Objective: To determine under what conditions lay people and health professionals find it acceptable for a physician to breach confidentiality to protect the wife of a patient with a sexually transmitted disease (STD).
Methods: In a study in France, breaching confidentiality in 48 scenarios were accepted by 144 lay people, 10 psychologists and 7 physicians. The scenarios were all possible combinations of five factors: severity of the disease (severe, lethal); time taken to discuss this with (little time, much time); intent to inform the spouse about the disease (none, one of these days, immediately); intent to adopt protective behaviours (no intent, intent); and decision to consult an expert in STDs (yes, no), 2×2×3×2×2. The importance and interactions of each factor were determined, at the group level, by performing analyses of variance and constructing graphs.
Results: The concept of breaching confidentiality to protect a wife from her husband’s STD was favoured much more by lay people and psychologists than by physicians (mean ratings 11.76, 9.28 and 2.90, respectively, on a scale of 0–22). The patient’s stated intentions to protect his wife and to inform her of the disease had the greatest impact on acceptability. A cluster analysis showed groups of lay participants who found breaching confidentiality “always acceptable” (n = 14), “depending on the many circumstances” (n = 87), requiring “consultation with an expert” (n = 30) and “never acceptable (n = 13)”.
Conclusions: Most people in France are influenced by situational factors when deciding if a physician should breach confidentiality to protect the spouse of a patient infected with STD.
- STD, sexually transmitted disease
Statistics from Altmetric.com
Confidentiality is the ethical cornerstone of good treatment and is essential for establishing trust between clinicians and patients. Without such trust, patients may not disclose all pertinent information, especially about unsafe behaviours (such as substance misuse or risky sexual practices). Without complete disclosure, clinicians may not be able to make rapid and accurate diagnoses, perform useful tests, order effective treatments and arrange for appropriate follow-up. In addition, trust is needed to achieve patient–clinician relationships that may themselves be therapeutic. Through its link to trust, therefore, confidentiality plays a key role in the healing process.
The importance of confidentiality is recognised in codes of medical conduct from Hippocrates to the present. Confidentiality, however, has its limits. When clinicians have good reason to suspect that their patients’ behaviours will put other people at risk, they face a moral dilemma: should they maintain confidentiality, or should they break it to try to protect the other people (whether by warning them or by alerting the authorities)? The answers given in the laws, judicial rulings and medical codes of ethics differ between the US and the UK, on the one hand, and France, on the other.
In the US and the UK, it is legitimate to breach confidentiality in some cases to protect other people at risk. In the widely commented on Tarasoff case, the Supreme Court of California asserted in 1974, and again in 1976, that once a doctor or psychotherapist determines that his or her patient intends to murder or commit serious harm to an identifiable person, the clinician has a duty to take reasonable measures to inform or protect the intended victim.1 The Tarasoff mandate was reaffirmed in judicial rulings around the country2,3 (except in Texas4). The American Medical Association’s Code of Medical Ethics (section E-5.05, last updated June 1994) states: “The obligation to safeguard patient confidences is subject to certain exceptions which are ethically and legally justified because of overriding social considerations. Where a patient threatens to inflict serious bodily harm to another person or to himself or herself and there is a reasonable probability that the patient may carry out the threat, the physician should take reasonable precautions for the protection of the intended victim, including notification of law enforcement authorities.”5 Similarly, in the UK, the General Medical Council’s ethical guidance about confidentiality (issued in April 1994) asserts (as principle 27): “Disclosure of personal information without consent may be justified in the public interest where failure to do so may expose the patient or others to risk of death or serious harm.”6 The BMA echoed this position in its policy statement of October 1999.7
In France, in contrast, the emphasis is more on preserving the confidentiality of the patient. The Ordre des Médecins, the body responsible for maintaining the ethical and professional integrity of French physicians, declared in Article 4 of the Code of Medical Ethics (as revised in 1995): “Professional confidentiality (le secret professionnel), instituted in patients’ interest, is obligatory for every physician within the conditions established by law. Confidentiality applies to everything the physician learns in the exercise of his profession, that is to say not only what has been confided to him, but also what he has seen, heard or understood.”8 No exception was made to protect other people. This position was affirmed by the parliament9 and upheld by French law. Article 226-13 of the New Penal Code (2002) stated: “The revelation of confidential information (information à caractère secret) by a person who possesses it either by profession or by reason of a function or of a temporary mission is punished by one year of prison and a fine of 15 000 Euros.”10
The issue of patient confidentiality is particularly sensitive and emotionally charged when it entails the possible transmission of a sexually transmitted disease (STD) to a patient’s spouse and more so, when the threat is infection with HIV. The issue of confidentiality merges with the fear of STDs, particularly of HIV-AIDS, and with the increasing concern about spousal abuse, in medical circles as well as among the public. The confidentiality of a patient’s infection with HIV is protected less in the US and the UK than in France. The American Psychiatric Association, for example, declared in 1988 that when a physician learns that a patient is infected by HIV and is certain the patient’s behaviour puts other people at risk of infection, the physician must inform the identifiable people who may be exposed to this risk.11 The BMA repeated in 1999 the General Medical Council’s 1997 advice that doctors “may disclose information to a known sexual contact of a patient with HIV where you have reason to think the patient has not informed that person, and cannot be persuaded to do so.”7 In Australia, two general practitioners were condemned by the Supreme Court of New South Wales in 2003 “for breach of contract and for negligence for failing to ensure that a man who tested positive for HIV told his wife about the results.”12 In contrast, in France, the Ordre des Médecins stated in 1992 that the law does not authorise the physician to disclose to the partner of a sero-positive patient the danger to him or her from the patient’s behaviour if the patient is obstinately opposed to any revelation.13 Sexual relations were considered to be private and the responsibility of the two partners.14
Revealing medical secrets would risk alienating patients in need of care.15 We chose to study this especially problematic issue of whether to break confidentiality when the patient’s wife is at risk of contracting an STD from her husband.
Several investigators have examined the attitudes of health professionals towards the breach of confidentiality with regard to HIV infection. In the US, mental health professionals’ assessments of the acceptability of breaching confidentiality varied in accordance with the principle of the Tarasoff decisions.16–22 In contrast, in France, Moatti and colleagues23 found that, in accordance with the Ordre des Médecins, general practitioners were in favour of maintaining confidentiality when patients had not consented to having their medical information disclosed, except when the information was to be communicated to another healthcare professional. The attitude of the general public in the US, the UK and France about breaching patient confidentiality have been little studied.24 In the UK, Jones25 asked 30 consecutive patients whether confidentiality should be breached in five scenarios and found considerable support for breaching confidentiality to protect third parties, including 50% who decided that, in the case of a man with an STD who would not tell his wife, the doctor should tell her.
The results of these studies and the statements of medical bodies have identified five factors that seem to have an important influence on the attitudes of health professionals towards a breach of patient confidentiality, particularly about an STD: (a) the level of danger from the patient to the other person (in this case, his wife); (b) the patient’s intent to inform;(c) the patient’s intent to protect this other person; (d) the time taken by the doctor to discuss the issue with the patient; and (e) the advice received from another, more expert doctor. We examined, in the specific context of a husband found to have an STD, the relative impact of these five factors on lay people’s judgements regarding the acceptability of breaching confidentiality, the possible interactions among these factors and the impact of their sociodemographic characteristics on these judgements (eg, age, sex and educational level). We also compared the acceptability judgements of lay people with those of health professionals (psychologists and physicians) and the judgements of physicians with those of psychologists.
The method used was an application of the Functional Theory of Cognition of Norman Anderson.26–28 The primary aim of Anderson’s method is to present the cognitive rules used by people to integrate information when they make a judgement or decision. Anderson’s method assumes that people place subjective values on different pieces of information and that they combine these values by means of cognitive algebra dominated by addition, multiplication and averaging. The method studies how they do this indirectly and functionally—that is, it infers from people’s judgements of the combined value of two or more stimuli (or pieces of information) the cognitive rules used to arrive at these judgements.
In Anderson’s method, participants evaluate combinations of factors, rather than single factors. Accordingly, we presented our participants with a series of patient vignettes rather than with a questionnaire, and thereby were able to simulate the way the issue would appear in real life—in the context of actual patients with particular characteristics. The method requires, in addition, a complete factorial design—that is, our set of vignettes had to consist of all possible combinations of the within-subject factors. This design not only facilitates the determination of the impact each factor has on the overall judgements, but is also necessary for the investigation of their interactions and of the cognitive rules participants used in combining them. Furthermore, Anderson found that the true importance of each factor and the cognitive rules people used were shown better by stable rather than by momentary judgements of combined values. His method also requires, therefore, that participants become familiar with the task, and with these combinations of variables in a familiarisation phase, before they give a final set of judgements.
The lay participants were unpaid volunteers recruited and tested by one of the authors (MG). She contacted 250 people walking along the sidewalks on the university campus and in the city of Toulouse, explained the study, asked them to participate, and, if they agreed, arranged where and when to carry out the experiment. Of the people contacted, 144 (58%) participated. MG also contacted 30 psychologists and physicians working in private offices or in the main hospitals of Toulouse. Of whom, 17 (57%) participated (10 psychologists and 7 physicians). Recruitment of physicians was difficult and was stopped once it was realised that their responses were extremely homogeneous.
The material consisted of 48 cards that had a story of a few lines, a question and a response scale. The vignettes were composed according to a five within-subject factor design that used the severity of the transmissible disease (severe, lethal) × the time taken to discuss this with the patient (little time, much time) × the level of intent to inform the spouse about the disease (no intent to inform, intent to inform one of these days, intent to inform immediately) × the intent to adopt protective behaviours (no intent, intent) × the decision to consult an expert in STDs (call to an expert, no call to an expert), 2×2×3×2×2. Other information was kept constant: notably, all participants were men, and in each case the doctor decided to call the patient’s partner personally to inform her that her husband had an STD. No specific STDs were mentioned in the vignettes.
Under each vignette was a question—“To what extent do you believe that the decision made by the doctor is acceptable?”—and a 22-cm linear response scale with anchors of “not acceptable at all” and “completely acceptable”. Two examples are given in the Appendix. The cards were arranged randomly and in a different order for each participant.
Finally, the participants answered additional questions about age, sex, educational level, religious belief and religious background.
The site was, for the lay people, either a vacant university classroom or the participant’s private home, and for the professionals, their office or a vacant hospital room. Each person was tested individually. The session had two phases. In the familiarisation phase, after the experimenter explained what was expected, the participant read each vignette out loud, was reminded by the experimenter of the items of information in it and indicated on the response scale the acceptability of breaching confidentiality. After completing the 48 ratings, the participant was allowed to look back at, compare and change his or her responses. In the experimental phase, the participant worked at his or her own pace, but was not allowed to look back at and change previous responses. In both phases, the experimenter made certain that each participant, regardless of age, educational level or professional status, was able to understand all the necessary information before making a rating.
Both the lay people and the professionals took 30–45 min to complete both phases. The experimental phase went quickly because they were already familiar with the task and the material. No lay person or professional complained about the number of vignettes or about their credibility.
For each of the 48 scenarios in the experimental phase, the distance between the left anchor and each answer given by the participant on the response scale was measured. All subsequent analyses were based on these measures.
In accordance with Anderson’s method,18–20 the data were analysed, at the group level (lay people, psychologists and medical doctors), by performing analyses of variance and by constructing graphs (with Statistica 5.0). The design of the analysis of variance was participant’s age × sex × severity of the disease × time taken × intent to inform × protective behaviour × expert, 2×2×2×2×3×2×2. Educational level was not introduced as a factor in this design because preliminary analyses showed that it had no appreciable effect and no noticeable interactions with the other factors. In light of the multiplicity of comparisons, the level of significance was set at 0.005. A cluster analysis was also conducted on the raw data from the lay participants group.
Characteristics of participants
The lay participants consisted of 144 people (95 women and 49 men) aged 18–59 years, with a mean age of 30 years 7 months. Of these, 81% had completed secondary education. All but eight reported that they were in good health. Fifty per cent were religious believers and 50% non-believers, although only 17% were churchgoers.
The 10 psychologists (7 women and 3 men) were aged 22–52 years, with a mean of 32 years 10 months. The 7 doctors (5 women and 2 men) were aged 26–56 years, with a mean of 30 years; 5 were generalists and 2 were specialists (1 working in a hospital).
For the lay participants, the overall mean value of all the ratings was 11.76 cm. The highest mean response, 17.01 cm, was still very distant from the possible maximal answer, 22 cm. Thus no ceiling effect complicated the interpretation of the results.
Each of the five within-subject factors under study had a salient effect. As shown in fig 1 (top panel), all sets of curves in each panel are ascending: the less willing the patient was to inform his wife, the more acceptable the decision (13.12–10.01 = 3.11), F(2, 286) = 77.29, p<0.001. All pairs of curves are clearly separated: if an expert was consulted, the decision was judged to be more acceptable than if an expert was not consulted (12.44–11.08 = 1.36), F(1, 143) = 28.60, p<0.001. The sets of curves in the two right panels were higher on the y axis than in the two left panels: the less concerned the patient appeared about the protection of his wife, the more acceptable the decision (14.00–9.53 = 4.47), F(1, 143) = 127.49, p<0.001. The sets of curves in the second and fourth panels were higher on the y axis than in the first and third panels: the more severe the consequences of the infection, the more acceptable the decision (12.35–11.17 = 1.18), F(1, 143) = 44.36, p<0.001. Finally, the time taken to discuss the severity of the disease with the patient also had an effect (not shown). The more the time taken, the more acceptable the decision (12.22–11.30 = 0.92), F(1, 143) = 32.20, p<0.001.
Thus the factors with the greatest impact on the acceptability of breaching confidentiality were the patient’s intent to protect his wife (mean difference between no intent and intent = 4.47 on the 22-cm scale) and the patient’s stated intention to inform his wife (mean difference between no intent to inform and intent to inform immediately = 3.11). None of the two between-subject factors (age and sex) had a significant effect.
Only one notable interaction was observed. This was between intent to adopt a protective behaviour and time taken to discuss the issue, F(1, 143) = 32.20, p<0.001. Time had a greater impact if the patient was unwilling to adopt a protective behaviour (14.01–12.66 = 1.35) than if the patient appeared willing to do so (9.27–8.94 = 0.33).
Cluster analysis showed four very different clusters. The first cluster (n = 14), named “always acceptable”, had a mean acceptability rating of 19.36—that is, close to the maximum value of 22. None of the five factors had a major effect. The second cluster, named “depending on the many circumstances”, contained most of the lay participants (n = 87, 60%). Its mean acceptability rating was 11.88—that is, close to the midpoint of the scale. All five factors had marked effects. The third cluster (n = 30), labelled “consulting an expert”, had a mean acceptability rating of 11.78—that is, also close to the midpoint of the scale. Consultation with an expert in STD was the only factor found to have a considerable effect. The fourth cluster (n = 13), called “never acceptable”, had a mean acceptability rating of 2.73—that is, close to the minimum value of 0. Again none of the five factors had a consequential effect. There were no noticeable differences in the composition of the clusters with regard to age, sex or educational level.
When analyses of variance were repeated using only the majority cluster as subjects, the results were essentially the same.
As shown in fig 1 (centre panel), the ratings of the psychologists were very similar to those of the lay participants. All five main effects were present, as well as the interaction between intent to adopt a protective behaviour and time taken to discuss the issue. The mean acceptability value was lower (9.28) than that for the lay people (11.26), but the difference was not significant (p = 0.12), because of the small size of the sample.
Finally, as shown in fig 1 (bottom panel), the ratings of the medical doctors were very different from the others. The mean acceptability value for the group was 2.90—that is, close to the minimum value. No factor had a detectable effect. The individual mean responses were 1.00, 1.00, 1.00, 1.08, 1.92, 4.06 and 9.25. The outlier was a 31-year-old female generalist.
When a physician learns that his or her patient has an STD and could, therefore, infect the patient’s spouse or other sexual partners, the physician must make a difficult decision: whether to maintain strict confidentiality and thereby put the partner at risk or to inform the partner and thereby violate the ethic of confidentiality. Our study of the acceptability among French people of breaching confidentiality in such situations provided some striking results.
Our first major finding was that lay participants and psychologists differed greatly from physicians. Lay people considered breaching confidentiality to be moderately acceptable (a mean rating of 11.76) although, in their personal experiences as patients, they would surely want their own health information to remain private. Psychologists also rated breaching confidentiality as moderately acceptable (a mean rating of 9.28), although they, like physicians, deal routinely with very sensitive information about their patients and are well aware of the importance of confidentiality. In contrast, breaching confidentiality was in all cases unacceptable to physicians (a mean rating of only 2.90 on a scale of 0–22).
Why were physicians so different? We can offer a few speculations.
The scenarios were specifically about physicians; the physician participants had to imagine themselves in such a situation, whereas the psychologists (and lay people) had to imagine others.
Medical students and doctors in France, the US and the UK are repeatedly taught the importance of confidentiality, which has achieved an almost sacred status. This position is more consistent with the medical codes, laws and judicial decisions in France than in the US and in the UK (as pointed out in the first section).
Even in France, with fewer malpractice suits than in the US, physicians may be more sensitive than psychologists to the threat of lawsuits, and they may deal with this threat by focusing on the single guiding principle of confidentiality. In addition, physicians are at risk of losing the right to practice if condemned for violating the code of medical conduct, and in France the body responsible for this oversight, the Ordre des Médecins, is adamant about patient confidentiality.
Busy physicians tend, in opposition to multiple pressures from within and from outside the profession, to want to restrict their responsibilities to the specific issues at hand—the particular patient in their office and the individual physical and, in some cases, psychological dimensions of this patient’s illness—rather than take on the broader and more time-consuming tasks that can be performed by a social worker or public health official.
Our second major finding was that, for most of the lay people (as well as for the psychologists), all of the five factors we studied had direct effects on the acceptability of breaching confidentiality. These were (1) the patient’s intention to adopt protective behaviour, (2) the patient’s intention to inform the person at risk (the patient’s spouse), (3) the physician’s consultation with an expert, (4) the severity of the risk (ie, of the consequences of acquiring the STD) and (5) the time taken to talk with the patient about the severity of the disease. The time spent talking with the patient had a greater impact on the acceptability of informing the spouse if the patient was unwilling to adopt a protective behaviour than if the patient appeared willing to do so. People were well aware that, even when the threat could eventually result in death, such decisions are not clear cut, are fraught with moral complexity and ambiguity, are dependent on the particular circumstances and require discussion with the patient, and, in some cases, with experts from outside. Lay people in France appear, in general, to think more in accordance with American and British laws than with French laws, legal decisions and medical ethical dictums.
Our third major finding was that lay participants can be separated into quite distinct groups. A minority was opposed to breaching confidentiality to protect a spouse in all cases under consideration. Another minority was in favour of it in all cases. To some extent, therefore, the French public appears to be polarised about this as well as about other controversial issues of medical ethics (eg, the acceptability of ending the life of a suffering patient).29 Yet most of them were more moderate and nuanced. It is the responses of this group that are responsible for the findings described here. Most of these people (60% of the total lay participants) took all factors into account in their judgements. Some of them (14% of the total) were influenced only by whether the physician sought the advice of an expert; they thought that the issue was so morally ambiguous that the physician needed such guidance.
Our study has several limitations:
The participants were limited to the people of Toulouse, France. Generalisations to other countries must, therefore, be made with care.
The samples of psychologists and physicians were small in size. The study findings will therefore need to be confirmed on other samples.
The vignette in which the patient planned to adopt protective behaviours but not to inform his spouse about his STD may have seemed unrealistic, although no participant pointed this out.
The ratings were made about hypothetical scenarios, rather than real cases.
The importance of factors depends on the way they are phrased. For example, the severity of the risk may have had greater impact if we had explicitly labelled the more serious infection as HIV-AIDS.
Multiple other factors influence, of course, the decisions of individual physicians and patients, even though, as stated in the first section, previous work suggested that the factors we studied have wide generalisability.
Despite these limitations, our findings should make physicians and policy makers aware that most people—including patients, voters and jury members—are sensitive to the influence of situational factors on the difficult moral decision about whether a physician should breach confidentiality when he or she suspects that a patient with an STD may put his wife at risk.
Two examples of scenarios
Patient 1 comes to see Dr 1. The results of analyses show that the patient is currently infected with a sexually transmitted disease (STD). Given the current state of our knowledge, the consequence of this infection will, after a medium amount of time, be fatal.
Dr 1 mentions only briefly to patient 1 the consequences of this infection, the risks of spreading it, and the precautions that can be taken against spreading it. Patient 1 manifests, during the visit, his intention not to mention his state of health to his wife.
It appears to Dr 1 that patient 1 will do nothing to protect his wife during sexual relations.
Worried about the health of patient 1’s wife, Dr 1 decides to call her and keep her informed about her husband’s infection and about the risks incurred. Before taking this step, Dr 1 takes the precaution of requesting the advice of Professor 1, a specialist in STDs.
To what extent do you believe that the decision made by Dr 1 is acceptable?
Not acceptable at all–Completely acceptable (on a 22-cm scale).
Patient 2 comes to see Dr 2. The results of analyses show that the patient is currently infected with an STD. Given the current state of our knowledge, the consequences of this infection will not be fatal, but will none the less be very serious.
Dr 2 spends much time in discussing with patient 2 the consequences of this infection, the risks of spreading it and the precautions that can be taken against spreading it. Patient 2 manifests, during the visit, his intention of telling his spouse immediately about his state of health.
It appears to Dr 2 that patient 2 will do everything to protect his spouse during sexual relations.
Worried about the health of patient 2’s wife, Dr 2 decides to call her and keep her informed about her husband’s infection and about the risks incurred. It is a decision that Dr 2 makes by himself.
To what extent do you believe that the decision made by Dr 2 is acceptable?
Not acceptable at all–Completely acceptable (on a 22-cm scale).
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.