Methods

Institutional Review Board approval was obtained to conduct this study with participants in the Kennedy Institute of Ethics' Intensive Bioethics Course. The course included a series of lectures and small group discussions on bioethical principles and theoretical approaches as well as specific topics (table 1). At on-site registration, each participant was asked to complete a survey before the course's opening session. Another version of the survey instrument was distributed at the conclusion of the course.

The pre-test instrument consisted of items regarding demographic characteristics and professional experience, scales measuring tolerance for uncertainty and ambiguity, and two of four clinical vignettes (table 2). The post-test instrument included the same uncertainty scales in addition to the two remaining vignettes, and was linked to the pre-test by a subject number. A randomised pre-test/post-test crossover design was used with the four groups so that each participant received every vignette, but some participants received it before and some after the course. In this way, we were able to control for bias from prior exposure while maintaining our ability to compare pre-test and post-test results in the aggregate. Participants were asked to describe in an open-ended fashion the ethical issues raised by each case as well as to explain how they would resolve the case from the perspective of an ethics committee member.

Data from the professional and demographic surveys were analyzed using SPSS software. A coding sheet was developed to quantify information contained in the open-ended responses, including the use of specific ethical principles and whether the use of the principle was implicit or explicit. This latter distinction was made based on whether the respondent named the principle in identifying issues relevant to the vignette or referred to the principle without using the term for it. For example, an explicit reference to autonomy from the lung tumour case: “If he's competent, tell the diagnosis to the patient based on respect for autonomy” can be contrasted with the implicit reference :“I believe this man has the right to know about his tumour and prognosis”. Two coders independently reviewed a sample of the cases and then met to resolve any discrepancies in coding. Once the coding system needed no further modifications and was consistent among coders, all of the cases were coded and the data were entered into a computerised database.

A coding manual was designed to analyze further the open-ended discussions elicited by the vignettes. Codes were created to identify particular text in the responses indicating respondents' methods of reasoning and proposed resolutions. Characteristics such as designation of power, communication, third party interests, hypothetical reasoning, directive tone, compassion, justification of resolutions, and recognition of competing interests were included in this coding system (table 3). A sample of the responses was coded to establish consistency, after which all responses were coded by a single individual (JM) and entered into the NUD*IST qualitative analysis software program (Qualitative Solutions & Research Pty Ltd, Melbourne, Australia). The resulting data were analyzed to uncover changes occurring between the beginning and end of the course.

Results

respondent characteristics and randomisation

One hundred and nine of the 131 conference participants (83%) returned the pre-test questionnaires. Respondents were fairly evenly distributed across all of the four groups into which conference participants were divided, with 79% responding from group 1, 91% responding from group 2, 85% from group 3, and 79% from group 4. The majority of respondents (55%) were physicians or nurses, with clergy, lawyers, philosophers and others constituting the remainder. Nearly half (49%) of the respondents reported providing direct patient care. Forty-four per cent had prior training in bioethics and 43% had conducted an ethics consultation.

We were able to confirm the successful randomisation of the pre-test respondents among the four groups using demographic and professional survey information. No differences across the groups were found in professional distribution, percentage of professional time spent doing direct patient care, the discussion of ethical issues with patients, or prior experience conducting a formal ethics consultation. Participants' sex, age, and religious affiliation were also evenly distributed. Sixty-six of these 109 participants (61%) returned the post-test questionnaire, and likewise were found to be evenly distributed across the four groups. The characteristics of those returning the post-test instrument did not differ from those participating in the pre-test with the exception of an increased response rate (p<0.03) from those who had conducted an ethics consultation.

responses

Clear differences in pre-test and post-test responses were discernible in three qualitative domains: justification, ranking of elements, and recognition of conflicting elements. Respondents justified their proposed resolutions more frequently after the course and did so in more depth and detail than before the course. Considering the Huntington's test case, representative pre-test responses included: “The MD should not give out the information based on the ethical rule of confidentiality,” and “the insurance company accepts the risk of the patient population as a whole”. In contrast, after the course a respondent wrote: “The ethical issue here is one of confidentiality. The duty of the physician is to the patient and not to the insurance company”. Another defended not releasing this information by claiming that “justice [guarantees the] rights of persons with illness to health care insurance despite expenses to insurers”. Although some of the same justifications for actions were cited before and after the course, responses to the vignettes after the course were more explicative and more frequently incorporated theoretical language.

Compared to justifications before the course that tended to rely upon the unique features of particular clinical vignettes, justifications following the course tended to invoke more generalisable constructs. For example, when presented with the lung tumour case before the course respondents offered: “The patient should know what is happening to his body so that he can make preparations to exit this life”, and “the patient will realise in the near future that he is dying and may become angry if info[rmation] is withheld from him”. Following the course, however, the nature of justification shifted. For example, “[because of] patient autonomy [the] patient is presumed to be competent and have the capacity to make decisions unless proven otherwise;” and “he has a right to know”.

After the course, participants also tended to rank, or weigh, issues and interests more frequently and clearly. In responding to the Huntington's test case before the course, a participant commented: “the physician's obligation to his patient outweighs his societal obligations”. After the course, a respondent stated: “confidentiality is supported by the principle of respect for autonomy. Even if we consider this a prima facie principle, the need for the insurance company to obtain these results does not override . . . the physician's obligation to his patient”.

Similarly, while participants identified many conflicting elements in analyzing the cases both before and after the course, they articulated these conflicts more precisely in post-test responses. Before the course, the “right to privacy v [the] right to know” was stated as an issue relevant to the Huntington's test case. Responses were more refined after the course, as illustrated by the explanation: “for the physician . . . to report information regarding [the] Huntington's gene to the insurance company without the patient's permission would violate his privacy and confidentiality and would be a conflict of the principles of beneficence and non-maleficence”.

Furthermore, the conflicting elements identified by participants tended to be simple before the course, but involved greater complexity after the course. A pre-test response to the lung tumor case states that “the family members' concerns are outweighed by the patient's own rights” which contrasts with the post-test response: “if he's not competent, withhold disclosure based on [the] principle of beneficence overriding [the] principle of autonomy”. The brain haemorrhage vignette given before the course elicited statements identifying a “conflict between the patient's wish and the wife's wish,” and the question of “whether to follow the desires of the individual or his closest proxy,” while after the course a respondent stated that respecting the “self-determination and autonomy of the patient may run counter to the patient's wellbeing or best interest”. Thus, more intricate and nuanced understandings of the vignettes are illustrated by the responses after the course.

Responses in most other domains differed very little from pre-test to post-test. Nevertheless, there was a trend within each vignette towards explicit use of ethical principles compared to implicit use (table 4). Before the course, 61% of the instances in which a participant identified a relevant principle were explicit for the brain haemorrhage case. After the course, 89% of the references to principles in the post-test responses were explicit. A similar trend from implicit use to explicit use was evident in the other three cases.

Experience conducting an ethics consultation may have affected the changes we observed before and after the course. Specifically, there was a trend towards those without prior ethics consultation experience being more likely to mention autonomy explicitly after than before the course (100% versus 67%; p=.09), whereas there was no difference among those who had conducted an ethics consultation. This pattern also held true for the principle of beneficence.

Discussion

The two greatest strengths of our study are also inherent weaknesses. Because the randomisation among the four groups was successful, the crossover study design allows us to see an accurate picture of the effects of this intensive bioethics course on its participants as a group. However, we are limited to interpreting these effects in the aggregate because no participant was given the same case twice. This prevents us from being able to examine any changes on an individual level.

The qualitative nature of our analysis is also an important strength because we are able to look at subtleties invisible to numeric data and thereby gain a more thorough understanding of the effects of this educational intervention. However, this type of qualitative analysis is necessarily subjective, requiring the differentiation between shades of grey rather than the reporting of more stark quantitative data. In addition, the process of analysis is a time-consuming endeavour, while describing these findings is itself a challenging task.

Adjustments

In retrospect, some adjustments might have been made to expand our findings. For example, the inclusion of additional details about the scenarios might have enabled respondents to analyze more thoroughly the issues involved. Developing a way to encourage longer, more discursive responses, perhaps through interviews, would enrich the results. Finally, a follow-up instrument could be designed to analyze long term effects of the course.

In interpreting our results, it is important to consider that these health care professionals are not unfamiliar with bioethical issues. Most had had experiences in clinical care or bioethics before taking the course, presumably making them familiar with some of the basic concepts in bioethics and therefore affecting what they learned from the course. In addition, a self-selection bias must be considered because an interest in bioethics most likely prompted the participants to register for the course. Similarly, a response bias may have affected our results in the post-test responses as a disproportionate number of participants who had conducted ethics consultations returned the surveys after the course. Finally, the course format and topics likely influenced our findings, making it reasonable to expect that a different course format and/or set of topics would yield other results.

Nevertheless, the three qualitative domains (justification, ranking and conflicting) in which we found noticeable changes following the course shed light on how respondents analyzed the vignettes and explained their resolutions. The trend towards more frequent, thorough, complex, and precise explanations of reasoning and defence of their analyses suggests that what these respondents took away from this intensive course in bioethics was an enhanced capacity to support better the resolutions they proposed. Similarly, they seemed to derive an improved ability to articulate their thoughts on relevant issues.

The systematic review of ethical theories offered in the course may have provided a contextual framework through which participants learned to express themselves. Lectures and discussions probably familiarised participants with a language that they might not have known before attending the course, the language of contemporary bioethics. In addition, these professionals may have come to understand better the tensions and conflicts involved with the cases they face and therefore have been able to explain why the issues they recognise are important.

Because of the uniqueness of this study as an evaluation of an intensive course intended for a group of health care professionals, our data expand current understanding of bioethics education. The published literature reports that medical ethics courses increase undergraduate medical students' ethical sensitivity^{1, 2, 14} and moral reasoning ability^{3–6, 10} as well as improve clinical competence and confidence surrounding ethical issues in house officers.^{15, 16} Studies have also found significant increases in ethical sensitivity in undergraduate students resulting from lectures and discussion workshops.^{1, 14} Together, these studies suggest that bioethics courses and workshops have a positive effect on the ability of medical students to identify issues, reason morally and act ethically. Perhaps students just beginning to be exposed to bioethical questions gain these abilities from courses in the discipline while experienced professionals instead gain the ability to defend more carefully their reasoning and to articulate more clearly the relevant issues.

While our methods seem too cumbersome to use as an evaluation tool for individual courses, our evaluation is important since it enhances our understanding of the effect of bioethics education on health care professionals. This more complete picture should help educators to improve subsequent educational efforts.