Article Text

Download PDFPDF

Engagement without entanglement: a framework for non-sexual patient–physician boundaries
  1. Jacob M Appel
  1. Psychiatry, Icahn School of Medicine at Mount Sinai, New York, New York, USA
  1. Correspondence to Dr Jacob M Appel, Psychiatry, Icahn School of Medicine at Mount Sinai, New York, NY, USA; jacobmappel{at}


The integrity of the patient–physician relationship depends on maintaining professional boundaries. While ethicists and professional organisations have devoted significant consideration to the subject of sexual boundary transgressions, the subject of non-sexual boundaries, especially outside the mental health setting, has been largely neglected. While professional organisations may offer guidance on specific subjects, such as accepting gifts or treating relatives, as well as general guidance on transparency and conflict of interest, what is missing is a principle-based method that providers can use to assess non-sexual interactions with patients that transcend norms of practice. This paper attempts to offer an operational model for such assessment that considers not only the traditional emphasis on beneficence, but also incorporates concerns over entanglement and concordance.

  • professional-professional relationship
  • professional misconduct
  • ethics

Statistics from

Request Permissions

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.

The importance of boundaries in the patient–physician relationship has been recognised since ancient times, when the Hippocratic Oath prohibited sexual contact during house calls.1 Considerable focus has been placed by professional organisations, regulatory authorities and ethicists on the specific subject of sexual boundary transgressions. Since the 1970s and the seminal American case of Roy versus Hartogs (1975), a professional and public consensus has arisen that sexual relations between medical providers and their current patients are both malpractice and misconduct.2 Whether providers may engage in such relationships with former patients differs by specialty, while prohibitions on sexual relations with the close relatives of patients vary by jurisdiction.3 4 In contrast, non-sexual boundaries have received far less scrutiny, except through the narrow lens of psychodynamic psychotherapy. Boundary transgressions by physicians other than psychiatrists also receive much less attention in the literature.5 The result is a dearth of guidance on how clinicians outside the field of mental health should handle ethical challenges relating to non-sexual boundaries. This differs substantially from other fields, such as law, where non-sexual boundaries are codified extensively.6 In fact, the norms of medical practice with regard to non-sexual boundaries are generally not codified beyond a handful of specific areas like offering inducements or accepting gifts.

The most well-known approach to the subject of non-sexual boundaries is that of Gutheil and Gabbard.7 Eying the issue through the lens of psychotherapy, they propose a distinction between ‘boundary crossings’ and ‘boundary violations’. The former reflects ‘a deviation from classical therapeutic activity that is harmless, non-exploitative and possibly supportive of the therapy itself’, while the latter ‘is harmful or potentially harmful, to the patient and the therapy’, and may constitute ‘exploitation’.8 It is worth noting that in the Gutheil-Gabbard model, such crossings can occur bidirectionally—from patient toward physician or from physician toward patient.8 Unfortunately, the theoretical framework offered by Gutheil and Gabbard can be difficult to operationalise for providers in the field. They note that ‘Clinicians tend to feel that they understand the concept of boundaries instinctively, but using it in practice or explaining it to others is often challenging.’7 In addition, their model largely evaluates the acceptability of boundary-transgressing conduct on the basis of whether it helps or harms the patient. However, some boundary crossings that may prove benign for a particular patient may nonetheless have negative implications for patient–physician relations more broadly. For example, allowing a victim of intimate partner violence (IPV) to take shelter in one’s home may protect that individual, but if abusers come to believe that doctors harbour IPV patients, other providers may find themselves in danger and might even avoid caring for IPV patients to avoid that risk. Moreover, certain forms of engagement may clearly benefit a patient—such as hiring him to work in one’s office or assisting her directly with rent money—yet may entangle the physician in such ways that raise complex issues relating to expectations and equity. For example, the choice of which patients to assist financially may exacerbate both bias and perceptions of bias, or may lead patients to place similar demands on other providers who are not inclined or equipped to offer non-medical assistance. It is important to note that even when such bias is not the driving factor underlying physician conduct, the patient’s misperception of being favoured can exacerbate perceptions of inequity, both real and misperceived, among other patients and in society at large, especially when such misperceptions are reported publicly. So even a provider who offers a particular form of assistance to all of his patients in need—such as free bus fare—should be careful that patients do not conclude they are being singled out for beneficial treatment. An effective, operationalisable model for managing non-sexual boundaries should take into account these concerns.

These issues have become more salient as the patient–physician model has undergone a radical transformation over the past two generations.9 In particular, the rise of what Szasz and Hollender termed the ‘model of mutual participation’ and subsequent forms of patient-centred medicine, often described as ‘shared decision-making’, have largely supplanted the paternalistic model of the previous era.10 11 On the whole, this shift had benefited patients through increased respect for their autonomy and interests. However, it has led to a typical patient–physician relationship that ‘has elements often associated with the notions of friendship and partnership.’10 What must be emphasised is that the modern patient–physician relationship may display aspects of a social relationship in a way that it usually did not prior to the 1960s, but this relationship is not an ordinary friendship. The physician is privy to information that patients may not have shared with others and possesses a range of professional privileges not afforded in ordinary friendships (like the opportunity to examine the patient unclothed and the power to write prescriptions). In addition, the physician is widely considered to have a fiduciary duty to patients under her care.12 Physicians should give meaningful reflection regarding any interactions with patients outside the medical setting, recognising the imbalances of knowledge and power inherent in these relationships. In the absence of clear guidance from authorities, physicians would be wise to consult colleagues on the propriety of such actions—an approach which is already recommended by the American College of Physicians regarding certain sexual boundaries.13

Although non-sexual boundaries are addressed piecemeal in the codes of various professional organisations and the rules of licensing boards, including in the American Medical Association’s (AMA) Principles of Medical Ethics in the USA and the General Medical Council’s (GMC) Good Medical Practice guidelines in the UK, these rules tend to focus almost exclusively on conflicts of interest. The emphasis is on transparency, fiduciary duty and beneficence. For instance, the American Psychiatric Association (APA), whose rules are derived from the AMA’s, requires merely that a psychiatrist ‘should not use the unique position of power afforded him/her by the psychotherapeutic situation to influence the patient in any way not directly relevant to the treatment goals’ and further requires that ‘when the psychiatrist’s outside relationships conflict with the clinical needs of the patient, the psychiatrist must always consider the impact of the relationship and strive to resolve conflicts in a manner that the psychiatrist believes to be beneficial to the patient.’14 Similarly, the GMC’s guidelines emphasise honesty in financial dealing and prohibit gifts or inducements ‘that may affect or be seen to affect the way you prescribe for, treat or refer patients or commission services for patients. You must not offer these inducements.’15 Broad principles that favour transparency and oppose self-dealing are necessary, but they are often not sufficient to guide physicians toward optimal ethical in situations where non-sexual boundary considerations arise. Physicians would benefit from a set of clear principles that can be operationalised in assessing situations that implicate non-sexual boundaries.

This paper proposes a three-prong test for physicians to determine whether non-sexual interactions that transcend traditional patient–physician boundaries are ethically permissible. First, echoing Gutheil and Gabbard, boundaries should only be traversed for the benefit of the patient, and such decisions should reflect intentionality and a risk–benefit assessment. Second, physicians should avoid unnecessary entanglement in their patients’ lives. Third, the physician and patient should have the same understanding of the motive and purpose underlying any engagement that transcends the traditional patient–physician relationship. Each of these principles is discussed further below. All three principles should be satisfied before embarking on an interaction with a patient that transcends traditional boundaries. Professional organisations and licensing authorities may wish to consider incorporating these principles into their codes of conduct.


The first proposed principle with regard to any non-traditional interaction with a patient is the most straightforward: the action should both be motivated by an intent to benefit the patient and it should have a reasonable likelihood of actually benefiting the patient. As important, the physician should have meaningfully reflected on whether or not this interaction is both motivated by beneficence and promises to achieve its goals. A risk–benefit analysis is often appropriate. Needless to say, all non-traditional interactions entail some risks. Patients may misinterpret motives, for instance, or may develop unrealistic expectations. It is the responsibility of the provider to weigh these concerns against the potential benefit to the patient. (One might also conceptualise this risk–benefit analysis in terms of the traditional core bioethical values of beneficence and non-maleficence.) Under some circumstances, when possible, it may also be appropriate to discuss these matters with a patient. For example, in accepting a wedding invitation from a patient or agreeing to receive an award from an organisation in which a patient is involved, providers may wish to directly (although politely) explain the boundary issues involved so that there are no misunderstandings. The examples of information sharing, accepting gifts and offering assistance may help clarify this principle.

Information sharing: in the routine course of care, patients may often make personal inquiries of physicians: Do you have children? Where did you go on vacation? Sharing non-controversial information to build rapport is generally reasonable. However, such revelations should be modulated in accordance with the needs and interests of a particular patient: sharing one’s marital status with a patient being treated for kidney failure might prove benign, but sharing the same information with a patient being treated for borderline personality disorder might compromise care. In addition, sometimes revealing more private personal or medical information may be appropriate, if tailored to a therapeutic end. For instance, a provider may wish to discuss her own path to weight loss or quitting smoking with patients struggling to achieve these goals. As the revelation becomes increasingly personal, more care should be taken to ensure that the primary motive is therapeutic, rather than conversational or gratuitous. Revelations that fundamentally alter the patient–physician relationship or run a significant risk of making the patient uncomfortable, even if intended therapeutically, should be avoided. Examples might include empathising with a patient’s recent divorce by discussing one’s own marital break-up, or sharing a history of childhood abuse with a trauma victim. Potentially controversial topics such as politics should generally be avoided, even if the provider and patients share similar views and the goal of discussion about the subject is to develop rapport. Reasons for avoiding such topics are that the patients’ views may change with time, small differences of opinion can still interfere with therapeutic relationship and, if the patient discusses the doctor’s views with others who do not share them, these conversations can affect the physician’s therapeutic relationships with other patients. Of course, the physicians’ views may become known to patients through other means, such as looking up political contributions online, but these discoveries require more intentional effort and do not raise the same sorts of expectations or concerns within the patient–physician encounter.

Accepting gifts: hospitals, medical schools and professional associations often have specific guidelines regarding gifts from providers to patients. As a general rule, the justification for a provider accepting a gift from a patient is to help that patient feel empowered through an expression of gratitude and to avoid offending that patient by rejecting the offer. At the same time, it is important to avoid allowing such gifts to compromise care, or to create a bias in favour of that patient ahead of others who cannot or do not offer gifts. The AMA recommends that providers decline ‘gifts that are disproportionately or inappropriately large, or when the physician would be uncomfortable to have colleagues know the gift had been accepted.’16 Providers should be particularly wary of gifts they personally desire or look forward to—such as an annual present of a case of wine from a well-heeled patient’s vineyard or cellar. Such gifts have significant implications for social justice, widely regarded as a core value in modern medicine. Most concerning are offers to fund a physician’s research or to endow a chair in her honour. In such cases, the primary benefit accrues to the provider or the institution, rather than the patient. The line between direct gifts to a provider and charitable gifts to the hospital or medical school that employs her is often not meaningful in that the patient–physician relationship is still altered (raising concerns of expectations and bias), while providers often do receive compensation indirectly from their institutions through increased funding or career advancement. At a minimum, providers should be reflective about the ethical implications of accepting such indirect gifts and their potential impact on clinical care and perceptions of the profession. Ideal practice might encourage the patient to bestow the gift without crediting or honouring the provider, achieving the same social end without undermining the therapeutic relationship.

Offers of assistance: situations may arise in which a compassionate physician considers offering a form of tangible, non-medical assistance to a patient. (Offers of assistance related to medical matters, such as negotiating with insurance companies, are widely considered within the scope of the patient–physician relationship and do not inherently raise ethical concerns.) Such offers might include intervening with non-medical bureaucracies, such as calling a government agency on the patient’s behalf, or providing the patient with a ride home after an appointment. At the extreme, they could involve offering a patient employment, lodging or even direct financial assistance. Most significant contributions will raise concerns related to entanglement, discussed further below. However, even those offers that do not rise to the level of entanglement should only be extended if they are aimed either primarily or only to serve a therapeutic purpose or to benefit the patient. For instance, allowing a patient to volunteer in one’s office might benefit that individual, but if doing so also helps the provider complete necessary work for free, one of the primary beneficiaries is now the provider, and such arrangements that are mutually beneficial should largely be avoided.


The second proposed principle, avoiding entanglement with patients, has received far less attention in the literature. Yet in many regards, the danger of entanglement is the most insidious in that actions initially intended to benefit a patient can result in either harm to that patient or damage to the profession and other patients more generally. Entanglement may best be thought of as a form of engagement that significantly immerses a physician in a patient’s personal or professional life to the degree that either the non-medical relationship overshadows the medical relationship or that extrication from the non-medical relationship impacts the medical relationship. Even entanglements that appear beneficial to the patient should be avoided. For instance, co-founding a non-profit health awareness organisation might empower that patient, improve the public health and draw valuable attention to a particular disease. However, the duty to the non-profit may come into conflict with fiduciary obligations to the patient; in addition, the perception of favouritism toward that patient, real or imagined, may have a negative impact on the healthcare of others. One key factor to assess with regard to entanglement is whether the interaction is likely to be repeated: attending the funeral of a patient’s spouse, for example, involves far less entanglement than accepting an invitation to a holiday dinner that is likely to be recurring. Another factor to assess may be the relationship of the particular entanglement to the overall course of care: attending a life event for a patient of 30 years standing may not raise the same boundary concerns or expectations as attending a life event for a patient new to care. It should also be noted that the level of acceptable entanglement may differ with regard to current and former patients. For instance, ‘giving away’ a patient in marriage, if she lacks family, might involve far too much entanglement, while ‘giving away’ a former paediatric patient who is now an adult would not raise the same degree of concern.

Social: social entanglements are likely to arise with the most frequency and implicate the largest swath of ethical grey area. As a general rule, physicians should not engage in extra medical social interactions with patients such as meeting them for meals or drinks, calling them for personal advice or actively involving themselves in the social lives of the patients. The pitfalls in the last of these interactions may not be readily apparent, but even social assistance offered with the best of intentions can backfire terribly. For example, Allen Collins, Chair of Psychiatry at New York’s Lenox Hill Hospital, lost $650 000 after setting up a patient on dates with other patients.17 He claimed his actions were ‘reasonably intended to address’ the patient’s ‘expressed feelings of isolation’ and ‘inability to meet people on her own’, but a jury found his conduct deleterious to her welfare and outside the standard of care.18 (Of note, Collins further entangled himself in the patient’s life by referring her to a divorce lawyer.) While offering patients general advice about their social lives can be an important part of the patient–physician relationship, especially for mental health providers, actively involving them in social activities of one’s own or others can prove problematic.

Difficulties are particularly likely to arise in two situations. First, physicians often care for the relatives of friends or colleagues, sometimes even extending professional courtesy and remitting charges. It would be unrealistic to expect physicians to entirely discontinue this practice. However, particular care should be taken to ensure that these relationships remain entirely professional and that the lines between professional and social relations do not blur. For example, even is the absence of a risk of a sexual boundary violation, making house calls off hours (unless this is a standard part of one’s practice) or combining social and professional activities (such as doing a physical in a private room at a wedding) can lead to an unnecessary and deleterious confusion of roles. Second, physicians may encounter patients with whom they have incidental or tangential social relationships, particularly in smaller or more insular communities: a child’s former teacher, for instance, or a clerk at a local shop. Such overlap between the medical and social worlds is sometimes unavoidable. What is essential for good practice is to avoid additional entanglement based on these existing loose relationships. For example, one should not shop more frequently at a store because one’s patient works there as a clerk, even though human nature may tempt the provider to entangle herself further in the non-medical relationship.

Professional: both seeking and offering non-medical professional assistance can have significant implications for a patient–physician relationship. In the former category, a provider might call a patient who is a lawyer for legal advice or ask an accountant for help with his taxes. One clear concern here is that the patient may feel obliged to offer assistance when he does not feel comfortable doing so, fearing refusal will have an impact on his medical care. Yet even if one disregards the power dynamics, these requests create a degree of entanglement that may impinge on the judgment of both provider and patient. What if the legal or tax advice proves wrong? One can imagine a scenario where the physician ends up wishing to sue his own patient for negligence. Physicians should also avoid offers of non-medical guidance that may entangle them with patients. At the far end of a continuum, a physician who also possesses an additional credential, such as a JD or a CPA, should refrain from offering legal or tax advice to a patient, even gratis. (This is important because physicians may overestimate their ability to offer non-medical guidance to patients, especially if they have additional credentials but are not actively engaged with the secondary field in question.) But as physicians are often held in high esteem by patients, even in areas where they lack expertise, the wise physician should forgo offering advice on non-medical matters, such as choosing a school or buying a home. A patient may rely on such advice and later regret it to the detriment of the patient–physician relationship.

Financial: most providers recognise that financial arrangements that serve their own interests at the expense of the patient are unacceptable. In contrast, financial relationships that either benefit both provider and patient, primarily the patient, or the broader public may appear more ambiguous. They are generally fraught with risks of conflict of interest. Among the forms of entanglement that physicians should avoid are employing a patient; investing in a patient’s product or company; serving as a consultant to a patient’s business or non-profit entity; facilitating business dealings between patients or between one’s patients and members of one’s own family; and holding property or significant funds on behalf of a patient. Although it has occurred in high-profile cases, accepting financial benefits from a deceased patient through a will or trust is also highly problematic, as it may lead the public to believe that patients will receive better care if they make their physicians their beneficiaries. Even something as seemingly innocuous as offering a stock tip or accepting a lead on a horse race can reshape the relationship between provider and patient. Ideally, the only financial transactions between doctors and patients should be those involving payments for medical services rendered. In reality, circumstances may arise where low-cost financial interactions that serve only the patient’s interests, such as assisting an indigent patient with bus fare after an appointment, may be justified. However, these seemingly low-cost interactions should be considered judiciously and implemented sparingly; as the value at stake increases, so do ethical concerns for excess entanglement.


A third proposed principle, that of concordance, does not appear to be discussed in the existing literature at all. Concordance, in this instance, refers to what in contract law is often termed a ‘meeting of the minds’: namely, that when a traditional boundary is broached, both provider and patient should have the same understanding of the nature and purpose of this breach. An example may help elucidate the importance of this principle. Let us say a patient is the dean of admissions at a prominent medical school and requests an appointment at the end of the workday—after the physician’s normal close of business—to accommodate her schedule. There is nothing inherently wrong with such an accommodation. However, let us further imagine that the physician’s child, who will soon be applying to medical school, volunteers in a nearby medical office and stops by the physician’s office at the end of the day for a ride home with the physician. While accommodating the patient would generally be fine, accommodating the patient with the purpose of having patient and child cross paths would prove problematic—beyond the possible issues of entanglement that might arise. Under such circumstances, the provider should make a full disclosure of the different motives so that both patient and provider have the same understanding of the interaction. Sometimes such full disclosure is not possible in advance, especially if non-concordant events occur spontaneously. Under such circumstances, the physician should address the difference in motivation or underlying conflict with the patient as soon as opportunity reasonably permits and should strive to prevent future occurrences.

Concordance may not seem nearly as important as beneficence or avoiding entanglements, but it appears differently when one considers that the opposite of concordance is self-interested deception. Transparency of the physician’s own interest is essential to prevent situations in which the patient later discovers hidden motives and feels betrayed, which can jeopardise care. Moreover, as a general rule, motives for interactions that transcend boundaries should be considered highly suspect if a provider does not feel comfortable sharing them with the patient. Rare circumstances may exist where it is permissible to withhold motives for medical interventions from the patient under the doctrine of therapeutic privilege. In contrast, withholding motives from the patient for non-medical transgressions is likely never justified. Of course, situations may arise where a physician believes they are acting in concordance with the patient and then discover otherwise; what should be expected of providers is a reasonable effort to ensure concordance in advance and then an effort to restore it through transparency if the provider’s original beliefs were mistaken.


The above principles are designed to be both operational and flexible, as considerable grey area exists between forms of engagement that are ideal and those that may prove problematic. The principles are likely to prove inadequate to address three specific situations, which merit further examinations.

Pre-existing relationships: circumstances will arise in which physicians are called on to provide medical care for individuals with whom they have pre-existing professional, social or familial relationships. Two distinct, related situations are likely to raise ethical issues: a close relative or associate requiring minor medical interventions or a more distant social contact requiring full-blown medical care. In the former case, it is unrealistic to expect providers to turn away all such patients. In fact, one study has shown that a majority of physicians have treated their own children at least once.19 Both the AMA and the American Pediatric Association allow treatment for ‘minor conditions’, although the former specifies that such interventions should be ‘short term’.20 21 More complex are situations in which the relationship is less direct, but the level of care more intense or of longer duration. For example, physicians will often rely on their own colleagues for care. These relationships are not inherently problematic, but providers should be carefully attuned to the power dynamics involved. Considerable variation exists regarding how medical boards approach this issue, so although many such cases arise without warning, physicians should—to the degree possible—familiarise themselves in advance with the rules in their own jurisdiction. In non-emergent circumstances, treating anyone who is also one’s direct employee or with whom one has a significant financial relationship should be avoided. The one caveat to this exception is that extreme care should be taken regarding even minor involvement with an individual with whom one is engaged in a sexual or romantic relationship. Finally, long-term or high stakes care relationships with close relatives or associates should be avoided. In short, both the proximity of the connection and the nature of the care should be weighed in determining whether to initiate a patient–physician relationship. The likelihood of additional conflicts arising from such a relationship, such as boundary or confidentiality implications for third parties, should also be considered. For example, treating a neighbour for a routine medical condition is not inherently problematic. However, if a provider is already treating another neighbour with whom the first neighbour is in a high-profile lawsuit, prudence may argue for referring the potential patient elsewhere to avoid possible conflicts.

Termination of patient–physician relationship: one distinctive feature of the patient–physician relationship is that, subject to limitations, it can be terminated by either party. If the terminating party is the physician, care must be taken to ensure that other providers are reasonably available to the patient so that he does not find himself abandoned. Even terminating the relationship in an appropriate manner may not prove sufficient to transcend all boundaries, such as prohibitions against sexual contact. The AMA Code of Ethics limits such future sexual relationships, while the APA prohibits them entirely.22 In contrast, in the absence of duress, physicians and patients are entitled to end their medical relationship and pursue other professional, financial and non-sexual social interactions. Under such circumstances, providers should be particularly aware of residual power dynamics between parties. However, no inherent barrier exists to terminating a patient–physician relationship to pursue such ends. For example, a provider may agree to refer a patient to a different physician and then engage in a mutual charitable endeavour that would not be advisable while the patient was still under his care. That is not to say that formal termination automatically eliminates dormant and subconscious power dynamics between patient and physicians, but rather that with appropriate care, such dynamics can sometimes be transcended.

Context and cultural norms: the above principles are designed for the patient–physician interactions in communities large enough so that provider and patient are relative strangers. In small, isolated communities, strict adherence to the principles may not always prove possible. For instance, in a small town where the only paediatrician is married to a local schoolteacher, it may prove logistically impossible for a child to avoid being taught by the husband and cared for by the wife, even though this might violate the entanglement principle. In addition, some cultural or religious communities may specify roles for medical providers that transcend those traditionally engaged in by allopathic providers. For example, if in such a community it is the norm for a patient’s physician to attend a patient’s bar mitzvah or confirmation or wedding, the likelihood of causing offence by declining to do so rises significantly, shifting the risk–benefit analysis related to concerns for entanglement. At the same time, the nature of small and insular communities may call for particular care with regard to maintaining confidence; a deidentified anecdote or case presentation may raise no ethical concerns in a large, heterogeneous city, but can lead to unwitting unmasking of a patient’s identity in a more circumscribed environment.


The absence of clear principles to assess actions that transcend traditional non-sexual patient–physician boundaries leaves providers in a quandary. They can develop their own casuistic conclusions by drawing analogies from specific guidance in the codes and canons of professional organisations such as the AMA and GMC. Or, as many do, they can use a ‘gestalt’ approach that principally considers the short-term impact on the patient—often not examining or undervaluing other indirect and long-term consequences for patients and the profession. By using a more rigorous, principle-based analysis that not only incorporates intent and potential benefit, but also concerns for entanglement and concordance, physicians should be able to navigate non-sexual boundary issues with more consistency and confidence.

Ethics statements

Patient consent for publication



  • Contributors JMA is the sole contributor and responsible for all content.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.