family perspectives on aggressive cancer care near the end of life

8/19/2019 Family Perspectives on Aggressive Cancer Care Near the End of Life

1/17

Family Perspectives onAggressive Cancer Care

Near the End of LifeAlexiA. Wright, MD, MPH; Nancy L. Keating,MD, MPH; John Z. Ayanian,MD,MPP; ElizabethA. Chrischilles, PhD; Katherine L. Kahn, MD;

ChristineS. Ritchie, MD, MSPH;JaneC. Weeks, MD, MSc†; CraigC. Earle,MD, MSc; Mary B.Landrum,PhD

IMPORTANCE Patientswith advanced-stage cancer are receiving increasingly aggressive

medical care near death, despite growing concerns thatthis reflects poor-quality care.

OBJECTIVE To assess the association of aggressive end-of-life care with bereaved familymembers’ perceptions of the quality of end-of-life care and patients’ goal attainment.

DESIGN, SETTING,ANDPARTICIPANTS Interviewswith 1146 family members of Medicare

patients with advanced-stagelung or colorectal cancer in theCancer Care Outcomes

Research and Surveillance study (a multiregional, prospective, observational study) who died

by theend of 2011 (median, 144.5 days after death;interquartile range,85.0-551.0 days).

EXPOSURES Claims-based quality measures of aggressive end-of-life care (ie, intensive care

unit [ICU] admission or repeated hospitalizations or emergency department visits during the

last month of life; chemotherapy2 weeks ofdeath;no hospice or3 days of hospice

services; and deaths occurring in the hospital).

MAINOUTCOMESANDMEASURES Family member–reportedquality rating of “excellent” for

end-of-life care. Secondary outcomes included patients’goal attainment (ie, end-of-life care

congruentwith patients’ wishes and locationof death occurred in preferred place).

RESULTS Of 1146 patients with cancer (median age,76.0 years [interquartile range, 65.0-87.0

years]; 55.8% male), bereaved family members reported excellent end-of-life care for 51.3%.

Family members reported excellentend-of-life care more oftenfor patients who received

hospice care forlonger than 3 days (58.8% [352/599]) than those whodid not receive

hospice care or received 3 or fewer days (43.1% [236/547]) (adjusted difference, 16.5

percentage points[95% CI,10.7 to 22.4percentage points]). In contrast,family members of

Supplementalcontent at

jama.com

CMEQuizat

jamanetworkcme.com and

CMEQuestionspage 300

Research

Original Investigation

http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2015.18604&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2015.18604http://www.jama.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2015.18604http://www.jamanetworkcme.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2015.18604http://www.jamanetworkcme.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2015.18604http://www.jama.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2015.18604http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2015.18604&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2015.18604


2/17

Patients with advanced-stage cancer receive aggressive

medical care at the end of life, despite increasing evi-

dencethat high-intensitytreatments maynot be associ-

ated with better patient quality of life,outcomes, or caregiver

bereavement. TheNational Quality Forum andthe American

Society of Clinical Oncology have jointly endorsed several in-

dicatorsof overlyaggressive end-of-lifecare (developedbyEarle

etal), including repeated hospitalizations,emergencydepart-

ment visits,or admission to an intensivecare unit (ICU) within

the last month of life,patient receipt of chemotherapy within

weeksprior to death, and late or absent hospicereferrals.

The end-of-life care measures of the National Quality

Forum and the American Society of Clinical Oncology are in-

tegratedintothe Society’s Quality Oncology Practice Initiative

and will likely become more important because the Patient

Protection and Affordable Care Act mandates public report-

ing of quality metrics and has linked incentive and penalty

payments to similar indicators (eg, the Centers for Medicare

& Medicaid Services Oncology Care Model). Improving end-

of-life care offers an opportunity to improve the dying expe-

rience of patientswith cancer, more effectivelyuse resources,

and potentially prolong survival.

Few studies haveexaminedwhether theseaggressive end-of-life care measures reflect patients’ preferences or bereaved

family members’ perceptions and expectations of the quality

of end-of-life care. In this study, we assessed the relationship

betwe en aggres sive end- of-li fe care and family member–

reported qualityratings of end-of-life careusing data fromthe

Cancer Care Outcomes Research and Surveillance (CanCORS)

Consortium.We alsoexamined patientgoal attainment,which

wasbased on () whether patients treated withaggressive care

near death received carethat was congruent withtheir prefer-

ences and () if theirdeath occurred in theirpreferred place.

Methods

vited to participate. Patients who were alive for the baseline

interview were asked to identify a primary family memberor

friend “familiarwith yourcare since diagnosis” and a second-

ary respondent (in case the first could not be reached). Be-

cause % of interviews were completed by family (defined

as spouse or partner, child, other relative, or friend), we will

hereafter refer to respondents as family members.

Wemeasuredthe intensity of end-of-life care received by

participants enrolled in the traditional fee-for-service Medi-

careplanby obtaining claims data withapproval fromthe Cen-

ters for Medicare & Medicaid Services privacy board. Poten-

tially eligible enrolleeswere linked using a common matching

algorithm based on Social Security number, date of birth,and

sex, which matched .% of CanCORS enrollees aged years or older to Medicare data.

StudyPopulation

The CanCORS participants who were diagnosed or experi-

encedadvanced lungor colorectalcancer recurrence, died by

, were aged years or older, and continuously enrolled

in Medicare partsA andB fee-for-service for monthsor lon-

ger before death were included.

Outcome Variables

We examined family member–reported ratings of the quality

of end-of-life care and measures of patient goal attainment.

To assess care quality, family members were asked: “Overall,

howwould yourate thecare received [atthe last place where

carewas provided]?Would yousay it wasexcellent, verygood,

good, fair, or poor?” We defined high-quality end-of-life care

as that which family members rated as excellent.

We assessed patient goal attainment using measures.

First, family members were asked “In your opinion, to what

extentwere[the patient’s] wishes followed in themedicalcare

received in the last month of life?” Goal attainment was de-

fined as care that “followed patients’ wishes a great deal.” In

FamilyPerspectives on AggressiveCancer Care Near theEnd of Life Original Investigation Research


3/17

literaturereviews andfocusgroupswith patients, families, and

anexpert panel.Weidentifiedaggressiveend-of-lifecare using

Medicare claims data., We examined deaths that occurred

inhospitals using family member–reportedlocation ofdeath be-

cause hospital deaths have been associated withworse family

member–reported patient quality of life near death compared

with homedeaths, andincreased psychiatric morbidity among

caregivers.

Control Variables

We obtained information about patients’ sociodemographic

characteristics (age, marital status, education, and income),

overall health status, comorbid conditions, depression, and

treatment preferences from family members who completedthe after-deathinterview. Race/ethnicity was analyzed as a po-

tential determinant of the quality of end-of-life care because

patients mayvary in theirtreatmentpreferences, advance care

planning, and health careuse. Cancer stage at diagnosisand

type was obtained through medical record abstraction or, if

unavailable, from cancer registries.

We also adjusted for geographic region and enrollment in

an integratedhealthsystem, which included sites andKaiser

Permanente of Northernor SouthernCalifornia. Inaddition,weadjustedfor time from diagnosisto death as a linearvariable,

family respondent(eg, spouse or partner, child, other relative,

orfriend),and time from deathto interviewcompletionas a lin-

ear variable. In additional analyses,we tested for interactions

betweentime from deathto interviewcompletionand respon-

dentrole; noneof theinteractions werestatisticallysignificant,

so the models included only themain effects.

StatisticalAnalyses

Descriptive statistics wereused tocharacterizefamilymember–

rated quality of end-of-life care, patient goal attainment, and

rates of aggressive end-of-life care according to patient pref-

erences. We examined associations between aggressive end-

good” vs “good, fair, and poor” and as an ordinal model. We

also fit models that included family members’ contact with

patients during the last week as a proxy for closeness. We

repeated these analyses among a subset of the total cohort

with information from the baseline interview and the -year

follow-up interview, both with and without family member–

reported frequency of discussing important medical deci-

sions with the patient. Next, we examined the effect of

including patients’ symptoms at the end-of-life (eg, pain and

difficulty breathing) and how well these were controlled

because uncontrolled symptoms often precipitate hospital-

izations near death. In addition, we repeated all analyses

stratified by disease site.

Two-sided P values % across all sites for pa-

tients aged ≥years). We excluded patientsfor whomwe

did not have an after-death interview from a family member

(theafter-deathinterviews wereavailable for .% of poten-

tially eligiblepatients). We also excluded patientswho were

not continuously enrolled in Medicare parts A and B fee-for-

service as of monthsbefore death and patients forwhom

thefamily memberdid notratethe quality ofend-of-lifecare,

leaving a final study population of patients (median age,

. years [interquartile range, .-. years]).

Most of the study population had stage IV disease at di-

agnosis(Table).Lungcancer decedentshad much shorter me-

Research Original Investigation FamilyPerspectives onAggressive CancerCare Near theEnd of Life


4/17

Table 1. PatientCharacteristicsa

Total(N = 1146)

Lung Cancer(n = 886)

Colorectal Cancer(n = 260)

Male sex 639 (55.8) 514 (58.0) 125 (48.1)Age at death, y

65-69 204 (17.8) 170 (19.2) 34 (13.1)

70-74 273 (23.8) 222 (25.1) 51 (19.6)

75-79 260 (22.7) 215 (24.3) 45 (17.3)

≥80 409 (35.5) 279 (31.5) 130 (50.0)

Race/ethnicity

White 923 (80.5) 726 (81.9) 197 (75.8)

Black 85 (7.4) 54 (6.1) 31 (11.9)

Hispanic 58 (5.1) 45 (5.1) 13 (5.0)

Asian/Pacific Islander 41 (3.6) 31 (3.5) 10 (3.9)

Otherb 39 (3.4) 30 (3.4) 9 (3.5)

Education level

≤High school graduate 285 (25.1) 231 (26.3) 54 (21.1)

High school graduate 658 (57.9) 514 (58.4) 144 (56.3)

College graduate 193 (17.0) 135 (15.3) 58 (22.7)

Household income, $


5/17

Table2. FamilyMemberCharacteristicsa

Total(N = 1146)

Lung Cancer(n = 886)

Colorectal Cancer(n = 260)

Family member relationship with patientb

Spouse or partner 498 (43.5) 401 (45.3) 97 (37.3)

Child 474 (41.4) 364 (41.1) 110 (42.3)

Other relative 113 (9.9) 75 (8.5) 38 (14.6)

Friend or other 61 (5.3) 46 (5.2) 15 (5.8)

Accompanied patient to physician appointments

Always 449 (58.0) 388 (58.3) 61 (56.5)

Usually 136 (17.6) 121 (18.2) 15 (13.9)

Sometimes 127 (16.4) 107 (16.1) 20 (18.5)

Never 62 (8.0) 50 (7.5) 12 (11.1)

Discussed important medical decisions regarding cancer

Always 464 (60.0) 415 (62.3) 49 (45.4)

Usually 92 (11.9) 81 (12.2) 11 (10.2)

Sometimes 136 (17.6) 109 (16.4) 27 (25.0)

Never 82 (10.6) 61 (9.2) 21 (19.4)

Male sex, No./total (%) 189/774 (24.4) 158/666 (23.7) 31/108 (28.7)

Race/ethnicity

White 622 (83.4) 548 (85.0) 74 (73.3)

Black 50 (6.7) 36 (5.6) 14 (13.9)

Hispanic 40 (5.4) 33 (5.1) 7 (6.9)

Asian/Pacific Islander 34 (4.6) 28 (4.3) 6 (5.9)

Days of co ntact w ith p atien t d ur ing l ast w k of l ife, med ian ( IQR) 7.0 (6 .0- 7.0) 7.0 (6 .0- 7.0) 1.0 ( 1.0- 7.0)

Time between death and after-death interview,median (IQR), d

144.5 (85.0-551.0) 131.0 (78.0-246.0) 539.0 (116.5-1560.0)

Abbreviation:IQR, interquartilerange.

a Data areexpressed as No. (%)unless otherwiseindicated. There were missing

datafor thefollowing characteristics: accompaniedpatient to physician

appointments (n = 372), discussed important medical decisions regarding

cancer (n = 372),male sex (n = 372),race/ethnicity (n = 400), and daysof

contact with patient duringlast week of life(n = 74).

b Respondents wereasked to describe his/her relationshipwith the patient and

reporthow many days he/shewas incontact with thepatientduringthelast

weekof life;however, questions aboutfrequencyof accompanying patient to

physician appointments, discussing important medical decisions regarding

cancer,sex, and race/ethnicitywere only askedduring the baseline decedent

interview and the decedentinterview (conducted1 yearafter diagnosis and

not thedecedentinterviewconducted5-7 yearsafter diagnosis).

Table3. FamilyMember–ReportedQualityofEnd-of-Life Careand PatientGoalAttainment

Research Original Investigation FamilyPerspectives onAggressive CancerCare Near theEnd of Life


6/17

Table4.RelationshipBetweenFamilyMember–ReportedTreatmentPreferencesof thePatient andthePatient’sReceipt ofAggressiveEnd-of-LifeCarea

Type of Aggressive End-of-Life CareTotal(N = 1146)

Among Patients Reporting Preferenceb

P Valuec

Extend Life

vs RelieveSymptoms(n = 332)

Relieve Symptoms

vs Extend Life(n = 776)

During last mo of life

Intensive care unit admission 151 (13.2) 53 (16.0) 94 (12.1) .08

≥2 Hospitalizations 70 (6.1) 32 (9.6) 37 (4.8) .002

≥2 Emergency department visits 90 (7.9) 31 (9.3) 57 (7.4) .26

Chemotherapy ≤2 wk before death 75 (6.5) 41 (12.4) 30 (3.9)


7/17

(.% [/]) and in those who did not receive chemo-

therapy (.% [/]) (adjusteddifference, −. percent-

age points [% CI, −. to . percentage points]).

reported contact with patients during the last week or fre-

quencyof discussionsof importantmedicaldecisionswith pa-

tients. Wealso found few differences in results from analyses

Table6.AssociationsBetweenAggressiveEnd-of-LifeCareand PatientReceipt ofGoal-ConcordantEnd-of-Life Care

Type ofAggressiveEnd-of-Life Care

Family Member Reported That End-of-LifeWishes of Patient Were Followeda

P Valued

Family Member Reported That PatientDied in Preferred Placec

P Valued

UnadjustedNo./Total (%)(n = 930)

Adjusted UnadjustedNo./Total (%)(n = 774)

Adjustedb

%Risk Difference(95% CI)b %

Risk Difference(95% CI)

During last mo of life

Intensive care unit admission

No 662/814 (81.3) 80.0−3.5 (−12.5 to 5.4) .43

404/682 (59.2) 59.4−22.8 (−33.5 to −12.0)


8/17

with advanced-stagecancer receivedcarecongruentwith their

preferences. Our findings suggest that efforts to increase ear-

lier hospice enrollment and avoidance of ICU admissionsand

hospital deaths might improve the quality of end-of-lifecare.

Forexample, implementationof multifaceted approaches (eg,

enhancedcounseling of patients and families, earlypalliative

care referrals, and an audit and feedback system to monitor

physicians’ use of aggressive end-of-life care) might result in

more preference-sensitive care for patients and overall im-

proved quality of end-of-life care.,

To our knowledge, this study is one of the first empirical

validationsof theseaggressive end-of-lifecare indicatorsusing

patient- and family member–centered outcomes in a popula-

tion-based cohort. These findings may have policy implica-tions because administrative databaseindicatorsare becom-

ing increasingly feasible to assess with the expansion of

electronichealth records underthe Health Information Tech-

nology for Economic and Clinical Health Act. This is particu-

larly noteworthy given increasing evidence that physicians’

characteristics andbeliefs areone of thestrongest predictors

of end-of-life care,, suggesting that modifications to phy-

sicians’ practices may result in significant improvements.

Severalstudieshavepreviouslydocumentedan association betweenless-aggressiveend-of-life care and improved patient

quality of life near death.,,,- Ourstudyextends this re-

search by demonstrating similar associations with family

member–ratedquality for end-of-life care in a large cohort of

patientstreated in community settings in multiple regions of

theUnitedStates. Inaddition, weassessedpatients’preferences

for careandfound thatmost olderpatientswith fee-for-service

Medicarewanted palliativetreatmentsmore thanlife-extending

care,and preferredto dieat home,which issimilarto the find-

ingsfrom otherstudies.,Suchpreferences werestronglyas-

sociated with the aggressiveness of end-of-life care received.

Although existing quality measures characterize the re-

peateduse of emergency visitsnear deathas poorquality care,

nosed with lung and colorectal cancer in the US regions cov-

ered by the Surveillance, Epidemiology, and End Results

registries, we studied older patients with lung and colorec-

tal cancers insured by Medicare who died predominantly be-

fore .Wehad toexcludemany patients whowerenoten-

rolled in Medicare fee-for-service plans or whose family

members could notbe reached. Ourfindings may notbe gen-

eralizable to patients who are younger, commercially in-

sured, uninsured, treated in Veteran’s Administration hospi-

tals, have other diseases, or died more recently. The data are

older, particularly for patientswho died soon afterdiagnosis,

dueto thetime involved withlinkinginterviewandclaims data.

However,one of thestudy’sstrengthsis that itincludespa-

tientswho were longer-termcancersurvivorswho mayhave dif-ferent experiencesnear death.In addition, recent datademon-

strate that end-of-life care among Medicare beneficiaries

remains intensive (ICU admissions and late hospice enroll-

mentsare increasingovertime); thus patients’and familymem-

bers’experiencesare likelyto be similartoday. Our results may

be subject to unmeasured confounding related to unsuccess-

fullinkages to Medicare claims. However,our family member–

reported rates of “excellent”qualityfor end-of-lifecare andre-

ceipt of carethatwas notconsistent withpatients’ preferencesare similar to another population-based study of patients who

died from diverse diseases and were insured by Medicare fee-

for-service and health maintenance organization plans.

Family members reported on patients’ end-of-life expe-

riencesat variabletime points after death, andsome mayhave

forgotten details over time. Although a recent study demon-

strates moderate to high stability in bereaved family mem-

bers’ assessments of patients’ end-of-life experiences over

time, future research should examine family members’ re-

call over longer durations. We relied on family members’ re-

portsof patients’ preferredand actualplace of deathto deter-

mine goal attainment. Nevertheless, these questions were

separatedby atleast questionsto minimize consciouscom-

FamilyPerspectives on AggressiveCancer Care Near theEnd of Life Original Investigation Research


9/17

Health, Universityof Michigan,Ann Arbor

(Ayanian); Gerald R. Ford School of PublicPolicy,

Universityof Michigan, Ann Arbor (Ayanian);

Departmentof Epidemiology, Collegeof Public

Health, Universityof Iowa,Iowa City(Chrischilles);

RandCorporation,Santa Monica, California (Kahn);

Division of General Internal Medicine,Universityof

California, Los Angeles (Kahn); Divisionof

Geriatrics, Departmentof Medicine,Universityof

California, San Francisco(Ritchie); Institutefor

Clinical EvaluativeSciences, Toronto, Ontario,

Canada(Earle).

Author Contributions:DrsWright andKeatinghad

full accessto allof the datain the study and take

responsibility forthe integrityof thedataand the

accuracy of thedataanalysis.Drs Earle and

Landrum shared seniorauthorship. Study concept and design:Wright, Keating, Ayanian,

Weeks, Earle,Landrum.

Acquisition, analysis, or interpretation of data: All

authors.

Drafting of themanuscript:Wright, Keating, Weeks,

Landrum.

Critical revision of themanuscriptfor important

intellectual content:Wright, Keating,Ayanian,

Chrischilles,Kahn, Ritchie, Earle, Landrum.

Statistical analysis:Wright, Keating,Landrum.

Obtained funding:Ayanian,Kahn, Weeks, Earle.

Administrative, technical, or material support:Wright, Keating,Ayanian,Weeks,Earle, Landrum.

Study supervision:Weeks.

Conflict of InterestDisclosures:Theauthors have

completedand submittedtheICMJEForm for

Disclosure of PotentialConflicts of Interest.Dr

Richiereportedreceiving institutional supportfor

servingas thecurrentpresidentfor theAmerican

Academy of Hospice and PalliativeMedicine;

receiving grantfunding from theNational Institute

of Nursing Research,National Instituteon Aging,

RetirementResearchRetirement Foundation,

California HealthcareFoundation, CommonwealthFund, and the West Health Institute;receiving

payment forlecturesfrom theMedicalGrand

Roundsat theUniversityof Wisconsin; and

receivingroyalties forserving as an editorfor

Roleof theFunder/Sponsor:TheNational Cancer

Institutehad a rolein thedesign andconduct of the

study; collection,management, analysis, and

interpretation of thedata; butnot inthe

preparation,review, or approval of the manuscript;

anddecisionto submitthe manuscriptfor

publication.

AdditionalContributions:Wethank LinDing,PhD

(Harvard Medical School), for conductingstatistical

analyses; herworkon this projectwas supported by

researchgrants to HarvardMedical Schoolfrom the

National Cancer Institute.

REFERENCES

1. TenoJM, GozaloPL, Bynum JP, etal. Changein

end-of-life care for Medicare beneficiaries. JAMA.

2013;309(5):470-477.

2. WrightAA, ZhangB, RayA, etal. Associations

between end-of-life discussions,patient mental

health, medical care neardeath, and caregiver

bereavement adjustment. JAMA. 2008;300(14):

1665-1673.

3. Earle CC,ParkER, LaiB, etal. Identifying

potential indicators of thequality of end-of-life

cancer carefrom administrative data. J Clin Oncol .

2003;21(6):1133-1138.

4. Earle CC,LandrumMB, Souza JM,et al.

Aggressiveness of cancercarenear theendof life. J Clin Oncol . 2008;26(23):3860-3866.

5. Jacobson JO, Neuss MN,Hauser R. Measuring

andimproving valueof care inoncology practices.

Am Soc Clin Oncol Educ Book . 2012;e70-e76.

6. Patient Protectionand Affordable Care Act,

§§119, 3001,3002, 3007, 3021, 3025, 10327

(2010).

7. Obermeyer Z,MakarM, Abujaber S,et al.

Associationbetween theMedicare hospice benefit

andhealth care utilizationandcostsfor patients

withpoor-prognosis cancer. JAMA. 2014;312(18):1888-1896.

8. TemelJS, GreerJA, Muzikansky A, etal. Early

palliative care for patients withmetastatic

non-small-cell lung cancer.N EnglJ Med. 2010;363

of end-of-life cancer care. IntJ Qual HealthCare .

2005;17(6):505-509.

16. WrightAA, KeatingNL, Balboni TA, etal. Place

of death. J Clin Oncol . 2010;28(29):4457-4464.

17. SmithAK, Earle CC,McCarthyEP.Racial andethnic differencesin end-of-life carein

fee-for-service Medicare beneficiarieswith

advanced cancer. J Am Geriatr Soc . 2009;57(1):153-

158.

18. Spiegelman D,Hertzmark E. Easy SAS

calculations forrisk or prevalence ratiosand

differences. Am J Epidemiol . 2005;162(3):199-200.

19. HeY,ZaslavskyAM, LandrumMB, etal.

Multiple imputationin a large-scalecomplex survey.

Stat Methods Med Res. 2010;19(6):653-670.

20. Obermeyer Z,Powers BW, Makar M,KeatingNL, Cutler DM.Physiciancharacteristics strongly

predictpatientenrollmentin hospice.HealthAff

(Millwood). 2015;34(6):993-1000.

21. CutlerD, Skinner J,SternAD,Wennberg D.

Physician beliefs and patient preferences: a new

lookat regional variationin healthcarespending.

http://www.nber.org/papers/w19320. Accessed

December 22, 2015.

22. GreerDS, MorV,MorrisJN, etal. Analternative

in terminal care. J Chronic Dis. 1986;39(1):9-26.

23. WallstonKA, BurgerC, SmithRA, BaugherRJ.Comparingthe quality of deathfor hospice and

non-hospicecancer patients.MedCare . 1988;26

(2):177-182.

24. Teno JM,ClarridgeBR, CaseyV, etal. Family

perspectiveson end-of-life care at thelastplaceof

care. JAMA. 2004;291(1):88-93.

25. ZhangB, WrightAA, Huskamp HA,et al.Health

carecosts in thelastweekof life. Arch InternMed .

2009;169(5):480-488.

26. Barnato AE,AnthonyDL, Skinner J,et al.Racial

and ethnic differencesin preferences for end-of-lifetreatment. J Gen InternMed . 2009;24(6):695-701.

27. HigginsonIJ,Sen-Gupta GJ. Placeof carein

advanced cancer. J PalliatMed . 2000;3(3):287-300.

28 SmithAK SchonbergMA FisherJ etal

Research Original Investigation FamilyPerspectives on AggressiveCancer Care Near theEnd of Life

http://www.ncbi.nlm.nih.gov/pubmed/23385273http://www.ncbi.nlm.nih.gov/pubmed/23385273http://www.ncbi.nlm.nih.gov/pubmed/23385273http://www.ncbi.nlm.nih.gov/pubmed/23385273http://www.ncbi.nlm.nih.gov/pubmed/18840840http://www.ncbi.nlm.nih.gov/pubmed/18840840http://www.ncbi.nlm.nih.gov/pubmed/18840840http://www.ncbi.nlm.nih.gov/pubmed/18840840http://www.ncbi.nlm.nih.gov/pubmed/12637481http://www.ncbi.nlm.nih.gov/pubmed/12637481http://www.ncbi.nlm.nih.gov/pubmed/12637481http://www.ncbi.nlm.nih.gov/pubmed/12637481http://www.ncbi.nlm.nih.gov/pubmed/18688053http://www.ncbi.nlm.nih.gov/pubmed/18688053http://www.ncbi.nlm.nih.gov/pubmed/18688053http://www.ncbi.nlm.nih.gov/pubmed/24451835http://www.ncbi.nlm.nih.gov/pubmed/24451835http://www.ncbi.nlm.nih.gov/pubmed/24451835http://www.ncbi.nlm.nih.gov/pubmed/25387186http://www.ncbi.nlm.nih.gov/pubmed/25387186http://www.ncbi.nlm.nih.gov/pubmed/25387186http://www.ncbi.nlm.nih.gov/pubmed/25387186http://www.ncbi.nlm.nih.gov/pubmed/20818875http://www.ncbi.nlm.nih.gov/pubmed/20818875http://www.ncbi.nlm.nih.gov/pubmed/15985505http://www.ncbi.nlm.nih.gov/pubmed/15985505http://www.ncbi.nlm.nih.gov/pubmed/15985505http://www.ncbi.nlm.nih.gov/pubmed/15985505http://www.ncbi.nlm.nih.gov/pubmed/20837950http://www.ncbi.nlm.nih.gov/pubmed/20837950http://www.ncbi.nlm.nih.gov/pubmed/20837950http://www.ncbi.nlm.nih.gov/pubmed/19054185http://www.ncbi.nlm.nih.gov/pubmed/19054185http://www.ncbi.nlm.nih.gov/pubmed/19054185http://www.ncbi.nlm.nih.gov/pubmed/19054185http://www.ncbi.nlm.nih.gov/pubmed/15987728http://www.ncbi.nlm.nih.gov/pubmed/15987728http://www.ncbi.nlm.nih.gov/pubmed/15987728http://www.ncbi.nlm.nih.gov/pubmed/19654173http://www.ncbi.nlm.nih.gov/pubmed/19654173http://www.ncbi.nlm.nih.gov/pubmed/19654173http://www.ncbi.nlm.nih.gov/pubmed/26056205http://www.ncbi.nlm.nih.gov/pubmed/26056205http://www.ncbi.nlm.nih.gov/pubmed/26056205http://www.ncbi.nlm.nih.gov/pubmed/26056205http://www.nber.org/papers/w19320http://www.ncbi.nlm.nih.gov/pubmed/3511081http://www.ncbi.nlm.nih.gov/pubmed/3511081http://www.ncbi.nlm.nih.gov/pubmed/3511081http://www.ncbi.nlm.nih.gov/pubmed/3339915http://www.ncbi.nlm.nih.gov/pubmed/3339915http://www.ncbi.nlm.nih.gov/pubmed/3339915http://www.ncbi.nlm.nih.gov/pubmed/3339915http://www.ncbi.nlm.nih.gov/pubmed/14709580http://www.ncbi.nlm.nih.gov/pubmed/14709580http://www.ncbi.nlm.nih.gov/pubmed/14709580http://www.ncbi.nlm.nih.gov/pubmed/19273778http://www.ncbi.nlm.nih.gov/pubmed/19273778http://www.ncbi.nlm.nih.gov/pubmed/19273778http://www.ncbi.nlm.nih.gov/pubmed/19273778http://www.ncbi.nlm.nih.gov/pubmed/19387750http://www.ncbi.nlm.nih.gov/pubmed/19387750http://www.ncbi.nlm.nih.gov/pubmed/19387750http://www.ncbi.nlm.nih.gov/pubmed/15859670http://www.ncbi.nlm.nih.gov/pubmed/15859670http://www.ncbi.nlm.nih.gov/pubmed/15859670http://www.ncbi.nlm.nih.gov/pubmed/15859670http://www.ncbi.nlm.nih.gov/pubmed/19387750http://www.ncbi.nlm.nih.gov/pubmed/19273778http://www.ncbi.nlm.nih.gov/pubmed/19273778http://www.ncbi.nlm.nih.gov/pubmed/14709580http://www.ncbi.nlm.nih.gov/pubmed/3339915http://www.ncbi.nlm.nih.gov/pubmed/3339915http://www.ncbi.nlm.nih.gov/pubmed/3511081http://www.nber.org/papers/w19320http://www.ncbi.nlm.nih.gov/pubmed/26056205http://www.ncbi.nlm.nih.gov/pubmed/26056205http://www.ncbi.nlm.nih.gov/pubmed/19654173http://www.ncbi.nlm.nih.gov/pubmed/15987728http://www.ncbi.nlm.nih.gov/pubmed/19054185http://www.ncbi.nlm.nih.gov/pubmed/19054185http://www.ncbi.nlm.nih.gov/pubmed/20837950http://www.ncbi.nlm.nih.gov/pubmed/15985505http://www.ncbi.nlm.nih.gov/pubmed/15985505http://www.ncbi.nlm.nih.gov/pubmed/20818875http://www.ncbi.nlm.nih.gov/pubmed/25387186http://www.ncbi.nlm.nih.gov/pubmed/25387186http://www.ncbi.nlm.nih.gov/pubmed/24451835http://www.ncbi.nlm.nih.gov/pubmed/18688053http://www.ncbi.nlm.nih.gov/pubmed/12637481http://www.ncbi.nlm.nih.gov/pubmed/12637481http://www.ncbi.nlm.nih.gov/pubmed/18840840http://www.ncbi.nlm.nih.gov/pubmed/18840840http://www.ncbi.nlm.nih.gov/pubmed/23385273http://www.ncbi.nlm.nih.gov/pubmed/23385273


10/17

American Medical Association All rights reserved

Supplementary Online Content

Wright AA, Keating NL, Ayanian JZ, et al. Family perspectives on aggressive cancer care near the end of life. JAMA.doi:10.1001/jama.2015.18604

eTable 1. Patient Deaths and Family Interviews by Year eTable 2. Comparison of Patients Included in the Cohort and Those Who Were Excluded Because Family After DeathInterviews Were Not Available or Family Members Did Not Rate the Quality of Patients’ End-of-Life (EOL) Care

eTable 3. Relationship between Family Member–Reported Patient Treatment Preferences and Receipt of AggressiveEnd-of-life (EOL) Care Among Those who Died within 90 days of DiagnosiseTable 4. Relationship between Family Member–Reported Patient Treatment Preferences and Receipt of AggressiveEnd-of-life (EOL) Care Among Those who Died within 180 days of DiagnosiseTable 5. Associations Between Aggressive End-of-Life (EOL) Medical Care Measures and Family Members’ OverallRating of the Quality of Patients’ Care Near Death Adjusted for Patients’ Symptoms Near DeatheTable 6. Associations Between Aggressive End-of-Life (EOL) Medical Care and Receipt of Goal-Concordant EOL Care Adjusted for Patients’ Symptoms Near Death

This supplementary material has been provided by the authors to give readers additional information about their work.


11/17


eTable 1. Patient Deaths and Family Interviews by Year

Year

Patient Deaths, No. (%) Family Interviews, No. (%)

Total(n=1146) Lung(n=886) Colorectal(n=260) Total(n=1146) Lung(n=886) Colorectal(n=260)2003 124 (10.8) 96 (10.8) 28 (10.8) 26 (2.3) 18 (2.0) 8 (3.1)

2004 502 (43.8) 439 (49.6) 63 (24.2) 396 (34.6) 337 (38.0) 59 (22.7)

2005 257 (22.4) 218 (24.6) 39 (15.0) 390 (34.0) 343 (38.7) 47 (18.1)

2006 81 (7.1) 48 (5.4) 33 (12.7) 33 (2.9) 27 (3.1) 6 (2.3)

2007 45 (3.9) 25 (2.8) 20 (7.7) 1 (0.1) 1 (0.1) 0 (0.0)

2008 56 (4.9) 28 (3.2) 28 (10.8) ** ** **

2009 34 (3.0) 12 (1.4) 22 (8.5) ** ** **

2010 34 (3.0) 14 (1.6) 20 (7.7) ** ** **

2011 13 (1.1) 6 (0.7) 7 (2.7) ** ** **

2012 ** ** ** 300 (26.2) 160 (18.1) 140 (53.9)** There were no patient deaths in 2012 and family interviews were not performed in 2008-2011.


12/17


eTable 2. Comparison of Patients Included in the Cohort and Those Who Were Excluded Because Family After Death InterviewsWere Not Available or Family Members Did Not Rate the Quality of Patients’ End-of-Life (EOL) Care

Characteristic, No. (%)

Total patients withMedicare claimsavailable (n=1621)

Patients with linkedfamily decedentinterviews (n=1146)

Patients withoutlinked family decedentinterviews or missing

information aboutEOL care quality

(n=475) P-value1 Lung cancer 1238 (76.4) 886 (77.3) 352 (74.1) 0.16

Male 895 (55.2) 639 (55.8) 256 (53.9) 0.49

Age at death, years


13/17


eTable 3. Relationship between Family Member–Reported Patient Treatment Preferences and Receipt of Aggressive End-of-life(EOL) Care Among Those who Died within 90 days of Diagnosis

Aggressive EOL Care Measure,No. (%)

Total Cohort(n=318)

Among patientsreporting preference toextend life vs. relievesymptoms (n=96)

Among patientsreporting preference torelieve symptoms vs.extend life(n=222) P-value1

Intensive care unit admission 31 (9.7) 14 (14.6) 17 (7.7) 0.06

≥ 2 Hospitalizations 17 (5.3) 11 (11.5) 6 (2.7) 0.001

≥ 2 Emergency department visits 23 (7.2) 10 (10.4) 13 (5.9) 0.15

Chemotherapy ≤2 weeks beforedeath

18 (5.7) 12 (12.5) 6 (2.7)


14/17


eTable 4. Relationship between Family Member–Reported Patient Treatment Preferences and Receipt of Aggressive End-of-life(EOL) Care Among Those who Died within 180 days of Diagnosis

Aggressive EOL Care Measure,No. (%)

Total Cohort(n=668)

Among patientsreporting preference toextend life vs. relievesymptoms (n=187)

Among patientsreporting preference torelieve symptoms vs.extend life(n=481) P-value1

Intensive care unit admission 86 (12.9) 31 (16.6) 55 (11.4) 0.07

≥ 2 Hospitalizations 42 (6.3) 18 (9.6) 24 (5.0) 0.03

≥ 2 Emergency department visits 56 (8.4) 18 (9.6) 38 (7.9) 0.47

Chemotherapy ≤2 weeks beforedeath

35 (5.2) 22 (11.8) 13 (2.7)


15/17


eTable 5. Associations Between Aggressive End-of-Life (EOL) Medical Care Measures and Family Members’ Overall Rating of theQuality of Patients’ Care Near Death Adjusted for Patients’ Symptoms Near Death

Aggressive EOL CareMeasure

Incidence inFull Cohort

Family Reports of Excellent Quality EOL CareUnadjusted Adjusted1

n % %RiskDifference1 95%CI P value1 (n=1146)

ICU admission in last month of life

No 995 (86.8) 520 52.3 52.1-7.1 -15.1 to 1.0 0.09

Yes 151 (13.1) 68 45 45

≥2 Hospitalizations in last month of life

No 1076 (93.9) 556 51.7 51.5-5.9 -17.5 to 5.6 0.31

Yes 70 (6.1) 32 45.7 45.6

≥2 Emergency department visits in last month of life

No 1056 (92.1) 542 51.3 51.2 -1.1 -11.4 to 9.2 0.84Yes 90 (7.9) 46 51.1 50.1

Chemotherapy ≤2 weeks before death

No 1071 (93.5) 551 51.5 51.2-1.0 -12.1 to 10.1 0.85

Yes 75 (6.5) 37 49.3 50.2

No hospice or ≤3 days before death

No 599 (52.3) 352 58.8 57.8-13.7 -19.7 to -7.6


16/17


eTable 6. Associations Between Aggressive End-of-Life (EOL) Medical Care and Receipt of Goal-Concordant EOL Care Adjusted forPatients’ Symptoms Near Death

Aggressive EOLCare Measure

Family Reports EOL Wishes Followed Family Reports Patient Died in Preferred PlaceUnadjusted Adjusted1 Unadjusted Adjusted1 Total2

(n=930) % %Risk

Difference1

95% CI

Pvalue

Total2 (n=774) % %

RiskDifference1

95%CI

P value

ICU admission in last month of life

No 662/814 81.3 78.7

-3.8-13.1 to5.5

0.42

404/682 59.2 58.2

-18.6-28.6to -8.6


17/17


1 Adjusted for gender, age, race/ethnicity, education, income, marital status, region, integrated health system, cancer stage, comorbid conditions,depression, time from diagnosis to death, respondent, treatment preferences, cancer type, time between death and after death interview, andfamily-reported measures of patients’ symptoms (pain and trouble breathing), and whether they were adequately controlled near death usinglinear-binomial models with an identity link. The model examining associations between preferred place of death and death in hospital did not

converge; we used a linear probability model in this case. We report 2-sided p values assessing differences for individuals who experienced themeasure of aggressive care vs. those who did not.2 Analyses examining whether patients’ EOL wishes were “followed a great deal” included 930 interviews from bereaved family members; 216family members who responded “unknown” were excluded. Analyses examining whether patients died in their preferred place of death included774 interviews from bereaved family members; 372 family members who responded “unknown” to a question about where patients wanted to diewere excluded.EOL=End of life; ICU=intensive care unit; CI=confidence interval

family perspectives on aggressive cancer care near the end of life

Documents