by scott d. halpern, george loewenstein, kevin g. … · other christian 1 2.0 4 9.3 2 5.0 ......

By Scott D. Halpern, George Loewenstein, Kevin G. Volpp, Elizabeth Cooney, Kelly Vranas,Caroline M. Quill, Mary S. McKenzie, Michael O. Harhay, Nicole B. Gabler, Tatiana Silva, Robert Arnold,Derek C. Angus, and Cindy Bryce

THE CARE SPAN

Default Options InAdvance Directives InfluenceHow Patients Set Goals ForEnd-Of-Life Care

ABSTRACT Although decisions regarding end-of-life care are personal andimportant, they may be influenced by the ways in which options arepresented. To test this hypothesis, we randomly assigned 132 seriously illpatients to complete one of three types of advance directives. Two typeshad end-of-life care options already checked—a default choice—but one ofthese favored comfort-oriented care, and the other, life-extending care.The third type was a standard advance directive with no options checked.We found that most patients preferred comfort-oriented care, but thedefaults influenced those choices. For example, 77 percent of patients inthe comfort-oriented group retained that choice, while 43 percent ofthose in the life-extending group rejected the default choice and selectedcomfort-oriented care instead. Among the standard advance directivegroup, 61 percent of patients selected comfort-oriented care. Our findingssuggest that patients may not hold deep-seated preferences regarding end-of-life care. The findings provide motivation for future researchexamining whether using default options in advance directives mayimprove important outcomes, including patients’ receipt of wanted andunwanted services, resource use, survival, and quality of life.

Most seriously ill patients valuecomfort and dignity over lifeextension,1 but routine careoften leads to treatmentoriented toward extending

life.2,3 Deviating from this life-extending normrequires that someone actively request or sug-gest doing so.4–6

Specifying one’s goals of care in the living willcomponent of an advance directive provides pa-tients with an opportunity to counter this ten-dency.7 However, the text and structure of com-monly used advance directives carry some of thesame implicit biases that tend to favor life

extension in the absence of advance directives.2

For example, in the widely used “Five Wishes”document,8 the option “I want to have life sup-port” is listed first in all three clinical scenarios,despite evidence that the ordering of choicesinfluences the choices selected and that theone presented first often dominates.9,10

Federal law encourages people to completeadvance directives,11 and their use appears tobe increasing.12 Given the importance of thechoices embedded in advance directives, it isessential to understand how the structure ofadvance directives affects patients’ stated pre-ferences. In studies providing hypothetical

doi: 10.1377/hlthaff.2012.0895HEALTH AFFAIRS 32,NO. 2 (2013): –©2013 Project HOPE—The People-to-People HealthFoundation, Inc.

Scott D. Halpern ([email protected]) is thedirector of the FosteringImprovement in End-of-LifeDecision Science Program,University of Pennsylvania, inPhiladelphia.

George Loewenstein is theHerbert A. Simon Professor ofEconomics and Psychology,Carnegie Mellon University, inPittsburgh, Pennsylvania.

Kevin G. Volpp is the directorof the Center for HealthIncentives and BehavioralEconomics, Leonard DavisInstitute of Health Economics,Wharton School, at theUniversity of Pennsylvania.

Elizabeth Cooney is theassistant director of theFostering Improvement inEnd-of-Life Decision ScienceProgram.

Kelly Vranas is a hospitalistin the medical intensive careunit at the PhiladelphiaVeterans Affairs MedicalCenter, in Pennsylvania.

Caroline M. Quill is a fellow inthe Division of Pulmonary,Allergy, and Critical Care atthe Hospital of the Universityof Pennsylvania, inPhiladelphia.

Mary S. Mckenzie is a fellowin the Division of Pulmonary,Allergy, and Critical Care atthe Hospital of the Universityof Pennsylvania.

February 2013 32:2 Health Affairs 1

End Of Life

directives to college students13 and elderly out-patients,14 Laura Kressel and colleagues foundthat people were significantly more likely to in-dicate preferences to forgo life-sustaining inter-ventions when completing advance directives inwhich forgoing these interventions was the de-fault than when they had to actively choose toforgo the interventions.14

These findings using hypothetical scenariosraise the possibility that people might not havewell-formulated, strongly held views on whatforms of care at the end of life will best promotetheir values. Indeed, insights from behavioraleconomics suggest that preferences for end-of-life care are likely to be “constructed” at the mo-ment people are asked to express them, ratherthan reflective of deeply ingrained preferences,because such choices are made infrequently andoften without opportunity for feedback onwhether the choices made promoted patients’interests.2,15 We tested this hypothesis by exam-ining whether default options influence thechoices of seriously ill patients in real advancedirectives, even after patients were alerted to thedefault option and their responses to it.

Study Data And MethodsDesign Overview, Setting, And ParticipantsWe randomly assigned real advance directives,which differed only in their embedded defaultoptions, to outpatients who were at least fiftyyears old, lacked prior advance directives, hadincurable diseases of the chest, and were notbeing considered for lung transplantation(Exhibit 1 and online Appendix Exhibit A).16

Patients were recruited in the thoracic oncologyand pulmonary outpatient clinics at theHospitalof theUniversity of Pennsylvania fromMay 2010through January 2012. The University ofPennsylvania Institutional Review Board ap-proved this study.Each week a research nurse screened elec-

tronic health records to identify eligible patientsand obtained permission from these patients’physicians to recruit them. The research nursethen met with potentially eligible patients inperson, described the study, reviewed the poten-tial benefits of completing advance directives,answered all questions, and provided patientswith a written informed-consent document.The consent form and nurse’s spoken guidance

Exhibit 1

Characteristics Of Patients In The Study To Assess The Affect Of Default Options In End-Of-Life Care Planning

Life-extensiondefault (n=49)

Standard advancedirective (n=43)

Comfort default(n=40)

Characteristic Number Percent Number Percent Number PercentAge (mean years) 64.6 — 64.4 — 64.8 —

Sex

Male 24 49.0 15 34.9 17 42.5Female 25 51.0 28 65.1 23 57.5

Race

Black or African American 11 22.4 14 32.6 10 25.0White or Caucasian 34 69.4 29 67.4 28 70.0Other/unknown 4 8.2 0 0.0 2 5.0

Religion

Catholic 12 24.5 10 23.3 12 30.0Protestant 15 30.6 13 30.2 13 32.5Other Christian 1 2.0 4 9.3 2 5.0Jewish 2 4.1 3 7.0 1 2.5Other faiths 13 26.5 10 23.3 6 15.0Unaffiliated 6 12.2 3 7.0 6 15.0

Diagnosis

Non–small cell lung cancer/other thoracicmalignancya 18 36.7 16.9 37.2 13 32.5

Chronic obstructive pulmonary disease 14 28.6 15 34.9 14 35.0Idiopathic pulmonary fibrosis 8 16.3 3 7.0 5 12.5Other incurable fibrotic lung diseases 6 12.2 7 16.3 5 12.5Otherb 3 6.1 2 4.7 3 7.5

SOURCE Authors’ analysis. aOther thoracic malignancies include malignant pleural effusion (for example, from breast cancer) andmesothelioma. bChronic obstructive asthma, bronchiectasis, cystic fibrosis, chronic pulmonary heart disease, other pulmonaryinsufficiency not elsewhere classified, other respiratory abnormalities, radiation pneumonitis, beryllium disease, lung involvementin systemic sclerosis, SLE (systemic lupus erythematosis), RA (rheumatoid arthritis).

Michael O. Harhay is adoctoral student inepidemiology, University ofPennsylvania.

Nicole B. Gabler is abiostatistician in the Centerfor Clinical Epidemiology andBiostatistics, University ofPennsylvania.

Tatiana Silva is a doctoralstudent in economics,University of Mannheim, inGermany.

Robert Arnold is the directorof the Institute for Doctor-Patient Communication,University of Pittsburgh, inPennsylvania.

Derek C. Angus is the directorof the Clinical Research,Investigation, and SystemsModeling of Acute IllnessesCenter, University ofPittsburgh.

Cindy Bryce is an associateprofessor of health policy andmanagement and of medicine,University of Pittsburgh.

End Of Life

2 Health Affairs February 2013 32:2

informed patients that different types of advancedirectives would be assigned by chance, that pa-tients in all groups could select or decline thesame interventions and treatment goals, andthat patients could change their choices atany time.

Interventions Consenting patientswere ran-domly assigned to complete one of three advancedirectives. All three were modified slightly fromthe professionally endorsed directive publishedby the Allegheny County Medical Society.17 Eachwas deemed consistent with Pennsylvania law bythe University of Pennsylvania Office of theGeneral Counsel. Each included an identical sec-tion for the designation of a durable power ofattorney for health care and a living will sectionthat was altered among the three versions, asdescribed below. Facsimiles of all of the advancedirective forms used in this study can be found inthe online Appendix.16

In all versions, patients were shown the exactsame options. The versions differed in whetheror not they contained a default—that is, whethera particular option was already marked with anX.When such a preselected defaultwas used, thatchoice was placed first of the three options.Patients first were asked to choose an overallplan of care that prioritized extending life orone that prioritized minimization of pain andsuffering.The precise language used was adapted from

that used by William Knaus in the Study toUnderstand Prognoses and Preferences for Out-comes and Risks of Treatments (SUPPORT), alandmark clinical trial.18 Patients could expandupon or clarify their choices by writing in theadditional space provided.Patients also were asked to choose whether or

not they wished to receive five potentially life-sustaining interventions, such as feeding tubeinsertion, if they became unable to make deci-sions themselves. Again, patients could opt notto make these choices, and they could expandupon their choices in the additional spaceprovided.Allocation of individual patients was deter-

mined by electronic number generation, withassignment probabilities of 33.3 percent to eachgroup. The research nurse recruiting patientsused a numbered packet containing the assignedadvance directive for each sequentially random-ized patient.One-third of patients were assigned to receive

a “comfort default advance directive” that de-faulted to the goal of relief of pain and sufferingand nonreceipt of life-sustaining interventions.Patients were instructed tomake other choices ifthey preferred by crossing out the default op-tions and initialing lines next to their selections.

Another one-third of patients, in the “life-extension default advance directive” group, re-ceived a directive that defaulted to the goal of lifeextension and receipt of potentially life-sustaining interventions. Again, patients wereshown how to make alternative selections.Patients in a third group were assigned to re-

ceive a “standard advance directive” that re-quired patients to actively choose their goalsof care or preferences for specific interventions.As in usual practice, if patients did not makeactive choices, surrogates and clinicians wouldmake decisions if patients lost capacity.All patients were encouraged to involve their

family members and physicians in completingtheir advance directives and to return them. Ifcompleted directives were not returned withinten days, the nurse telephoned patients up tothree times to remind them and, if desired, toschedule a clinic visit specifically for help in com-pleting the advance directive. For an advancedirective to be considered complete, the signa-tures of two witnesses or a notary were required,as per Pennsylvania law.Debriefing After patients returned completed

advance directives, one investigator called pa-tients to debrief19 them about the precisedifferences between their assigned advance di-rective and the advance directives that otherstudy participants received. The investigatorused an Institutional Review Board–approveddebriefing script.16 After explaining the goalsof the study, including the concept of defaultoptions, the investigator read patients’ choicesback to them and asked if they wished to makeany changes.Outcomes The primary outcome was the pro-

portion of patients across the three interventiongroups who selected a comfort-oriented goal ofcare. Given the propensity of the health caresystem to try to extend life in the absence of adirective otherwise,3,4,7 patients who selected alife-extending goal of care and those who didnot select an overall goal of care were jointlyconsidered to have not selected a comfort-oriented goal.We also assessed patients’ satisfaction with

their advance care planning two months afterthe debriefing. One of two authors blinded topatients’ group assignments contacted patientsby phone and administered the CanadianHealthcareEvaluation Project (CANHELP) ques-tionnaire. This thirteen-item questionnaire hasbeen validated for assessing satisfaction withend-of-life care planning.20,21

Statistical Analysis Our protocol specifieda primary analysis in which we included onlypatients who returned completed advance direc-tives. This analysis assessed the efficacy of


default options in advance directives among pa-tients who completed them.Because such analyses are susceptible to selec-

tion effects, intention-to-treat analyseswere alsoconducted in which all patients who were ran-domly assigned an advance directive were in-cluded in the analyses. These intention-to-treatanalyses were considered secondary becausethey are heavily biased toward the null. Specif-ically, they make the implausible assumptionthat all patients who did not complete advanceddirectives chose not to receive comfort-orientedplans of care and chose not to forgo any poten-tially life-sustaining interventions.Chi-square tests, Cochran-Armitage tests of

trend, and t tests were used as appropriate fortwo-group binary, three-group binary, and con-tinuous outcomes data, respectively. In secon-dary analyses, logistic regression models werecreated to adjust for chance imbalance acrossarms in patient-level variables.The clinic where patients were recruited for

the study was modeled as a random effect toadjust forpotentially correlatedoutcomeswithinclinics and to prevent confounding by clinic.22

Analyses were performed using the softwareStata, version 11.0, except for the Cochrane-Armitage tests, which were performed usingSAS, version 9.3.We targeted a sample size of ninety-three pa-

tients. If evenly distributed across the three or-dered groups, this samplewould yield 81 percentpower to declare significance at p ¼ 0:05 for adifference in the proportion of patients selectingcomfort-orientedgoals of careof 35percent.Thiscalculation assumed that the proportion in thestandard advance directive group (the middlegroup) would be roughly equidistant betweenthe proportions in the life-extension and com-fort default groups.Limitations This study was designed to enroll

a relatively small number of patients from a sin-gle health system. Although it was a randomizedtrial, the sample size doesnot allowus to rule outthe possibility that results were confounded byunmeasured variables, such as howwell patientsunderstood their illnesses or how often theyspoke with their physicians about prognosis.Second, we did not randomly assign the order-

ing of options within the standard advance di-rective but instead listed the comfort-orientedgoal of care and the options to forgo specificinterventions first for all patients in that arm.Because the first-listed option tends to be morecommonly selected,9,10 the observed differencesin selections between the comfort-default ad-vance directive and standard advance directivemay actually underestimate themagnitude of thedefault effect.

Third, we enrolled only patients with seriousthoracic diseases—primarily lung cancer and ob-structive and restrictive lung diseases. The find-ings from this group of patients might not begeneralizable to all patients or to patients withother specified health conditions.

Study ResultsOne thousand and sevety-nine patients werescreened and determined to be eligible for thisstudy based upon reviews of their electronichealth records. Of these, physicians requestedthat 43 not be contacted for study enrollmentat the time of their visit, and 332 missed or re-scheduled their visit or were not approached bythe research nurse because of scheduling con-flicts. Of the remaining 704 patients, 391(55.5 percent) were deemed ineligible whenin-person questioning revealed that they hadexisting advance directives.16 Thus, there were313 fully eligible patients, 132 (42.2 percent)of whom consented to participate.One patient was excluded because he com-

pletely rewrote the assigned advance directive,making choices that were not classifiable usingour coding scheme. The other 131 consented pa-tients were included in intention-to-treatanalyses.Completed advance directives were returned

by ninety-five patients (72.0 percent), only twoofwhom(2.1 percent) elected to reconsider theirchoices during the debriefing. One of these pa-tients returned a new advance directive in whichthe only change was his selection of a new dura-ble power of attorney; the other patient did notreturn a completed advance directive by the endof the study. Thus, ninety-four patients were in-cluded in per protocol analyses. Principal diag-noses and demographic characteristics amongthe 132 patients who consented to participateare shown in Exhibit 1.Goals Of Care The specific goals selected by

patients in each group are shown in onlineAppendix Exhibit F.16 This exhibit shows thatfifty-four (57.4 percent) of the ninety-four pa-tients who returned a completed advance direc-tive made a choice regarding their overall goalsthat differed from the default option.Nonetheless, in per protocol analyses, ad-

vance directive default options significantly in-fluenced the proportions of patients who chosecomfort-oriented goals of care. Among the com-fort default group, 77 percent retained comfortas their overall goal of care; 61 percent in thestandard advance directive group chose comfortas their goal; and 43 percent of those in the life-extension default group rejected the defaultchoice and indicated comfort as their primary

End Of Life


goal (p < 0:01 for test of trend; Exhibit 2).Intention-to-treat analyses produced a similar

trend in proportions: 50 percent in the comfortdefault group, 47 percent in the standard ad-vance directive group, and 31 percent in thelife-extension default group (p ¼ 0:04).16

In secondary analyses adjusting for race, sex,age, marital status, and the recruiting researchnurse, group assignment remained significantlyassociated with selections of comfort-orientedgoals of care inper-protocol analyses (odds ratio:2.12; 95% confidence interval: 1.21–3.72;p < 0:01). Intention-to-treat analyses yieldedsimilar but nonsignificant results (odds ratio:1.51; 95% confidence interval: 0.96–2.37;p ¼ 0:07). These results were robust to differentmodeling strategies.16

Among the thirty-six patients in whom educa-tion level was measured, patients who had neverattended college (10=17, 59 percent) were aslikely as patients who had attended college(10=19, 53 percent) to make selections otherthan the default option (odds ratio: 1.33; 95%confidence interval: 0.26–6.68).

Choices To Receive Potentially Life-Sustaining Interventions Patients complet-ing different advance directive versions alsohad different probabilities of choosing to forgopotentially life-sustaining interventions. For ex-ample, the proportions of patients choosing toforgo feeding-tube insertion were 54 percent inthe comfort-default group, 45 percent in thestandard advance directive group, and 26 per-cent in the life-extension default group (p ¼ 0:01for test of trend;Exhibit 3). Forcardiopulmonaryresuscitation, corresponding proportions were42 percent, 32 percent, and 20 percent(p ¼ 0:03).Similar but nonsignificant trends were noted

for intensive care unit admission (p ¼ 0:06),mechanical ventilation (p ¼ 0:06), and hemodi-alysis (p ¼ 0:08). Similar trends were also notedin intention-to-treat analyses, although only thetest for feeding-tube insertion was statisticallysignificant.16

Satisfaction With End-Of-Life CarePlanning Assessments of patients’ satisfactionwith end-of-life care planning were completedfor seventy-eight of the ninety-four patientswho completed advance directives (83.0 per-cent). Of the remaining sixteen patients, at leastfive died within twomonths; the remainder werelost to follow-up, and it was not known withcertainty that they were still living. Global andaverage satisfaction scores were high across thethree intervention groups (greater than 4.5 outof a possible 5) and no significant between- oramong-group differences were identified.16

DiscussionDefault options, or the events or conditions thatwill be set into place if no alternative is activelychosen,23,24 have been shown to influence deci-sions in domains as diverse as drivers’ insur-ance,25 retirement savings,26,27 influenza vaccina-tion,28 and organ donation.29,30 A hallmark ofdefaults is that they lead gently, without restrict-ing any options.Thus, when people have strong preferences,

such as for low-deductible health care insurance,they commonly make choices that counter thedefault. This is precisely what happened, for ex-ample, with the roll-out of Medicare Part D drugcoverage, when despite the default annualdeductible of $250, most Americans chose planswith no deductible at all.24

The study on which we report here shows thatdefault options have large influences on

Exhibit 2

Percentage Of Patients Choosing A Comfort-Oriented Goal Of Care (Per Protocol Population)

SOURCE Authors’ analysis. NOTES Standard error bars denote 95% confidence intervals..

Exhibit 3

Percentage Of Patients Choosing To Forgo Each Intervention, By Type Of Advance Directive(Per Protocol Population)

SOURCE Authors’ analysis. NOTES Standard error bars denote 95% confidence intervals. CPR iscardiopulmonary resuscitation.


seriously ill patients’ actual choices for healthcare interventions at the end of life. Overall,most patientswith terminal illnesses stated pref-erences for comfort-oriented care when offeredthe opportunity to state these preferences in realadvance directives, but the proportions of pa-tients choosing this option differed markedlyas a function of how the default was set.Importantly, these effects manifested even

after patients were made aware of the defaultsand shownhow theyhad responded to them, andafter it was made easy to choose counter to thedefault, whichmany patients did, particularly inthe life-extension default and standard advancedirective groups.16 Only 2.1 percent of patients inthis study elected to reconsider their selectionsafter being alerted to the manipulation of thedefault option, but ultimately these patientsdid not change their original selections.Furthermore, intentionally setting defaults wasnot associated with any changes in patients’ sat-isfaction with their choices, which suggests thatpatients were content to be guided in suchdecisions.Although one might expect patients to hold

strong prior preferences about end-of-life treat-ments, the findings that people were heavilyswayed by defaults, and content to be swayed,suggest that many seriously ill patients lackdeep-seated preferences about their end-of-lifecare.Despite the importanceof end-of-life healthcare decisions, it should come as no surprise thatmany patients lack well-established preferencesin this domain. People commonly lackpriorpref-erences for decisions that are made infrequentlyand provide few opportunities for learning afterthe fact whether the choices made did or did notpromote their goals.15 These are precisely thecharacteristics of end-of-life care choices.The power of defaults in determining stated

end-of-life care preferences underlines the im-portance of selecting defaults carefully withoutlimiting patients’ options.24,31 At the same time,clinicians should recognize that it is often diffi-cult to avoid defaults, and they should thereforeconsider carefully the predictable consequencesof which defaults are used or allowed toremain.2,24,32

Indeed, there is a default option embedded inthe standard approach to advance care planning:If patients donot actively choose specific goals inadvance directives, clinicians and surrogatesmustmakedecisions for patientswho lose capac-ity. Because many patients do not complete ad-vance directives or merely designate a durablepower of attorney, this “default to surrogate de-cision making” not only is prevalent, but alsocarries important bereavement consequencesfor family members.33–35

Given that most patients place a high priorityon not burdening their loved ones35–39 and thatmost patients in our study selected comfort-ori-ented goals even in the standard advance direc-tive group, there is reason to believe that thecurrent systemic default of life extension mightnot optimally promote patients’ wishes andvalues.These results can also be interpreted as evi-

dence that advance directive forms, absent awell-structured conversation among patients,family members, and providers, will not mean-ingfully promote patients’ values.40 A preferableapproach to advance care planning may be onethat relies not on forms but on carefully struc-tured conversations that explore patients’ valuesin the presence of their potential surrogate de-cision makers.41

For particular patients and families cared forby particular clinicians—for example, patientswith good access to physicians well trained inend-of-life communication and with familymembers experienced in advance care plan-ning—such coordinated communicationmay in-deed prove optimal. But it is uncertain whetherthis approach could be implemented across di-verse populations with differential access toskilled clinicians and experienced familymembers.By contrast, designing an advance directive

that would help the majority of patients makedecisions that promote their goals could providea way to improve end-of-life care more broadly,for all Americans. Recent evidence provides sub-stantial motivation to try, as observational stud-ies in the United States show that patients whocomplete advance directives less commonly diein a hospital,12,42–44 more often receive care con-sistent with their preferences,12 have surrogateswho are less likely to report concerns with com-munication near the end of life,43 and, in certainregions, receive less costly care.44

This study shows that using default options inadvance directives strongly influences the end-of-life care choices that people make withoutaffecting their satisfaction with their advancecare planning. Furthermore, because the effectsof defaults were identical even after patientswere directly told about the default, this studysuggests that for many patients, “preferences”for end-of-life care are not deeply held.This study also provides motivation for future

research examining whether using default op-tions in advance directives may improve impor-tant outcomes, including patients’ receipt ofwanted and unwanted services, survival, qualityof life, resource use, and family members’ be-reavement. If such research shows that settingdefaults in advance directives improves such

End Of Life


outcomeswhile adhering to ethical standards fordefault setting (including assurances that pa-tients are aware of the decisions to be madeand that countering the default can be done

easily),3,24,45,46 then the clinical use of default op-tions in advance directives may provide a novelway to improve end-of-life care for large popula-tions of seriously ill patients. ▪

Some portions of this article werepreviously presented at the 33rd AnnualMeeting of the Society for MedicalDecision-Making, Chicago, IL,October 22–24, 2011, and at theAcademyHealth Annual Research

Meeting, Orlando, FL, June 24–26, 2012.This work was supported by researchgrants from the Greenwall Foundation,Kornfeld Foundation, Leonard DavisInstitute of Health Economics at theUniversity of Pennsylvania, and Center

for Excellence in Cancer CommunicationResearch of the Universities of Michiganand Pennsylvania. The authors thankSusan Metzger for her assistance withpatient recruitment.

NOTES

1 Fried TR, Bradley EH, Towle VR,Allore H. Understanding the treat-ment preferences of seriously ill pa-tients. N Engl J Med. 2002;346(14):1061–6.

2 Halpern SD. Shaping end-of-lifecare: behavioral economics and ad-vance directives. Semin in RespirCrit Care Med. 2012;33:393–400.

3 Blinderman CD, Krakauer EL,Solomon MZ. Time to revise theapproach to determining cardiopul-monary resuscitation status. JAMA.2012;307(9):917–8.

4 White DB, Arnold RM. The evolutionof advance directives. JAMA. 2011;306(13):1485–6.

5 Schroeder SA. Personal reflectionson the high cost of Americanmedicalcare. Arch Intern Med. 2011;171:722–7.

6 Neuman MD. Surgeons’ decisionsand the financial and human costs ofmedical care. N Engl J Med.2010;363(25):2382–3.

7 Halpern SD, Emanuel EJ. Advancedirectives and costs of care: greaterclarity and perpetual confusion.Arch Intern Med. 2012;172:266–8.

8 Aging with Dignity. Five wishes[Internet]. Tallahassee (FL): Agingwith Dignity; [cited 2012 Mar 13].Available from: http://www.agingwithdignity.org/forms/5wishes.pdf

9 Krosnick JA, Alwin DF. An evalu-ation of a cognitive theory of re-sponse-order effects in surveymeasurement. Public OpinionQuarterly. 1987;51(2):201–19.

10 Bishop GF, Smith AE. Response-order effects in public opinion sur-veys: the plausibility of rival hy-potheses [Internet]. Papers pre-sented at: 52nd Annual Conferenceof the American Association forPublic Opinion Research, SurveyResearch Methods Section, Norfolk,Virginia, 1997 May 15–18 [cited 2013Jan 22]. Available from: http://www.amstat.org/sections/srms/Proceedings/papers/1997_179.pdf

11 Wolf SM, Boyle P, Callahan D, FinsJJ, Jennings B, Nelson JL, et al.Sources of concern about the PatientSelf-Determination Act. N Engl JMed. 1991;325(23):1666–71.

12 Silveira MJ, Kim SYH, Langa KM.Advance directives and outcomes of

surrogate decision making beforedeath. N Engl J Med. 2010;362(13):1211–8.

13 Kressel LM, Chapman GB. The de-fault effect in end-of-life medicaltreatment preferences. Med DecisMaking. 2007;27(3):299–310.

14 Kressel LM, Chapman GB, LeventhalE. The influence of default optionson the expression of end-of-lifetreatment preferences in advancedirectives. J Gen Intern Med.2007;22:1007–10.

15 Slovic P. The construction of pref-erence. Am Psychol. 1995;50:364–71.

16 To access the Appendix, click on theAppendix link in the box to the rightof the article online.

17 Allegheny County Medical Society.Advance health care directive: healthcare power of attorney and living will[Internet]. Pittsburgh (PA): TheSociety; 2009 [cited 2013 Jan 9].Available from: http://www.acba.org/Public/Legal-information/LivingWillPoweroAttyform.pdf

18 Knaus WA, Connors AF, Dawson NV,Desbiens NA, Fulkerson WJ,Goldman L, et al. A controlled trial toimprove care for seriously ill hospi-talized-patients—the Study toUnderstand Prognoses andPreferences for Outcomes and Risksof Treatments (SUPPORT). JAMA.1995;274(20):1591–8.

19 Miller FG, Gluck JP, Wendler D.Debriefing and accountability indeceptive research. Kennedy InstEthics J. 2008;18(3):235–51.

20 Heyland DK, Cook DJ, Rocker GM,Dodek PM, Kutsogiannis DJ,Skrobik Y, et al. The developmentand validation of a novel question-naire to measure patient and familysatisfaction with end of life care: theCanadian Health care EvaluationProject (CANHELP) Questionnaire.Palliat Med. 2010;24(7):682–95.

21 Heyland DK, Frank C, Tranmer J,Paul N, Pichora D, Jiang XR, et al.Satisfaction with end-of-life care: alongitudinal study of patients andtheir family caregivers in the lastmonths of life. J Palliat Care.2009;25(4):245–56.

22 Localio AR, Berlin J, Ten Have TR,Kimmel SE. Adjustments for centerin multicenter studies: an overview.

Ann Intern Med. 2001;135(2):112–23.

23 Carroll GD, Choi JJ, Laibson D,Madrian BC, Metrick A. Optimaldefaults and active decisions. Q JEcon. 2009;124(4):1639–74.

24 Halpern SD, Ubel PA, Asch DA.Harnessing the power of defaultoptions to improve health care. NEngl J Med. 2007;357(13):1340–4.

25 Johnson EJ, Hershey JC, Meszaros J,Kunreuther H. Framing, probabilitydistortions, and insurance decisions.J Risk Uncertain. 1993;7:35–51.

26 Choi JJ, Laibson D, Madrian BC,Metrick A. Defined contributionpensions: plan rules, participantdecisions, and the path of least re-sistance. In: Poterba JM, editor. Taxpolicy and the economy. Vol. 16.Cambridge (MA): MIT Press; 2002.p. 67–113.

27 Madrian BC, Shea DF. The power ofsuggestion: inertia in 401(k) par-ticipation and savings behavior. Q JEcon. 2001;116(4):1149–87.

28 Chapman GB, Li M, Colby H,Yoon H.Opting in vs. opting out of influenzavaccination. JAMA. 2010;304(1):43–4.

29 Horvat LD, Cuerden MS, Kim SJ,Koval JJ, Young A, Garg AX.Informing the debate: rates of kidneytransplantation in nations with pre-sumed consent. Ann Intern Med.2010;153(10):641–68.

30 Johnson EJ, Goldstein D. Do de-faults save lives? Science. 2003;302(5649):1338–9.

31 Loewenstein G, Brennan T, VolppKG. Asymmetric paternalism to im-prove health behaviors. JAMA.2007;298(20):2415–7.

32 Camerer C, Issacharoff S,Loewenstein G, O’Donoghue T,Rabin M. Regulation for conserva-tives: behavioral economics and thecase for asymmetric paternalism.Univ PA Law Rev. 2003;151(3):1211–54.

33 Azoulay E, Pochard F, Kentish-Barnes N, Chevret S, Aboab J, AdrieC, et al. Risk of post-traumatic stresssymptoms in family members of in-tensive care unit patients. Am JRespir Crit Care Med. 2005;171(9):987–94.

34 Wendler D, Rid A. Systematic review:


the effect on surrogates of makingtreatment decisions for others. AnnIntern Med. 2011;154(5):336–46.

35 Wright AA, Zhang BH, Ray A, MackJW, Trice E, Balboni T, et al.Associations between end-of-lifediscussions, patient mental health,medical care near death, and care-giver bereavement adjustment.JAMA. 2008;300(14):1665–73.

36 Terry PB, Vettese M, Song J, FormanJ, Haller KB, Miller DJ, et al. End-of-life decision making: when patientsand surrogates disagree. J ClinEthics. 1999;10(4):286–93.

37 Singer PA, Martin DK, Lavery JV,Thiel EC, Kelner M, MendelssohnDC. Reconceptualizing advance careplanning from the patient’s per-spective. Arch Intern Med. 1998;158(8):879–84.

38 Mead GE, O’Keeffe ST, Jack CI,Maèstri-Banks AM, Playfer JR, Lye

M. What factors influence patientpreferences regarding cardiopulmo-nary resuscitation? J R CollPhysicians Lond. 1995;29(4):295–8.

39 Steinhauser AE, Christakis NA, ClippEC, McNeilly M, McIntyre L, TulskyJA. Factors considered important atthe end of life by patients, family,physicians, and other care providers.JAMA. 2000;284(19):2476–82.

40 Sudore RL, Fried TR. Redefining the“planning” in advance care plan-ning: preparing for end-of-life deci-sion making. Ann Intern Med.2010;153(4):256–61.

41 Kirchhoff KT, Hammes BJ, Kehl KA,Briggs LA, Brown RL. Effect of adisease-specific planning interven-tion on surrogate understanding ofpatient goals for future medicaltreatment. J Am Geriatr Soc. 2010;58(7):1233–40.

42 Degenholtz HB, Rhee Y, Arnold RM.

Brief communication: the relation-ship between having a living will anddying in place. Ann Intern Med.2004 Jul;141(2):113–7.

43 Teno JM, Gruneir A, Schwartz Z,Nanda A, Wetle T. Association be-tween advance directives and qualityof end-of-life care: a national study. JAm Geriatr Soc. 2007;55(2):189–94.

44 Nicholas LH, Langa KM, IwashynaTJ, Weir DR. Regional variation inthe association between advance di-rectives and end-of-life medicareexpenditures. JAMA. 2011;306(13):1447–53.

45 Blumenthal-Barby JS, Burroughs H.Seeking better health care outcomes:the ethics of using the “nudge.” Am JBioeth. 2012;12(2):1–10.

46 Quill CM, Halpern SD. Decipheringthe appropriateness of defaults: theneed for domain-specific evidence. JMed Ethics. 2012;38(12):721–2.

ABOUT THE AUTHORS: SCOTT D. HALPERN, GEORGE LOEWENSTEIN,KEVIN G. VOLPP, ELIZABETH COONEY, KELLY VRANAS,CAROLINE M. QUILL, MARY S. MCKENZIE, MICHAEL O. HARHAY,NICOLE B. GABLER, TATIANA SILVA, ROBERT ARNOLD,DEREK C. ANGUS & CINDY BRYCE

Scott D. Halpern isthe director of theFosteringImprovement inEnd-of-LifeDecision ScienceProgram, Universityof Pennsylvania.

In this month’s Health Affairs, ScottHalpern and coauthors report ontheir study testing whether end-of-life care decisions are influencedby the ways in which options arepresented. They randomly assigned132 seriously ill patients tocomplete one of three types ofadvance directives that had eitherdifferent “default” choices alreadyselected for comfort-oriented careor life-extending care, or nooptions checked at all. The resultsindicated that the defaults clearlyinfluenced patients’ choices andthat patients might not hold deep-seated preferences regarding end-

of-life care, the authors write. Theauthors recommend furtherresearch to determine whether useof advance directives with thesedefault options leads to outcomesimportant to patients, families, andsociety.Halpern is the director of the

Fostering Improvement in End-of-Life Decision Science Program anddeputy director of the Center forHealth Incentives and BehavioralEconomics at the Leonard DavisInstitute of Health Economics,Wharton School, University ofPennsylvania. He is also anassistant professor of medicine,specializing in pulmonary andcritical care, epidemiology, andmedical ethics and health policy atthe university.Halpern is a member of the

Secretary of Health and HumanServices Advisory Committee onBlood and Tissue Safety andAvailability and of the TechnicalAdvisory Committee to the

Scientific Registry of TransplantRecipients. He received a RobertWood Johnson Foundation YoungLeader Award in 2012. Halpernearned master’s degrees inbioethics and clinicalepidemiology, a doctorate inepidemiology, and a medical degreefrom the University ofPennsylvania.

GeorgeLoewenstein is theHerbert A. SimonProfessor ofEconomics andPsychology,Carnegie MellonUniversity.

George Loewenstein is theHerbert A. Simon Professor ofEconomics and Psychology atCarnegie Mellon University andone of the founders of the fields ofbehavioral economics and

End Of Life


neuroeconomics. He is the formerpresident of the Society forJudgment and Decision Makingand a member of the Russell SageFoundation Behavioral EconomicsRoundtable, and he served on theboard of the Society forNeuroeconomics. Loewensteinearned a doctorate in economicsfrom Yale University.

Kevin G. Volpp isthe director of theCenter for HealthIncentives andBehavioralEconomics.

Kevin Volpp is the director of theCenter for Health Incentives andBehavioral Economics and of theNational Institute on Aging–fundedPenn–Carnegie Mellon UniversityRoybal P30 Center on BehavioralEconomics and Health at theUniversity of Pennsylvania. He isalso a tenured professor ofmedicine in the Perelman School ofMedicine and of the WhartonSchool’s Health Care ManagementDepartment, both at Penn, as wellas a staff physician at thePhiladelphia Veterans Affairs (VA)Medical Center and core facultymember of its Center for HealthEquity Research and Promotion.Volpp earned a doctorate in healtheconomics, public policy, andmanagement and a medical degreefrom the University ofPennsylvania.

Elizabeth Cooney isthe assistantdirector of theFosteringImprovement inEnd-of-LifeDecision ScienceProgram.

Elizabeth Cooney is the assistantdirector of the FosteringImprovement in End-of-Life

Decision Science Program. She hasmanaged several clinical trialsaddressing public health issues,including end-of-life care anddecision making, substance abuse,and college student health. Cooneyearned a master’s degree in publichealth from Boston University.

Kelly Vranas is ahospitalist in themedical intensivecare unit at thePhiladelphia VAMedical Center.

Kelly Vranas is a hospitalist inthe medical intensive care unit atthe Philadelphia VA MedicalCenter. She completed herinternship and residency ininternal medicine at the Hospitalof the University of Pennsylvaniaand will start her fellowship inpulmonary and critical caremedicine at Stanford University inJune 2013. She earned a medicaldegree from Cornell University.

Caroline M. Quill isa fellow at theHospital of theUniversity ofPennsylvania.

Caroline Quill is a fellow in theDivision of Pulmonary, Allergy,and Critical Care at the Hospital ofthe University of Pennsylvania,where she also completed herinternship and residency ininternal medicine. Her researchfocuses on sources of variation inend-of-life care, particularly amongintensive care units and providers.She earned a medical degree fromthe University of Rochester and isin the process of completing amaster’s degree in health policyresearch at the University ofPennsylvania.

Mary S. McKenzieis a fellow at theHospital of theUniversity ofPennsylvania.

Mary McKenzie is also a fellowin the Division of Pulmonary,Allergy, and Critical Care at theHospital of the University ofPennsylvania. She completed herinternship and residency ininternal medicine at the BethIsrael–Deaconess Medical Center.McKenzie holds a master’s degreein death and society from theUniversity of Reading, England,and a medical degree from OregonHealth and Science University. Sheis in the process of earning amaster’s degree in health policyresearch at the University ofPennsylvania.

Michael O. Harhayis a doctoralstudent inepidemiology,University ofPennsylvania.

Michael Harhay is a doctoralstudent in epidemiology at theUniversity of Pennsylvania. Hisresearch focuses on methodologicalstrategies to improve the designand analysis of clinical trials andobservational studies in criticalcare medicine. He holds master’sdegrees in demography, healthpolicy, and bioethics, all from theUniversity of Pennsylvania.


Nicole B. Gabler isa biostatistician inthe Center forClinicalEpidemiology andBiostatistics.

Nicole Gabler is a biostatisticianin the Center for ClinicalEpidemiology and Biostatistics atthe University of Pennsylvania. Herresearch interests include criticalcare health services research,treatment-effect heterogeneity,pulmonary arterial hypertension,and end-of-life decision making.She earned master’s degrees inpublic health and healthadministration from TulaneUniversity and a doctorate inepidemiology from the Universityof California, Davis.

Tatiana Silva is adoctoral student ineconomics,University ofMannheim.

Tatiana Silva is a doctoralstudent in economics at theUniversity of Mannheim. Shepreviously worked as a researchcoordinator at the Perelman Schoolof Medicine and the Wharton

School, both at Penn. She has amaster’s degree in publicadministration from the Universityof Pennsylvania.

Robert Arnold isthe director of theInstitute forDoctor-PatientCommunication,University ofPittsburgh.

Robert Arnold is the director ofthe Institute for Doctor-PatientCommunication and codirector ofthe Institute to Enhance PalliativeCare, both at the University ofPittsburgh. He is also a professorin the Division of General InternalMedicine, Department of Medicine,University of Pittsburgh, and in theuniversity’s Center for Bioethicsand Health Law. Arnold earned amedical degree from the Universityof Missouri–Kansas City.

Derek C. Angus isthe director of theClinical Research,Investigation, andSystems Modelingof Acute IllnessesCenter, Universityof Pittsburgh.

Derek Angus is a professor inand chair of the Department ofCritical Care Medicine and director

of the Clinical Research,Investigation, and SystemsModeling of Acute Illnesses Centerat the University of Pittsburgh. Heholds secondary appointments inmedicine, health policy andmanagement, and clinical andtranslational science at theuniversity. He is editor of theJournal of the American MedicalAssociation’s Caring for theCritically Ill section. Angus earneda master’s degree in public healthfrom the University of Pittsburghand a medical degree from theUniversity of Glasgow, Scotland.

Cindy Bryce is anassociate professorof health policy andmanagement and ofmedicine, Universityof Pittsburgh.

Cindy Bryce is an associateprofessor of health policy andmanagement and of medicine atthe University of Pittsburgh. She isalso a core member of theuniversity’s Center for Research onHealth Care and of the ClinicalResearch, Investigation, andSystems Modeling of AcuteIllnesses Center. She holds both amaster’s degree and a doctorate inpolicy analysis from CarnegieMellon University.

End Of Life


by scott d. halpern, george loewenstein, kevin g. … · other christian 1 2.0 4 9.3 2 5.0 ......

Documents