Despite growing concern over the prevalence and consequences of physicians violating professional boundaries, lack of a brief and reliable measure of physician knowledge in this area impedes assessment and intervention.
A 25-item exam was developed to assess an array of topics related to professional boundaries in health care such as legal terms, risk factors for boundary violations, and responses to case vignettes requiring clinical judgment in health care situations. To establish its psychometric properties the exam was completed by three different groups: 46 physicians participating in a professional boundaries program (PBP), 42 university-affiliated family physicians, and 36 physicians participating in CME courses on practice improvement.
Psychometric analyses revealed a one-factor structure with good internal consistency and temporal stability. The measure’s sensitivity to change was established by showing that while PBP physicians did not score significantly different from university-affiliated family physicians prior to program participation, they achieved significantly higher scores after participation (p <.05). Mean scores for these two groups were significantly higher than for those physicians completing required ‘practice improvement’ CME coursework (p <.05).
In order to promote the education of physicians across their career in the area of professional boundaries, a reliable measure of such knowledge was established. Identifying risk factors for crossing professional boundaries is of growing importance to maintain the public’s trust.
Despite express prohibition by state medical boards and professional associations, instances of physician sexual misconduct have increasingly been brought to the attention of legislative bodies. The percentage of physicians disciplined for sex-related offenses by state medical boards more than doubled from 1989 to 1996, and sexual misconduct constituted 10 percent of disciplinary actions by the Medical Board of California from October 1995 through April 1997.1,2 Moreover, the immense psychological harm done to patients who are victims of physician sexual misconduct has been documented. As a result of such misconduct, patients have been shown to become psychologically troubled, at increased risk for suicide and may receive substandard medical care in the future because of subsequent inability to trust and communicate with medical professionals.3–5 Despite a demonstrated need for identification of physicians at risk for professional sexual misconduct, further research employing standardized methods is still needed to effectively distinguish physicians who violate professional boundaries from the general physician population. Outcome measures for programs devoted to the remediation of professional boundary violators are absent. In fact, Pope3,6 warns against placing too much confidence in efforts at remediation of professional boundary violators specifically because so few outcome measures exist.
Despite the lack of validated assessment tools of professional boundaries, informative research has been reported on the prevalence of physicians who cross sexual boundaries. An independent survey of physicians drawn from the American Medical Association’s master file of physicians estimate approximately nine percent of physicians report having had a sexual relationship with one or more patients.7 This survey also revealed physicians who cross sexual boundaries are significantly more likely than non-violators to approve of physician-patient sexual contact and to reproach legal limitations on sexual misconduct, though overall 63 percent of physicians in the study deemed physician sexual misconduct “always harmful” to the patient. Despite this prevalence more than half the respondents reported receiving no instruction or training in these issues, with only three percent reporting completion of a continuing education course on this topic. Physicians at risk for sexually-related disciplinary action tend to be male, older than the average physician, established in practice, and are more likely to be obstetrician/gynecologists, family/general practice doctors, or psychiatrists.
Multiple etiological explanations for professional sexual misconduct exist in the literature, most of which address sexual misconduct as part of a character pathology or sexual addiction.4,8,9 For example, Abel and his colleagues have developed a treatment program based on a cognitive-behavioral model of professional sexual misconduct.4,10,11 Their treatment paradigm involves extended cognitive-behavioral therapy, identification of misconduct antecedents, correcting relevant cognitive distortions, building victim empathy, and bibliotherapy. According to this model, sexual misconduct closely resembles paraphilia, and is treated as such. Though no specific outcomes measures are mentioned, the creators of the treatment cite research demonstrating success of cognitive-behavioral therapy for paraphilias as an indicator of their program’s efficacy, though this research is seen as controversial. Additionally, the intrusive “surveillance” portion of the program following remediation would make interpretation of outcome measures of treatment difficult.
A similar perspective has been taken and applied in a continuing education course for physicians who cross professional boundaries. Taught by a panel of addiction specialists, this course focuses on boundary crossing as the result of an addiction, often a sexual one.9,12 This educational model focuses on physicians’ psychology, power imbalances in medical practice, empathy toward the victim, and behavioral antecedents to misconduct. While the professional literature on professional sexual misconduct emphasizes the importance of education in both treating and preventing boundary violations5,9,13–15 , there remains a need to effectively measure knowledge in this area.
Given the growing appreciation for the importance of education and training in this area at all stages of professional development, it appears necessary to develop measures of knowledge and judgment in this area. Once validated measures of knowledge in this area are established, we may better identify gaps in the knowledge of physicians as well as measure the effectiveness of education/intervention programs. Thus we propose to develop a psychometrically sound measure of professional boundary knowledge and judgment, and that this measure will be sensitive to changes in knowledge by participants in a professional boundaries program.
Study participants came from three distinct groups. The first group consisted of 42 physicians who participated in a Professional Boundaries Program (PBP), most of whom were referred by a state medical board as part of a stipulated agreement. This 38-hour CME program assumes most physicians who violate sexual boundaries do not suffer from a sexual addiction, but rather from a lack of knowledge about themselves and the risk factors for boundary violations. This program, conducted in a small group format of four to seven participants, includes a combination of seminars, discussions and action methods to help facilitate integration of learning to effect behavior change. Participants create their own boundary maintenance contract and are followed for one year after completing the three-day course (more information about the program can be obtained from the authors).
The second group consisted of 46 physicians who, as part of disciplinary action by a state medical board, participated in either a CME course on medication prescribing practices or on medical record documentation. No one in this group had been referred to the program for professional boundary violations. The third group consisted of 36 faculty physicians and medical residents in the Division of Family Medicine from a local medical school (none having been disciplined by a state medical board).
The authors created 25 multiple-choice items relevant to knowledge of professional boundaries. The content of these items was drawn from the medical literature on boundaries, AMA ethics guidelines and publications and newly created case vignettes of difficult clinical situations that pertain to maintaining appropriate boundaries. A review of correct response options, clarity, and comprehensibility of items was performed by a group of physicians and faculty. Preliminary testing showed an average completion time of 17 minutes. Sample items are provided in Figure 1.
In addition to completing the multiple-choice exam, all study participants completed an “Intake form” containing items regarding socio-demographic, medical practice, and health behavior variables.
All PBP physicians completed the exam twice — once prior to the course and a second time after completion of the 3-day course. In order to calculate temporal stability of the measures, all comparison family physicians completed the exam twice – between three and seven days apart; this test-retest timeframe is admittedly short but was used to approximate the same learning and practice effects time-frame as that of the PBP physicians. The CME (non-PBP) physicians completed the exam questions only once. The university human subjects committee approved this protocol and the research involved no external funding source.
Standard psychometric analyses of the 25 items were conducted (i.e., principal components analysis, internal consistency) to identify the optimal set of items with the strongest statistical properties. These analyses were followed with comparisons between the three groups of physicians (i.e., ANOVAs were used when dependent variables were continuous with distributions approaching normal), and regression analyses were used to identify factors predictive of exam scores (physician group, gender, practice setting, etc.).
Table 1 shows only socio-demographic variables for the three groups of physicians completing the Professional Boundaries (PB) exam. A one-way ANOVA showed these convenience samples differed significantly in respect to their age and the number of years since graduating medical school, while Chi-Square analysis showed a greater proportion of female physicians in the academic group (all p <.05). These group differences were expected given the academic group included younger medical residents.
Each exam item with a correct response was given one point and a total exam score was calculated by summing all correct items for each participant. Responses from all 113 participants were subjected to a principal component analysis that resulted in a one factor structure. Specifically, the first factor had an Eigenvalue of 3.4 with no other factor having a value greater than 1.0; a scree plot showed a significant ‘elbow’ between the first and second factor, confirming that the measure is a one-factor instrument. Internal reliability was estimated with a Cronbach’s alpha calculated to be .61 for all 25 items combined. However, a final set of 23 items with a standardized alpha of .69 was selected after removal of two items that significantly lowered the alpha. This was considered an acceptable level of internal consistency given the diversity of the construct being assessed (i.e., legal terms related to boundaries, cultural influences on intimate relationships, ‘best practices’ in medicine).
Table 2 shows scores on the 23-item scale for all three groups, and across time for the two groups who completed the exam at two different time points. T-tests revealed no significant differences between residents and faculty physicians, therefore all reported analyses have collapsed these participants into a single ‘academic’ group. First, an ANOVA between groups at Time 1 showed a significant difference (F(2,110)=25.61; p< .001). Planned comparisons revealed no difference between PBP and academic physicians, though the CME doctors did score significantly lower than the other two groups. A 2 x 2 ANOVA comparing the PBP participants to the academic group over time revealed a main effect of time (F(1,62)=26.35; p< .01) and a significant interaction of time by group (F(1,62)=12.25; p< .01). These results suggest both the impact of the PBP on exam scores as well as an initial demonstration of the temporal stability of the measure in the academic group. Specifically, the .66 intra-class correlation for this group between Time 1 and Time 2 suggests an acceptable test-retest reliability.
To follow-up on the differences observed between the CME physicians and the other two groups at Time 1, characteristics of physicians (as reported by them on an intake form) were examined for their impact on exam scores. Table 3 shows variables from the Intake form for all three groups. Post hoc comparisons between groups examining these differences revealed that the academic group was more often in primary care, had lower patient volume though reported working more hours overall per week, and while reporting greater interference of work on their personal lives, reported a lower frequency of relationship distress (e.g., divorce/separated, multiple marriages). The other group difference — on proxy measures of experience and clinical knowledge (e.g., board certified, years of residency) — reflect the fact that approximately half of the academic group was still completing medical residency requirements.
Finally, stepwise regression analyses were performed using variables in Table 3 as predictors of exam scores. Group membership (i.e., PBP, academics, CME) was entered first into the model and accounted for approximately 32 percent of the variance (R=.58, adjusted R2 =.32, p<.01). Of all other variables, only country of medical school attended (i.e., international medical graduate) accounted for significant additional variance (total R=.72, adjusted R2 = .50, p<.01), with IMGs scoring significantly lower.
There are at least four preliminary conclusions that can be drawn from the present data. First, a 23-item exam has been created showing acceptable psychometric properties. Initial analyses reported here suggest the exam has satisfactory internal consistency with a one-factor structure, congruent with the broad nature of the content and construction of items. Temporal stability has been established in that a group of academic family physicians show no significant change in scores over a short period of time. Additional items will be created and then studied in future administrations of the exam, with exploration of whether a multi-factorial structure would provide value to those using this instrument as a measure of continuing medical education.
A second conclusion is the Professional Boundaries Program studied appears to positively impact knowledge of professional boundary issues. Relative to the short-term stability of scores of the group of academic physicians, exam scores for PBP participants significantly increased. Whether this increase is associated with other outcomes that are more clinically relevant (e.g., professionalism as assessed by colleagues and patients) is under current investigation by our program.
A third conclusion is knowledge of professional boundaries issues does not differ significantly between those referred to our program prior to their intervention and a convenience sample of academic family physicians. While our original hypothesis that such a difference would be found (i.e., construct validity), there are at least three explanations of the current results. First, our experience is most of those who participate in the PBP have well above average intellectual abilities and may have advanced exam-taking skills. Thus, scores on the exam may reflect level of general medical knowledge as much as knowledge specifically in professional boundaries. To address this possibility, we have begun to develop more generalizable comparison groups (i.e., non-academic community physicians) and to assess medical knowledge as a potential covariate of exam scores. A second explanation is physicians who begin our program have often been exposed to the issues of professional boundaries through years of legal and regulatory interactions with their lawyer and state medical board. Such exposure to terms and information is likely to increase one’s knowledge relative to other physicians. A third explanation is knowledge of boundary issues may not differentiate those who are accused of sexual misconduct and those not accused. Although the current study has reported on the development and testing of one measure of program effectiveness (i.e., knowledge), based on years of clinical observation and follow-up of participants, there are several risk factors and areas of impact yet to be examined. That is, the PBP is designed to increase empathy for others, improve self-assessment, self-monitoring, and awareness of personal needs, enhance communication skills, and facilitate personal growth. These areas, seen as so important for improving adherence to professional boundaries, have been notoriously difficult to assess. As program facilitators we hear innumerable stories from program participants of unsatisfactory personal relationships, social and professional isolation, and heightened sense of entitlement secondary to chronic feelings of deprivation. We continue to expand and improve our assessment of program participants in these areas and are exploring avenues to collect such information across a longer timeframe. Future reports based on this data will advance the science of this very clinical issue, as well as inform medical education and continuing professional development endeavors.
A fourth finding from the current study points most directly to the need for further research. Exams scores were lowest for those physicians referred for specialized CME courses. It may be these physicians are less intelligent and less skilled at test-taking than the general physician population. In fact, a prior study of physicians (non-PBP participants) undergoing medical competency assessment, demonstrated that among those physicians who identified English as their first language the average I.Q. was 103.16 Although the I.Q. of a comparison group of physicians was not measured concurrently in the study, prior measurements of medical student I.Q. in both the 1940s and 1970s showed an average score of 125.17 In addition, initial analyses on the current dataset suggest physicians trained in medical schools outside the United States score significantly lower than those graduating from U.S. medical schools regardless of whether they have had sexual misconduct charges filed against them. Thus the difference in exam scores between groups may be an issue of language; a few items on the examination are grammatically complex (e.g., double negatives in some answer options). Future work is underway to improve the readability of the items and to expand the comparison groups to include more international medical graduates. This will help assess language barriers as potential factors in the significantly lower scores obtained from these physicians.
There is a lack of outcome measures for assessing continuing education programs that focus on professional boundaries.
A 23-item multiple choice exam is developed that demonstrates internal consistency and temporal stability.
Participants in the Professional Boundaries Program increased scores on an exam of knowledge and decision-making skills in this area.
International medical graduates (IMGs) were more likely to score poorly on the exam; future work is need to determine if this is due to language, quality of medical education or other factors.
Longitudinal collection of outcome data that is clinically-relevant (e.g., physician practice improvement) or related to patient health/satisfaction will provide better measurement of impact from continuing education programs targeting physicians accused of crossing professional boundaries.
Several exam items were case vignettes that posed difficult interpersonal situations in clinical settings, and thus we believe the improvement in scores for those participating in the PBP suggests improvement in attitudes, knowledge, and decision-making skills in complex interpersonal situations. However, the ultimate proof of efficacy of the intervention will be found in the future practice behaviors of PBP physicians. Future work will be most informative in which data is collected from state medical boards and hospital staffs that more directly addresses the issue of recidivism of medical board-disciplined physicians. Additional case vignettes will also enhance the relevance of this measure to assess the general medical community.
Our study has several shortcomings. First, the number of participants recruited for participation was small. We have begun to execute a study that includes both larger numbers of participants and a comparison group of the general “never-disciplined” physician population, as opposed to our unique academic sample. Such a study might allow us to better illuminate some of the causative factors associated with violations of professional boundaries and poor performance on our measure of knowledge.
The test instrument itself is limited. This instrument predominantly measures knowledge, and it may be that knowledge of ethics and professional boundaries has less to do with unprofessional behavior than other factors. In addition, future research should demonstrate temporal stability of greater than three days, perhaps develop a measure with a multi-factorial structure, and one less influenced by language factors. Regardless, it would seem imperative that medical educators have an obligation to be certain that medical students, resident physicians, and faculty have demonstrated competency in this body of knowledge. A future instrument might be more vignette-based, or even use simulated patients, similar to the USMLE Step 2 Clinical-Skills Examination.
Although the conclusions that can be drawn from this pilot study must be considered modest, our data combined with the extant data on this subject in the medical literature suggest that increased focus on continuing education of health professionals in principles of maintaining professional boundaries is warranted. Such offerings should not only teach the legal, ethical, and psychological components of maintaining professional boundaries, but also provide the learner with interactive clinical experiences that improve recognition of risk factors for boundary transgressions and teach healthy behaviors to handle such clinical situations. It is only when clinically-relevant outcomes from these efforts are documented, that the public can trust that the medical profession is demonstrating due diligence in an effort to adequately maintain professional standards.