Four Models of thePhysician-Patient RelationshipEzekiel J. Emanuel, MD, PhD, Linda L. Emanuel, MD, PhD

DURING the last two decades or so,there has been a struggle over the pa-tient's role in medical decision makingthat is often characterized as a conflictbetween autonomy and health, betweenthe values of the patient and the valuesof the physician. Seeking to curtail phy-sician dominance, many have advocatedan ideal ofgreater patient control.1,2Oth-ers question this ideal because it fails toacknowledge the potentially imbalancednature ofthis interaction when one partyis sick and searching for security, andwhen judgments entail the interpreta-tion of technical information.3,4 Still oth-ers are trying to delineate a more mutualrelationship.5,6 This struggle shapes theexpectations ofphysicians and patients aswell as the ethical and legal standards forthe physician's duties, informed consent,and medical malpractice. This struggleforces us to ask, What should be the idealphysician-patient relationship?

We shall outline four models of thephysician-patient interaction, emphasiz-ing the different understandings of (1)the goals of the physician-patient inter¬action, (2) the physician's obligations,(3) the role of patient values, and (4) theconception ofpatient autonomy. To elab¬orate the abstract description of thesefour models, we shall indicate the typesof response the models might suggest ina clinical situation. Third, we shall alsoindicate how these models inform thecurrent debate about the ideal physician-patient relationship. Finally, we shallevaluate these models and recommendone as the preferred model.

As outlined, the models are Weberianideal types. They may not describe anyparticular physician-patient interactionsbut highlight, free from complicating de¬tails, different visions of the essentialcharacteristics of the physician-patient

From the Division of Cancer Epidemiology andControl, Dana-Farber Cancer Institute, Boston, Mass(E.J.E.); Program in Ethics and the Professions,Kennedy School of Government, Harvard University,Cambridge, Mass (EJE. and L.L.E.); and Division ofMedical Ethics, Harvard Medical School, Boston, Mass(L.L.E.). L.L.E. is also a Teaching and ResearchScholar of the American College of Physicians.

Reprint requests to Division of Cancer Epidemiologyand Control, Dana-Farber Cancer Institute, 44 BinneySt, Boston, MA 02115 (Dr E. J. Emanuel).

interaction.7 Consequently, they do notembody minimum ethical or legal stan¬dards, but rather constitute regulativeideals that are "higher than the law" butnot "above the law."8THE PATERNALISTIC MODEL

First is the paternalistic model, some¬times called the parental9 or priestly10model. In this model, the physician-pa¬tient interaction ensures that patientsreceive the interventions that best pro¬mote their health and well-being. Tothis end, physicians use their skills todetermine the patient's medical condi¬tion and his or her stage in the diseaseprocess and to identify the medical testsand treatments most likely to restorethe patient's health or ameliorate pain.Then the physician presents the patientwith selected information that will en¬

courage the patient to consent to theintervention the physician considersbest. At the extreme, the physician au¬

thoritatively informs the patient whenthe intervention will be initiated.

The paternalistic model assumes thatthere are shared objective criteria fordetermining what is best. Hence thephysician can discern what is in the pa¬tient's best interest with limited patientparticipation. Ultimately, it is assumedthat the patient will be thankful for de¬cisions made by the physician even if heor she would not agree to them at thetime.11 In the tension between the pa¬tient's autonomy and well-being, be¬tween choice and health, the paternal¬istic physician's main emphasis is to¬ward the latter.

In the paternalistic model, the physi¬cian acts as the patient's guardian, artic¬ulating and implementing what is best forthe patient. As such, the physician has ob¬ligations, including that ofplacing the pa¬tient's interest above his or her own andsoliciting the views ofothers when lackingadequate knowledge. The conception ofpatient autonomy is patient assent, eitherat the time or later, to the physician's de¬terminations of what is best.

THE INFORMATIVE MODELSecond is the informative model,

sometimes called the scientific,9 engi-

neering,10 or consumer model. In thismodel, the objective of the physician-patient interaction is for the physicianto provide the patient with all relevantinformation, for the patient to select themedical interventions he or she wants,and for the physician to execute the se¬lected interventions. To this end, thephysician informs the patient of his orher disease state, the nature of possiblediagnostic and therapeutic interven¬tions, the nature and probability ofrisksand benefits associated with the inter¬ventions, and any uncertainties ofknowl¬edge. At the extreme, patients couldcome to know all medical informationrelevant to their disease and availableinterventions and select the interven¬tions that best realize their values.

The informative model assumes afairly clear distinction between facts andvalues. The patient's values are well de¬fined and known; what the patient lacksis facts. It is the physician's obligationto provide all the available facts, andthe patient's values then determine whattreatments are to be given. There is norole for the physician's values, the phy¬sician's understanding of the patient'svalues, or his or her judgment of theworth of the patient's values. In theinformative model, the physician is a

purveyor of technical expertise, provid¬ing the patient with the means to ex¬ercise control. As technical experts, phy¬sicians have important obligations to pro¬vide truthful information, to maintaincompetence in their area of expertise,and to consult others when their knowl¬edge or skills are lacking. The concep¬tion ofpatient autonomy is patient con¬trol over medical decision making.THE INTERPRETIVE MODEL

The third model is the interpretivemodel. The aim of the physician-patientinteraction is to elucidate the patient'svalues and what he or she actually wants,and to help the patient select the avail¬able medical interventions that realizethese values. Like the informative phy¬sician, the interpretive physician pro¬vides the patient with information onthe nature of the condition and the risksand benefits of possible interventions.

Downloaded From: http://jama.jamanetwork.com/ on 07/21/2012

Comparing the Four Models

Informative Interpretive Deliberative PaternalisticPatient values Defined, fixed, and known to the

patientInchoate and conflicting, requir¬

ing elucidationOpen to development and revi¬

sion through moral discussionObjective and shared by physi¬

cian and patientPhysician's

obligationProviding relevant factual infor¬

mation and implementing pa¬tient's selected intervention

Elucidating and interpreting rele¬vant patient values as well asinforming the patient and im¬plementing the patient's se¬lected intervention

Articulating and persuading thepatient of the most admirablevalues as well as informingthe patient and implementingthe patient's selected inter¬vention

Promoting the patient's well-being independent of the pa¬tient's current preferences

Conception ofpatient's autonomy

Choice of, and control over,medical care

Self-understanding relevant tomedical care

Moral self-development relevantto medical care

Assenting to objective values

Conception ofphysician's role

Competent technical expert Counselor or adviser Friend or teacher Guardian

Beyond this, however, the interpretivephysician assists the patient in eluci¬dating and articulating his or her valuesand in determining what medical inter¬ventions best realize the specified val¬ues, thus helping to interpret the pa¬tient's values for the patient.

According to the interpretive model,the patient's values are not necessarilyfixed and known to the patient. Theyare often inchoate, and the patient mayonly partially understand them; theymay conflict when applied to specificsituations. Consequently, the physicianworking with the patient must elucidateand make coherent these values. To dothis, the physician works with the pa¬tient to reconstruct the patient's goalsand aspirations, commitments and char¬acter. At the extreme, the physicianmust conceive the patient's life as a nar¬rative whole, and from this specify thepatient's values and their priority.12·13Then the physician determines whichtests and treatments best realize thesevalues. Importantly, the physician doesnot dictate to the patient; it is the pa¬tient who ultimately decides which val¬ues and course of action best fit who heor she is. Neither is the physician judg¬ing the patient's values; he or she helpsthe patient to understand and use themin the medical situation.

In the interpretive model, the physi¬cian is a counselor, analogous to a cab¬inet minister's advisory role to a head ofstate, supplying relevant information,helping to elucidate values and suggest¬ing what medical interventions realizethese values. Thus the physician's ob¬ligations include those enumerated inthe informative model but also requireengaging the patient in a joint processof understanding. Accordingly, the con¬

ception of patient autonomy is self-un¬derstanding; the patient comes to knowmore clearly who he or she is and howthe various medical options bear on hisor her identity.THE DELIBERATIVE MODEL

Fourth is the deliberative model. Theaim of the physician-patient interactionis to help the patient determine and

choose the best health-related valuesthat can be realized in the clinical situ¬ation. To this end, the physician mustdelineate information on the patient'sclinical situation and then help elucidatethe types ofvalues embodied in the avail¬able options. The physician's objectivesinclude suggesting why certain health-related values are more worthy andshould be aspired to. At the extreme,the physician and patient engage in de¬liberation about what kind of health-related values the patient could and ul¬timately should pursue. The physiciandiscusses only health-related values, thatis, values that affect or are affected bythe patient's disease and treatments; heor she recognizes that many elements ofmorality are unrelated to the patient'sdisease or treatment and beyond thescope of their professional relationship.Further, the physician aims at no morethan moral persuasion; ultimately, co¬ercion is avoided, and the patient mustdefine his or her life and select the or¬

dering of values to be espoused. By en¬

gaging in moral deliberation, the phy¬sician and patient judge the worthinessand importance ofthe health-related val¬ues.

In the deliberative model, the physi¬cian acts as a teacher or friend,14 en¬

gaging the patient in dialogue on whatcourse of action would be best. Not onlydoes the physician indicate what the pa¬tient could do, but, knowing the patientand wishing what is best, the physicianindicates what the patient should do,what decision regarding medical ther¬apy would be admirable. The concep¬tion of patient autonomy is moral self-development; the patient is empowerednot simply to follow unexamined pref¬erences or examined values, but to con¬

sider, through dialogue, alternativehealth-related values, their worthiness,and their implications for treatment.

COMPARING THE FOUR MODELSThe Table compares the four models

on essential points. Importantly, all mod¬els have a role for patient autonomy; amain factor that differentiates the mod¬els is their particular conceptions of pa-

tient autonomy. Therefore, no singlemodel can be endorsed because it alonepromotes patient autonomy. Instead themodels must be compared and evalu¬ated, at least in part, by evaluating theadequacy of their particular conceptionsof patient autonomy.

The four models are not exhaustive.At a minimum there might be added afifth: the instrumental model. In thismodel, the patient's values are irrele¬vant; the physician aims for some goalindependent of the patient, such as thegood of society or furtherance of scien¬tific knowledge. The Tuskegee syphilisexperiment15"17 and the Willowbrook hep¬atitis study18·19 are examples of thismodel. As the moral condemnation ofthese cases reveals, this model is not anideal but an aberration. Thus we havenot elaborated it herein.

A CLINICAL CASETo make tangible these abstract de¬

scriptions and to crystallize essential dif¬ferences among the models, we will il¬lustrate the responses they suggest in aclinical situation, that of a 43-year-oldpremenopausal woman who has recentlydiscovered a breast mass. Surgery re¬veals a 3.5-cm ductal carcinoma with no

lymph node involvement that is estro¬gen receptor positive. Chest roentgen-ogram, bone scan, and liver functiontests reveal no evidence of metastaticdisease. The patient was recently di¬vorced and has gone back to work as a

legal aide to support herself. Whatshould the physician say to this patient?

In the paternalistic model a physicianmight say, "There are two alternativetherapies to protect against recurrenceof cancer in your breast: mastectomy orradiation. We now know that the sur¬vival with lumpectomy combined withradiation therapy is equal to that withmastectomy. Because lumpectomy andradiation offers the best survival andthe best cosmetic result, it is to be pre¬ferred. I have asked the radiation ther¬apist to come and discuss radiation treat¬ment with you. We also need to protectyou against the spread of the cancer toother parts of your body. Even though

Downloaded From: http://jama.jamanetwork.com/ on 07/21/2012

the chance of recurrence is low, you are

young, and we should not leave any ther¬apeutic possibilities untried. Recentstudies involving chemotherapy suggestimprovements in survival without re¬currence of breast cancer. Indeed, theNational Cancer Institute recommendschemotherapy for women with your typeofbreast cancer. Chemotherapy has sideeffects. Nevertheless, a few months ofhardship now are worth the potentialadded years of life without cancer."

In the informative model a physicianmight say, "With node-negative breastcancer there are two issues before you:local control and systemic control. Forlocal control, the options are mastec¬tomy or lumpectomy with or withoutradiation. From many studies we knowthat mastectomy and lumpectomy withradiation result in identical overall sur¬

vival, about 80% 10-year survival.Lumpectomy without radiation resultsin a 30% to 40% chance of tumor recur¬rence in the breast. The second issuerelates to systemic control. We knowthat chemotherapy prolongs survival forpremenopausal women who have axil¬lary nodes involved with tumor. Therole for women with node-negativebreast cancer is less clear. Individualstudies suggest that chemotherapy is ofno benefit in terms of improving overallsurvival, but a comprehensive review ofall studies suggests that there is a sur¬vival benefit. Several years ago, the NCIsuggested that for women like yourself,chemotherapy can have a positive ther¬apeutic impact. Finally, let me informyou that there are clinical trials, forwhich you are eligible, to evaluate thebenefits of chemotherapy for patientswith node-negative breast cancer. I canenroll you in a study if you want. I willbe happy to give you any further infor¬mation you feel you need."

The interpretive physician might out¬line much of the same information as theinformative physician, then engage indiscussion to elucidate the patient'swishes, and conclude, "It sounds to meas if you have conflicting wishes. Un¬derstandably, you seem uncertain howto balance the demands required for re¬

ceivingadditional treatment, rejuvenat¬ing your personal affairs, and maintain¬ing your psychological equilibrium. Letme try to express a perspective that fitsyour position. Fighting your cancer isimportant, but it must leave you with a

healthy self-image and quality time out¬side the hospital. This view seems com¬

patible with undergoing radiation ther¬apy but not chemotherapy. A lumpec¬tomy with radiation maximizes yourchance of surviving while preservingyour breast. Radiotherapy fights yourbreast cancer without disfigurement.

Conversely, chemotherapy would pro¬long the duration of therapy by manymonths. Further, the benefits of che¬motherapy in terms of survival aresmaller and more controversial. Giventhe recent changes in your life, you havetoo many new preoccupations to undergomonths of chemotherapy for a question¬able benefit. Do I understand you? Wecan talk again in a few days."

The deliberative physician might be¬gin by outlining the same factual infor¬mation, engage in a conversation to elu¬cidate the patient's values, but continue,"It seems clear that you should undergoradiation therapy. It offers maximal sur¬vival with minimal risk, disfigurement,and disruption of your life. The issue ofchemotherapy is different, fraught withconflicting data. Balancing all the op¬tions, I think the best one for you is toenter a trial that is investigating thepotential benefit of chemotherapy forwomen with node-negative breast can¬cer. First, it ensures that you receiveexcellent medical care. At this point, wedo not know which therapy maximizessurvival. In a clinical study the scheduleof follow-up visits, tests, and decisionsis specified by leading breast cancer ex¬

perts to ensure that all the women re¬ceive care that is the best available any¬where. A second reason to participatein a trial is altruistic; it allows you tocontribute something to women withbreast cancer in the future who will facedifficult choices. Over decades, thou¬sands of women have participated instudies that inform our current treat¬ment practices. Without those women,and the knowledge they made possible,we would probably still be giving youand all other women with breast cancermastectomies. By enrolling in a trial youparticipate in a tradition in which womenof one generation receive the higheststandard of care available but also en¬hance the care of women in future gen¬erations because medicine has learnedsomething about which interventions arebetter. I must tell yoti that I am notinvolved in the study; if you elect toenroll in this trial, you will initially seeanother breast cancer expert to planyour therapy. I have sought to explainour current knowledge and offer my rec¬ommendation so you can make the bestpossible decision."

Lacking the normal interchange withpatients, these statements may seem

contrived, even caricatures. Neverthe¬less, they highlight the essence of eachmodel and suggest how the objectivesand assumptions of each inform a phy¬sician's approach to his or her patients.Similar statements can be imagined forother clinical situations such as an ob¬stetrician discussing prenatal testing

or a cardiologist discussing cholesterol-reducing interventions.

THE CURRENT DEBATE AND THEFOUR MODELS

In recent decades there has been acall for greater patient autonomy or, assome have called it, "patient sover¬

eignty,"20 conceived as patient choiceand control over medical decisions. Thisshift toward the informative model isembodied in the adoption of businessterms for medicine, as when physiciansare described as health care providersand patients as consumers. It can alsobe found in the propagation of patientrights statements,21 in the promotion ofliving will laws, and in rules regardinghuman experimentation. For instance,the opening sentences of one law state:"The Rights of the Terminally 111 Actauthorizes an adult person to controldecisions regarding administration oflife-sustaining treatment. . . . The Actmerely provides one way by which a

terminally-ill patient's desires regard¬ing the use of life-sustaining procedurescan be legally implemented" (emphasisadded).22 Indeed, living will laws do notrequire or encourage patients to discussthe issue of terminating care with theirphysicians before signing such docu¬ments. Similarly, decisions in "right-to-die" cases emphasize patient control overmedical decisions. As one court put it23:The right to refuse medical treatment is ba¬sic and fundamental. ... Its exercise re¬quires no one's approval. . . . [T]he control¬ling decision belongs to a competentinformed patient. ... It is not a medicaldecision for her physicians to make.

. . .

It isa moral and philosophical decision that, be¬ing a competent adult, is [the patient's]alone, (emphasis added)

Probably the most forceful endorse¬ment ofthe informative model as the idealinheres in informed consent standards.Prior to the 1970s, the standard for in¬formed consent was "physicianbased."2426 Since 1972 and the Canter¬bury case, however, the emphasis hasbeen on a "patient-oriented" standard ofinformed consent in which the physicianhas a "duty" to provide appropriate med¬ical facts to empower the patient to use hisor her values to determine what interven¬tions should be implemented.25-27True consent to what happens to one's self isthe informed exercise of a choice, and thatentails an opportunity to evaluate knowl-edgeably the options available and the risksattendant upon each. . . . [I]t is the prerog¬ative of the patient, not the physician, to de¬terminefor himselfthe direction in which hisinterests seem to lie. To enable the patient tochart his course understandably, some fa¬miliarity with the therapeutic alternativesand their hazards becomes essential.27 (em¬phasis added)

Downloaded From: http://jama.jamanetwork.com/ on 07/21/2012

SHARED DECISION MAKINGDespite its dominance, many have

found the informative model "arid."20The President's Commission and otherscontend that the ideal relationship doesnot vest moral authority and medicaldecision-making power exclusively in thepatient but must be a process of shareddecision making constructed around"mutual participation and respect."20·28The President's Commission argues thatthe physician's role is "to help the pa¬tient understand the medical situationand available courses of action, and thepatient conveys his or her concerns andwishes."20 Brock and Wartman29 stressthis fact-value "division of labor"—hav¬ing the physician provide informationwhile the patient makes value deci¬sions—by describing "shared decisionmaking" as a collaborative processin which both physicians and patients makeactive and essential contributions. Physi¬cians bring their medical training, knowl¬edge, and expertise—including an under¬standing of the available treatmentalternatives—to the diagnosis and manage¬ment of patients' condition. Patients bringknowledge of their own subjective aims andvalues, through which risks and benefits ofvarious treatment options can be evaluated.With this approach, selecting the best treat¬ment for a particular patient requires thecontribution of both parties.

Similarly, in discussing ideal medicaldecision making, Eddy30 argues for thisfact-value division of labor between thephysician and patient as the ideal:It is important to separate the decision pro¬cess into these two steps. . . . The first stepis a question of facts. The anchor is empiri¬cal evidence. .

. .

[T]he second step is a

question not of facts but ofpersonal values or

preferences. The thought process is not an¬

alytic but personal and subjective..

. . [I]tis the patient's preferences that should de¬termine the decision. . . . Ideally, you and I[the physicians] are not in the picture. Whatmatters is what Mrs. Smith thinks.

This view of shared decision makingseems to vest the medical decision-making authority with the patient whilerelegating physicians to technicians"transmitting medical information andusing their technical skills as the patientdirects."20 Thus, while the advocates of"shared decision making" may aspire to¬ward a mutual dialogue between physi¬cian and patient, the substantive view in¬forming their ideal reembodies the infor¬mative model under a different label.

Other commentators have articu¬lated more mutual models of the phy¬sician-patient interaction.5·6·25 Promi¬nent among these efforts is Katz'31 TheSilent World of the Doctor and Patient.Relying on a Freudian view in whichself-knowledge and self-determination

are inherently limited because of un¬conscious influences, Katz views dia¬logue as a mechanism for greater self-understanding of one's values andobjectives. According to Katz, this viewplaces a duty on physicians and patientsto reflect and communicate so thatpatients can gain a greater self-understanding and self-determination.Katz' insight is also available on

grounds other than Freudian psycho¬logical theory and is consistent with theinterpretive model.13OBJECTIONS TO THEPATERNALISTIC MODEL

It is widely recognized that the pater¬nalistic model is justified during emer¬

gencies when the time taken to obtain in¬formed consent might irreversibly harmthe patient.1·2·20 Beyond such limited cir¬cumstances, however, it is no longer ten¬able to assume that the physician and pa¬tient espouse similar values and views ofwhat constitutes a benefit. Consequently,even physicians rarely advocate the pa¬ternalistic model as an ideal for routinephysician-patient interactions.32OBJECTIONS TO THEINFORMATIVE MODEL

The informative model seems both de¬scriptively and prescriptively inaccu¬rate. First, this model seems to have no

place for essential qualities of the idealphysician-patient relationship. The in¬formative physician cares for the pa¬tient in the sense of competently imple¬menting the patient's selected interven¬tions. However, the informative physi¬cian lacks a caringapproach that requiresunderstanding what the patient valuesor should value and how his or her ill¬ness impinges on these values. Patientsseem to expect their physician to havea caring approach; they deem a techni¬cally proficient but detached physicianas deficient, and properly condemned.Further, the informative physician isproscribed from giving a recommenda¬tion for fear of imposing his or her willon the patient and thereby competingfor the decision-making control that hasbeen given to the patient.25 Yet, if oneof the essential qualities of the ideal phy¬sician is the ability to assimilate medicalfacts, prior experience of similar situa¬tions, and intimate knowledge of thepatient's view into a recommendationdesigned for the patient's specific med¬ical and personal condition,35·25 then theinformative physician cannot be ideal.

Second, in the informative model theideal physician is a highly trained subspe-cialistwhoprovidesdetailed factual infor¬mation and competently implements thepatient's preferred medical intervention.Hence, the informative model perpetu-

ates and accentuates the trend towardspecialization and impersonalizationwithin the medical profession.

Most importantly, the informativemodel's conception of patient autonomyseems philosophically untenable. The in¬formative model presupposes that per¬sons possess known and fixed values,but this is inaccurate. People are oftenuncertain about what they actually want.Further, unlike animals, people havewhat philosophers call "second order de¬sires,"3335 that is, the capacity to reflecton their wishes and to revise their owndesires and preferences. In fact, free¬dom of the will and autonomy inhere inhaving "second order desires" and be¬ing able to change our preferences andmodify our identity. Self-reflection andthe capacity to change what we wantoften require a "process" of moral de¬liberation in which we assess the valueof what we want. And this is a processthat occurs with other people who knowus well and can articulate a vision ofwhowe ought to be that we can assent to.13Even though changes in health or im¬plementation of alternative interven¬tions can have profound effects on whatwe desire and how we realize our de¬sires, self-reflection and deliberation playno essential role in the informativephysician-patient interaction. The infor¬mative model's conception ofautonomy isincompatible with a vision of autonomythat incorporates second-order desires.

OBJECTIONS TO THEINTERPRETIVE MODEL

The interpretive model rectifies thisdeficiency by recognizing that personshave second-order desires and dynamicvalue structures and placing the eluci¬dation of values in the context of thepatient's medical condition at the centerof the physician-patient interaction.Nevertheless, there are objections tothe interpretive model.

Technical specialization militatesagainst physicians cultivating the skillsnecessary to the interpretive model.With limited interpretive talents andlimited time, physicians may unwittinglyimpose their own values under the guiseof articulating the patient's values. Andpatients, overwhelmed by their medicalcondition and uncertain of their own

views, may too easily accept this impo¬sition. Such circumstances maypush theinterpretive model toward the pater¬nalistic model in actual practice.

Further, autonomy viewed as self-un¬derstanding excludes evaluative judg¬ment of the patient's values or attemptsto persuade the patient to adopt othervalues. This constrains the guidance andrecommendations the physician can of¬fer. Yet in practice, especially in pre-

Downloaded From: http://jama.jamanetwork.com/ on 07/21/2012

ventive medicine and risk-reduction in¬terventions, physicians often attemptto persuade patients to adopt particularhealth-related values. Physicians fre¬quently urge patients with high choles¬terol levels who smoke to change theirdietary habits, quit smoking, and beginexercise programs before initiating drugtherapy. The justification given for thesechanges is that patients should valuetheir health more than they do. Simi¬larly, physicians are encouraged to per¬suade their human immunodeficiency vi¬rus (HlV)-infected patients who mightbe engaging in unsafe sexual practiceseither to abstain or, realistically, to adopt"safer sex" practices. Such appeals arenot made to promote the HIV-infectedpatient's own health, but are groundedon an appeal for the patient to assume

responsibility for the good of others.Consequently, by excluding evaluativejudgments, the interpretive modelseems to characterize inaccurately idealphysician-patient interactions.

OBJECTIONS TO THEDELIBERATIVE MODEL

The fundamental objections to the de¬liberative model focus on whether it isproper for physicians to judge patients'values and promote particular health-related values. First, physicians do notpossess privileged knowledge of the pri¬ority of health-related values relative toother values. Indeed, since ours is a

pluralistic society in which people es¬

pouse incommensurable values, it islikely that a physician's values and viewof which values are higher will conflictwith those of other physicians and thoseof his or her patients.

Second, the nature of the moral de¬liberation between physician and pa¬tient, the physician's recommended in¬terventions, and the actual treatmentsused will depend on the values of theparticular physician treating the patient.However, recommendations and care

provided to patients should not dependon the physician's judgment of the wor¬thiness of the patient's values or on thephysician's particular values. As onebioethicist put it36:The hand is broken; the physician can repairthe hand; therefore the physician must re¬

pair the hand—as well as possible—withoutregard to personal values that might lead thephysician to think ill of the patient or of thepatient's values. . . . [A]t the level of clinicalpractice, medicine should be value-free inthe sense that the personal values of thephysician should not distort the making ofmedical decisions.

Third, it may be argued that the de¬liberative model misconstrues the pur¬pose of the physician-patient interac¬tion. Patients see their physicians to

receive health care, not to engage inmoral deliberation or to revise theirvalues. Finally, like the interpretivemodel, the deliberative model may eas¬

ily metamorphose into unintended pa¬ternalism, the very practice that gen¬erated the public debate over theproper physician-patient interaction.

THE PREFERRED MODEL AND THEPRACTICAL IMPLICATIONS

Clearly, under different clinical cir¬cumstances different models may be ap¬propriate. Indeed, at different times allfour models may justifiably guide phy¬sicians and patients. Nevertheless, it isimportant to specify one model as theshared, paradigmatic reference; excep¬tions to use other models would not beautomatically condemned, but would re¬

quire justification based on the circum¬stances of a particular situation. Thus,it is widely agreed that in an emergencywhere delays in treatment to obtain in¬formed consent might irreversibly harmthe patient, the paternalistic model cor¬

rectly guides physician-patient interac¬tions. Conversely, for patients who haveclear but conflicting values, the interpre¬tive model is probably justified. For in¬stance, a 65-year-old woman who hasbeen treated for acute leukemia may haveclearly decided against reinduction che¬motherapy if she relapses. Severalmonths before the anticipated birth ofherfirst grandchild, the patient relapses. Thepatient becomes torn about whether toendure the risks of reinduction chemo¬therapy in order to live to see her firstgrandchild or whether to refuse therapy,resigning herself to not seeing her grand¬child. In such cases, the physician mayjustifiably adopt the interpretive ap¬proach. In other circumstances, wherethere is only a one-time physician-patientinteraction without an ongoing relation¬ship in which the patient's values can beelucidated and compared with ideals,such as in a walk-in center, the informa¬tive model may be justified.

Descriptively and prescriptively, weclaim that the ideal physician-patientrelationship is the deliberative model.We will adduce six points to justify thisclaim. First, the deliberative model more

nearly embodies our ideal of autonomy.It is an oversimplification and distortionof the Western tradition to view respect¬ing autonomy as simply permitting a

person to select, unrestricted by coer¬

cion, ignorance, physical interference,and the like, his or her preferred courseof action from a comprehensive list ofavailable options.34·35 Freedom and con¬trol over medical decisions alone do notconstitute patient autonomy. Autonomyrequires that individuals critically as¬sess their own values and preferences;

determine whether they are desirable;affirm, upon reflection, these values asones that should justify their actions;and then be free to initiate action torealize the values. The process of de¬liberation integral to the deliberativemodel is essential for realizing patientautonomy understood in this way.

Second, our society's image ofan idealphysician is not limited to one who knowsand communicates to the patient rele¬vant factual information and compe¬tently implements medical interven¬tions. The ideal physician—often em¬bodied in literature, art, and popularculture—is a caring physician who in¬tegrates the information and relevantvalues to make a recommendation and,through discussion, attempts to per¬suade the patient to accept this recom¬mendation as the intervention that bestpromotes his or her overall well-being.Thus, we expect the best physicians toengage their patients in evaluative dis¬cussions ofhealth issues and related val¬ues. The physician's discussion does notinvoke values that are unrelated or tan-gentially related to the patient's illnessand potential therapies. Importantly,these efforts are not restricted to situ¬ations in which patients might make "ir¬rational and harmful" choices29 but ex¬tend to all health care decisions.

Third, the deliberative model is not a

disguised form of paternalism. Previ¬ously there may have been category mis¬takes in which instances of the deliber¬ative model have been erroneously iden¬tified as physician paternalism. And no

doubt, in practice, the deliberative phy¬sician may occasionally lapse into pa¬ternalism. However, like the idealteacher, the deliberative physician at¬tempts topersuade the patient ofthe wor¬thiness of certain values, not to imposethose values paternalistically; the physi¬cian's aim is not to subject the patient tohis or her will, but to persuade the patientof a course of action as desirable. In theLaws, Plato37 characterizes this funda¬mental distinction between persuasionand imposition for medical practice thatdistinguishes the deliberative from thepaternalistic model:A physician to slaves never gives his patientany account of his illness . . . the physicianoffers some orders gleaned from experiencewith an air of infallible knowledge, in thebrusque fashion of a dictator.

. . .

The freephysician, who usually cares for free men,treats their diseases first by thoroughly dis¬cussing with the patient and his friends hisailment. This way he learns something fromthe sufferer and simultaneously instructs him.Then the physician does not give his medica¬tions until he has persuaded the patient; thephysician aims at complete restoration ofhealth by persuading the patient to complywith his therapy.

Downloaded From: http://jama.jamanetwork.com/ on 07/21/2012

Fourth, physician values are relevantto patients and do inform their choice ofa physician. When a pregnant womanchooses an obstetrician who does notroutinely perform a battery of prenataltests or, alternatively, one whostrongly favors them; when a patientseeks an aggressive cardiologist whofavors procedural interventions or onewho concentrates therapy on dietarychanges, stress reduction, and life-stylemodifications, they are, consciously or

not, selecting a physician based on thevalues that guide his or her medical de¬cisions. And, when disagreements be¬tween physicians and patients arise,there are discussions over which valuesare more important and should be real¬ized in medical care. Occasionally, whensuch disagreements undermine thephysician-patient relationship and a

caring attitude, a patient's care is trans¬ferred to another physician. Indeed, inthe informative model the grounds fortransferring care to a new physician iseither the physician's ignorance or in¬competence. But patients seem toswitch physicians because they do not"like" a particular physician or thatphysician's attitude or approach.

Fifth, we seem to believe that physi¬cians should not only help fit therapiesto the patients' elucidated values, butshould also promote health-related val¬ues. As noted, we expect physicians topromote certain values, such as "safersex" for patients with HIV or abstain¬ing from or limiting alcohol use. Simi¬larly, patients are willing to adjust theirvalues and actions to be more compati¬ble with health-promoting values.38This is in the nature of seeking a caringmedical recommendation.

Finally, it may well be that many phy¬sicians currently lack the training and ca¬

pacity to articulate the values underlyingtheir recommendations and persuade pa¬tients that these values are worthy. But,in part, this deficiency is a consequence ofthe tendencies toward specialization andthe avoidance of discussions of values byphysicians that are perpetuated and jus¬tified by the dominant informative model.Therefore, if the deliberative modelseems most appropriate, then we need toimplement changes in medical care andeducation to encourage a more caring ap¬proach. We must stress understandingrather than mere provisions of factual in¬formation in keeping with the legal stan¬dards of informed consent and medicalmalpractice; we must educate physiciansnot just to spend more time in physician-patient communication but to elucidateand articulate the values underlying theirmedical care decisions, including routineones; we must shift the publicly assumedconception of patient autonomy that

shapes both the physician's and the pa¬tient's expectations from patient controlto moral development. Most important,we must recognize that developing a de¬liberative physician-patient relationshiprequires a considerable amount of time.We must develop a health care financingsystem that properly reimburses—rather than penalizes—physicians fortaking the time to discuss values withtheir patients.CONCLUSION

Over the last few decades, the dis¬course regarding the physician-patientrelationship has focused on two ex¬tremes: autonomy and paternalism.Many have attacked physicians as pa¬ternalistic, urging the empowerment ofpatients to control their own care. Thisview, the informative model, has be¬come dominant in bioethics and legalstandards. This model embodies a de¬fective conception of patient autonomy,and it reduces the physician's role tothat of a technologist. The essence ofdoctoring is a fabric of knowledge, un¬

derstanding, teaching, and action, inwhich the caring physician integratesthe patient's medical condition andhealth-related values, makes a recom¬mendation on the appropriate course ofaction, and tries to persuade the patientof the worthiness of this approach andthe values it realizes. The physician witha caring attitude is the ideal embodiedin the deliberative model, the ideal thatshould inform laws and policies that reg¬ulate the physician-patient interaction.

Finally, it may be worth noting that thefour models outlined herein are not lim¬ited to the medical realm; they may in¬form the public conception of other pro¬fessional interactions as well. We suggestthat the ideal relationships between law¬yer and client,14 religious mentor and la¬ity, and educator and student are well de¬scribed by the deliberative model, at leastin some of their essential aspects.

We would like to thank Robert Mayer, MD,Craig Henderson, MD, Lynn Peterson, MD, andJohn Stoeckle, MD, as well as Dennis Thompson,PhD, Arthur Applbaum, PhD, and Dan Brock,PhD, for their critical reviews of the manuscript.We would also like to thank the "ethics and the pro¬fessions" seminar participants, especially RobertRosen, JD, Francis Kamm, PhD, David Wilkins, JD,and Oliver Avens, who enlightened us in discussions.

References1. Veatch RM. A Theory of Medical Ethics. NewYork, NY: Basic Books Inc Publishers; 1981.2. Macklin R. Mortal Choices. New York, NY:Pantheon Books Inc; 1987.3. Ingelfinger FJ. Arrogance. N Engl J Med.1980;304:1507.4. Marzuk PM. The right kind of paternalism.N Engl J Med. 1985;313:1474-1476.5. Siegler M. The progression of medicine: fromphysician paternalism to patient autonomy to bu-reaucratic parsimony. Arch Intern Med. 1985;145:

713-715.6. Szasz TS, Hollender MH. The basic models ofthe doctor-patient relationship. Arch Intern Med.1956;97:585-592.7. Weber M; Parsons T, ed. The Theory of Socialand Economic Organization. New York, NY: TheFree Press; 1947.8. Ballantine HT. Annual discourse\p=m-\thecrisis inethics, anno domini 1979. N Engl J Med. 1979;301:634-638.9. Burke G. Ethics and medical decision-making.Prim Care. 1980;7:615-624.10. Veatch RM. Models for ethical medicine in a

revolutionary age. Hastings Cent Rep. 1975;2:3-5.11. Stone AA. Mental Health and Law: A System inTransition. New York, NY:JasonAronsonInc; 1976.12. Maclntyre A. After Virtue. South Bend, Ind:University of Notre Dame Press; 1981.13. Sandel MJ. Liberalism and the Limits ofJus-tice. New York, NY: Cambridge University Press;1982.14. Fried C. The lawyer as friend: the moral foun-dations of the lawyer client relationship. Yale LawJ. 1976;85:1060-1089.15. Jones JH. Bad Blood. New York, NY: FreePress; 1981.16. Final Report of the Tuskegee Syphilis StudyAd Hoc Advisory Panel. Washington, DC: PublicHealth Service; 1973.17. Brandt AM. Racism and research: the case ofthe Tuskegee Syphilis Study. Hastings Cent Rep.1978;8:21-29.18. Krugman S, GilesJP. Viral hepatitis: new lighton an old disease. JAMA. 1970;212:1019-1029.19. Ingelfinger FJ. Ethics of experiments on chil-dren. N Engl J Med. 1973;288:791-792.20. President's Commission for the Study of EthicalProblems in Medicine and Biomedical and BehavioralResearch. Making Health Care Decisions. Washing-ton, DC: US Government Printing Office; 1982.21. Statement on a Patient's Bill of Rights. Chi-cago, Ill: American Hospital Association; Novem-ber 17, 1972.22. Uniform Rights of the Terminally Ill Act. In:Handbook of Living Will Laws. New York, NY:Society for the Right to Die; 1987:135-147.23. Bouvia V Superior Court, 225 Cal Rptr 297(1986).24. Natanson V Kline, 350 P2d 1093 (Kan 1960).25. Appelbaum PS, Lidz CW, Meisel A. InformedConsent: Legal Theory and Clinical Practice. NewYork, NY: Oxford University Press Inc; 1987:chap 3.26. Faden RR, Beauchamp TL. A History andTheory of Informed Consent. New York, NY: Ox-ford University Press Inc; 1986.27. Canterbury V Spence, 464 F2d 772 (DC Cir 1972).28. Brock D. The ideal of shared decision-makingbetween physicians and patients. Kennedy Insti-tute J Ethics. 1991;1:28-47.29. Brock DW, Wartman SA. When competentpatients make irrational choices. N Engl J Med.1990;322:1595-1599.30. Eddy DM. Anatomy of a decision. JAMA.1990;263:441-443.31. Katz J. The Silent World of Doctor and Pa-tient. New York, NY: Free Press; 1984.32. Tannock IF, Boyer M. When is a cancer treat-ment worthwhile? N Engl J Med. 1990;322:989-990.33. Frankfurt H. Freedom of the will and the con-

cept of a person. J Philosophy. 1971;68:5-20.34. Taylor C. Human Agency and Language. NewYork, NY: Cambridge University Press; 1985:15-44.35. Dworkin G. The Theory and Practice of Au-tonomy. New York, NY: Cambridge UniversityPress; 1988:chap 1.36. Gorovitz S. Doctors' Dilemmas: Moral Con-flict and Medical Care. New York, NY: OxfordUniversity Press Inc; 1982:chap 6.37. Plato; Hamilton E, Cairns H, eds; EmanuelEJ, trans. Plato: The Collected Dialogues. Prince-ton, NJ: Princeton University Press; 1961:720 c-e.38. Walsh DC, Hingson RW, Merrigan DM, et al.The impact of a physician's warning on recoveryafter alcoholism treatment. JAMA. 1992;267:663-667.

Downloaded From: http://jama.jamanetwork.com/ on 07/21/2012