, empathy, and especially considerate biopsychosocial interactions. The limitations of this work include questions about both the internal and external validity. It is unclear how the number of cases completed by students could have affected the stability of the humanism measure or influenced the results. The potentially problematic reliability of the Risk in Clinical Practice and Decision-making Style scales could have contributed to the failure to detect relationships between those scales and humanism. Even the significant correlations we did observe are small, and account for little of the variability in humanism, so their statistical significance simply may be due to the sample size. The generalizability of these results is limited, since our study involved only one institution with approximately two class cohorts of students. Studies in other medical schools with additional cohorts of students would determine whether these apparent associations are stable. Further instrument development may improve the ability of attitude measures to predict later humanistic behaviors. The Physician Belief Scale seems to measure well attitudes associated with considerate biopsychosocial interactions. Since contemporary definitions of humanism include so many concepts (compassion, respect, empathy, integrity, fidelity, authenticity, spirituality, and virtue), we may need one comprehensive instrument, or several separate instruments, to measure the attitudes corresponding to the many facets of humanism. Demographic variables that may predict humanism, such as gender, could be included in predictive models, but the fundamental purpose of the work we report here is the identification of potentially modifiable predictors. Including demographic variables may improve the explanatory power of a multivariate model, but will it lead to an admission policy of preferentially selecting students with the unchangeable demographic variables positively associated with humanism? Or should we concentrate our efforts on curricula and the attitudes that we may be able to influence? Curricula in many medical schools have courses emphasizing the physician—patient relationship and problem-based learning, which may influence the students' preclinical attitudes found in our work to be associated with humanism. Future work may reveal relationships between other attitudes, which are conceptually related to humanism, and students' humanistic behaviors in clinical rotations. Measurements of both attitudes and behaviors may be important outcome measures for curricular interventions, and may help assess the need for additional innovations.