400 Journal of Pain and Symptom Management Vol. 30 No. 5 November 2005

Original Article

Late Referral to Hospice and BereavedFamily Member Perception of Qualityof End-of-Life CareErica R. Schockett, MD, Joan M. Teno, MD, MS, Susan C. Miller, PhD,and Brad Stuart, MDCenter for Gerontology and Health Care Research, and Department of Community Health,

Brown Medical School (E.R.S., J.M.T., S.C.M.), Providence, Rhode Island;

and Sutter VNA & Hospice (B.S.), Emeryville, California, USA

AbstractThe Family Evaluation of Hospice Services was used to document bereaved family members’perceptions of whether their loved ones were referred too late to hospice and to examine theassociation of that perception and quality of end-of-life care. A mortality follow-back survey ofbereaved family members from two not-for-profit hospices found that 13.7% of decedents werereferred at a time too late for hospice services. Family members of persons referred too late re-ported lower satisfaction with hospice services, a higher rate of unmet needs for informationabout what to expect at time of death, lower confidence in participating in patient care athome, more concerns with coordination of care, and lower overall satisfaction. Familiesreported physicians as an important barrier to earlier hospice referral in nearly one-half of cases.These results indicate a need for improved services for shorter-stay hospice patients/familiesand for physicians to help facilitate earlier hospice admission. J Pain Symptom Manage2005;30:400--407. � 2005 U.S. Cancer Pain Relief Committee. Published by Elsevier Inc.All rights reserved.

Key WordsHospice access, quality of end-of-life care

IntroductionHospice focuses on providing comprehen-

sive physical, emotional, and spiritual care topatients and their loved ones at a crucial pointin the life process.1 While experts recommenda 3 month hospice stay to allow time to providepatients and their families with adequate

Address reprint requests to: Erica R. Schockett, MD,Center for Gerontology and Health Care Research,1500 Locust St., Apt. 3319, Philadelphia, PA 19102,USA.

Accepted for publication: April 28, 2005.

� 2005 U.S. Cancer Pain Relief CommitteePublished by Elsevier Inc. All rights reserved.

physical and psychological support,2 themedian length of hospice service in Americadeclined from 29 days in 1995 to 22 days in2003,3 and, in 2003, 36.9% of enrollees diedwithin 1 week of hospice care initiation.3

There are a limited number of studies inves-tigating the underutilization of hospice care.In-depth interviews and surveys of health careprofessionals, as well as chart reviews and pa-tient focus groups, have illuminated potentialreasons for nonreferral and later referral tohospice. Two of the most common barriers totimely referral reported by health professio-nals are patient and family readiness and

0885-3924/05/$--see front matterdoi:10.1016/j.jpainsymman.2005.04.013

Vol. 30 No. 5 November 2005 401Late Referral to Hospice and Quality of Care

acceptance of end-of-life.4--7 Additional bar-riers to referral include physician, patient,and/or family lack of knowledge, comfort,and less positive attitudes about hospice anddeath.7--10 Physician inability to identifyhospice-appropriate diagnoses and prognoses,11--13 and concern that patients would perceivereferral as a cost-saving decision,13 are alsoreported barriers. Hospice underutilization al-so may stem from physicians’ difficulty in ac-cepting patient mortality and focus on cure,due to either training or financial incentive.Additionally, patients and families may resisttransitioning care from their health care pro-viders.7 Underutilization of hospice amonglow-income urban Hispanic/Latino and Afri-can American families, in particular, may resultfrom general concern of the costliness ofhealth care, as well as cultural beliefs of familyprovision of care to the dying and mistrust ofthe medical system.10 The American Collegeof Physicians-American Society of InternalMedicine End-of-Life Consensus Panel con-cluded that some minority groups may notfavor the current emphasis on patient autono-my and informed consent involved in end-of-life decision making.14

Research has found that, over time, hospicestays have become shorter in general.15 In1995--1996, hospice referrals and length of staywere further diminished secondary to aggres-sive federal investigation of alleged fraud andabuse in hospice.16 Additional factors associat-ed with shorter stays include diagnoses such asrenal disease and certain cancers,17,18 malegender,19--21 white race,21,22 private insurancestatus,20 enrollment in Medicare’s fee-for-service program (vs. a Medicare HMO orhealth maintenance organization),23 livingwith a caregiver,20 and having a hospitalreferral source.20 One study, though, foundno difference in patient length of hospice staybetween patients referred from academic andnonacademic medical centers.24 Hospicelengths of stay have also been found to varyby the site of hospice care provision (nursinghome vs. nonnursing home setting),25 acrossstates and health care markets,21,26 and byhospice provider characteristics.22

No study to date has asked bereaved familymembers their perceptions of referral timingand the reasons for late referral, and therehas been very limited research on the impact

of late hospice referral on satisfaction withend-of-life care. Only one study has investigat-ed family members’ perceptions of referraltiming and hospice care. Based on responsesto the Hospice Care Family Survey, the studyfound lower overall ratings of hospice servicesfrom families who reported referral as too latecompared to those who reported referral as ap-propriately timed.27,28 The key limitation ofthis study is its reliance on rating questionsonly to measure satisfaction, as these are oftenskewed and difficult to interpret.

There are newer tools available that usedifferent response tasks to provide informa-tion on unmet needs and concerns that fami-lies report about the quality of end-of-lifecare. One such tool is the Family Evaluationof Hospice Care, which has been validatedand used in a national study of dying in theUnited States,29,30 and is used by the NationalHospice and Palliative Care Organization tomeasure quality of hospice care. This tool re-ports bereaved family members’ perceptionsof patient and family unmet needs, concernsabout quality of care, and quality ratings ofmultiple care components.

The research effort presented here had twomain goals. First, the Family Evaluation of Hos-pice Care survey was used to examine associa-tions of bereaved family members’ report ofwhether or not referral was made ‘‘too late’’,and their satisfaction with home hospice care.Second, family members’ responses to anopen-ended question regarding reason for latereferral were examined to identify importanttrends in hospice referral.

MethodsThe study sample was drawn from a random

sample of deaths for home teams in two hos-pice programs, one in Rhode Island and onein California, which provided contact informa-tion for a total of 416 family members of homehospice patients who had died in the prior 3--6months. A total of 237 surveys were completedby telephone interview or mail; the responserates for those who completed the survey bytelephone interview or self-administrationwere 63.7% or 50.2%, respectively. The origi-nal goal of this study was to test the mode ofadministration on survey response patterns.Because the mode of administration was found

402 Vol. 30 No. 5 November 2005Schockett et al.

to yield no significant differences for this sur-vey tool,29 we report here the survey resultsin aggregate.

The dependent variable of perceived hos-pice referral timing was elicited with the fol-lowing question: ‘‘In your opinion, was thepatient referred to hospice too early, at theright time, or too late?’’ All those reporting‘‘too late’’ referral were asked to respond tothe following open-ended item: ‘‘In your ownwords, please write the reason why the patientwas referred too late to hospice.’’ Verbatim orwritten responses were entered into a computerto conduct content analysis and identify themajor barriers to earlier referral for end-of-lifecare. This analysis was completed by two inves-tigators (E.S. and J.M.T.), with disagreementresolved by discussion.

The Family Evaluation of Hospice Care wasused to evaluate the association between hos-pice referrals that were perceived as being‘‘too late’’ and the perceptions of quality ofend-of life care. This tool elicits family memberperceptions of hospice care across five do-mains: patient physical comfort and emotionalsupport, respectful treatment of the patient,attendance to family needs, coordination ofcare, and overall quality of hospice care. It usesfamily-centered reports, unmet needs, self-efficacy, and rating questions to examine thequality of end-of-life care. Family-centered re-ports were summed to report the number ofconcerns within a domain. For ease of

presentation, results are presented as one ormore concerns in that particular domain.Responses from the six study participantswho reported ‘‘too early’’ referral to hospicecare were excluded from this analysis due tothe small sample size. Self-efficacy was the sum-mation of the responses of ‘‘very confident’’(scored a 3) to ‘‘not at all confident’’ (scoreda 1). Rating questions included five items thatasked the respondent to rate whether care was‘‘best possible’’ (scored a 10) to ‘‘worst’’ (scoreda 0). These five questions were summarizedfrom 0 to 50.

All analyses of sociodemographics, hospicecare outcome measures, and perception ofreferral timing were conducted with the SAS(Statistical Application System) version 8.2program. For nominal variables, the chi-squared test was used. The Mantzel Hanzelchi-squared test or nonparametric one-wayanalysis of variables was used for ordinal varia-bles. Statistical significance was defined at the0.05 level.

ResultsSample Characteristics

Table 1 summarizes characteristics of re-spondents and decedents stratified by whethera bereaved family member reported that his orher loved one was referred ‘‘too late’’ or ‘‘at theright time.’’ Only 2.5% (n 5 6) reported thatthe patient was referred too early, while

Table 1Sample Description (n[ 237)

Characteristics ‘‘Too Early’’ (n 5 6, 2.5%)‘‘At the Right

Time’’ (n 5 196, 82.7%) ‘‘Too Late’’ (n 5 32, 13.7%)P

Value

Respondent relation to decedent (%) 0.02Spouse/partner 16.7 36.2 31.4Child 50.0 44.4 43.8

Patient characteristicsAge (mean years) 81.7 78.6 73.8 0.11Male (%) 0.0 42.9 59.4 0.02White (%) 100.0 97.5 93.8 0.89

Diagnosis (%) 0.63Cancer 33.3 52.6 59.4Heart disease 33.3 9.2 15.6Lung disease 0.0 8.2 6.2Dementia 0.0 7.6 0.0

Education (% lessthan high school)

33.2 19.9 25.0 0.56

Length of hospice stay[mean (median) days]

117.8 (139.5) 97.6 (31.0) 46.9 (21.0) 0.05

Vol. 30 No. 5 November 2005 403Late Referral to Hospice and Quality of Care

13.7% (n5 32) reported that the referral tohospice was made too late. There were no sta-tistically significant differences in mean ageand in the proportions of patients by race, di-agnoses, or education levels within the threereferral categories (P$ 0.11). However, com-pared to the proportion of males in the non-late referral categories, there was a significantlygreater proportion of male patients in thereferred too late category, 59.4% vs. 42.9%,respectively (P! 0.02). Family members’ per-ception of referral timing also was related tohospice length of stay. Persons referred ‘‘toolate’’ had a significantly shorter (P! 0.05) me-dian length of stay of 21.0 days compared to31.0 days in those cases where the family mem-ber believed that hospice referral was made atthe right time.

Hospice Care OutcomesTable 2 demonstrates that respondents’ per-

ception of late hospice referral was associatedwith statistically significant differences in un-met needs, number of concerns about thequality of hospice care, and ratings of the qual-ity of end-of-life care provided. Overall satisfac-tion with care was significantly lower amongthose who perceived hospice referral as ‘‘toolate’’ than among those who perceived hospicereferral as ‘‘at the right time’’ (56.3% reportedexcellent overall quality of end-of-life carecompared to 75.0%, P ! 0.001). Similarly,there was a significant (P ! 0.001) 5.9-pointdifference in family members’ mean rating offive aspects of quality end-of-life care, each as-sessed from 0 (worst) to 10 (best care). Whilethere were no statistically significant differen-ces in unmet needs for dyspnea and personalADL (activities of daily living) care, familymembers reported a higher rate of unmetneeds regarding emotional support for pa-tients referred too late, 20% vs. 9.5%(P ! 0.03). The only category in which thosereferred ‘‘too late’’ reported less dissatisfactionwas for help with pain management, 3.8% vs.5.9% among those referred ‘‘at the right time’’(P ! 0.05). Still, as outlined in Table 2, familymembers of those referred too late vs. those re-ferred at the right time reported higher ratesof unmet needs for information they receivedregarding the patient’s condition and what toexpect at time of death (P ! 0.001). Thistrend of unmet needs among those who

reported late referral is also evident from theirreports of higher rates of concerns about thehospice team’s coordination of care(P ! 0.001) and a lower overall degree of con-fidence in caring for the patient (P 5 0.02)than those who reported referral of their lovedones at the right time.

Reasons for ‘‘Too Late’’ ReferralFamily members who reported ‘‘too late’’ re-

ferral were asked, ‘‘In your own words, please

Table 2Outcomes by Perception of Referral Timinga

Outcome

‘‘At theRightTime’’(%)

‘‘TooLate’’(%) P Value

Provided desired physical comfort and emotional supportto patient

Patient received less help than was wanted withPain 5.9 3.8 0.05Dyspnea 4.0 5.9 0.61Personal care 24.3 33.3 0.09Emotional support 9.5 20.0 0.03

Treated patient with respectNot always treating

patient with respect2.1 6.3 0.15

Attended to needs of the familyAt least one or more

concern(s) aboutemotional or spiritualsupport to family

19.9 31.6 0.06

At least one or moreconcern(s) aboutinformation to familyregarding patient’scondition and death

29.6 68.8 !0.001

At least one or moreconcern(s) aboutinformation to familyregarding patient’ssymptoms

8.7 12.5 0.56

Self-efficacy (mean[standarddeviation])b

10.2 [2.0] 9.2 [2.0] 0.02

Coordinated careAt least one or more

concern(s) aboutcoordination of care

21.4 46.9 !0.001

Overall assessment of quality of careOverall rating of

hospice team care(mean [standarddeviation])c

46.6 [5.3] 40.7 [11.0] !0.001

Overall satisfactionwith hospice care(% excellent)

75.0 56.3 !0.001

aResults are presented only for the 228 respondents who reportedthe timing of hospice referral as ‘‘at the right time’’ or ‘‘too late.’’bResults are based on 201 available observations, due to missingvalues or deaths not at home.cResults are based on 214 observations, due to missing values.

404 Vol. 30 No. 5 November 2005Schockett et al.

write the reason why the patient was referredtoo late to hospice.’’ All 32 respondents whoreported late referral provided an explanationin their own words. Of the 32 family members,46.9% (n5 15) reported a physician-centeredbarrier to earlier referral, and 15.6% (n5 5)of all late referral respondents noted physi-cian’s accuracy of prognosis as the reason forlate referral.

Three late referral respondents indicatedphysician’s accuracy of diagnosis as a barrierto earlier referral to hospice. The daughter ofan 83-year-old woman with cancer commented

‘‘[The] physician did not diagnose mymom properly until it was too late forher to survive.’’

The brother of a 42-year-old woman with can-cer noted

‘‘[the doctor was] treating an illness hedidn’t even know he was treating. Aftershe died he said he didn’t even knowwhat kind of cancer it was.’’

Three other family members reported physi-cian disagreement with hospice philosophyas the reason for late referral. A close friendof a 95-year-old male with heart diseasedescribed

‘‘His physician did not believe in or un-derstand hospice. It had to be pushedby me to get it.’’

Three other respondents described theirphysicians’ desires to pursue further treatmentand/or testing before referral. The wife ofa 52-year-old man with cancer complained

‘‘Because he had an optimistic oncologistand we were getting ready for clinical tri-als.we could have used it 1 month earlier.Our GP was encouraging hospice beforeour oncologist.’’

The other main category of barriers to earlierreferral, reported by 28.1% (n 5 9) of re-spondents, related to patient and familyknowledge and beliefs. Patient decline of anearlier hospice referral accounted for 12.5%(n 5 4) of all reasons, second only to physi-cian accuracy of diagnosis and/or prognosisas a single reason for later referral. One ex-planation provided by the wife of a 75-year-old man with cancer was

‘‘It was his own choice; he wasn’t ready.I would have taken the referral earlierbut he wasn’t ready.’’

The most commonly reported family-relatedreason for later referral, reported by 9.4%(n 5 3) of the 32 respondents, was the family’slack of awareness of hospice or its eligibilityrequirements. The daughter of a 90-year-oldwoman simply stated

‘‘I didn’t ask earlierdI didn’t know.’’

DiscussionToo many dying persons are referred to hos-

pice services in the last days of life despiteproven benefits of hospice in end-of-life care.17

Our results found that one in seven bereavedfamily members stated that his or her lovedone was referred too late to hospice. While itis reassuring that the majority of respondentsbelieved hospice referral was at the right time,there is still an important opportunity toimprove, in that one in seven persons was re-ferred too late and his or her family reporteda higher rate of concerns, more unmet needs,and greater dissatisfaction with the quality ofend-of-life care than bereaved family memberswho state that hospice referral was at the righttime.

About half of families who reported ‘‘toolate’’ referral cited physician barriers. Severalstudies suggest that physicians are overly opti-mistic in their prognostic estimates for patientswith advanced disease, particularly terminalcancer.31 A recent meta-analysis showed thatsurvival of late-stage cancer patients was 30%less than that predicted by their physicians,32

and a prospective cohort study including bothcancer and nononcology patients showed thatphysicians overestimated survival in 63% ofcases by a factor of more than five, and errorsoccurred despite medical specialty and patientdiagnostic groups.33 Moreover, physicians re-port that even when cancer patients specifical-ly request prognostic estimates, they wouldwithhold their opinion or provide a willfully in-accurate figure in almost two-thirds of cases.34

Terminally ill cancer patients who have inap-propriately optimistic beliefs about their sur-vival time often choose futile aggressivetreatment rather than palliative measures.35

Vol. 30 No. 5 November 2005 405Late Referral to Hospice and Quality of Care

Physicians, therefore, can act as a barrier toearlier hospice referrals and patient avoidanceof futile testing and treatment, as reported bysome families in this study.

Those who deemed referral as too late re-ported higher rates of unmet needs for coordi-nation or care and self-efficacy, and loweroverall ratings of and satisfaction with hospiceservices. These results highlight both the needto assess the services provided to shorter-stayhospice patients and the necessity of educa-tional and other efforts to change physicianhospice referral behavior.

Important limitations of this study should benoted in interpreting these findings. The sur-vey results rely on respondents’ perceptionsof both patients’ and their own quality ofend-of-life care. Respondents may have inaccu-rately reported on patients’ unmet needs forphysical comfort and emotional support. Arecent review of studies on patient vs. proxyreport on end-of-life care found better agree-ment regarding observable symptoms andquality of services than subjective aspects ofthe patient experience.36 However, it wouldnot be expected that this discrepancy woulddiffer among the groups compared in thisstudy. Also, since this study’s sample was drawnfrom only two hospices, the results may not begeneralizable to other hospice programs. Addi-tionally, because the hospices did not releasesubject identifiers without their permission,we are unable to characterize the nonresponsebias. Finally, there was a small number of re-spondents who stated that their loved oneswere referred ‘‘too late.’’ Because of this smallsample size, the power of study to find statisti-cally significant differences is limited. Furtherresearch is needed to examine whether ourfindings hold with a larger, more generalizablesample.

Only one previous study examined bereavedfamily members’ satisfaction with hospice carestratified by perceived timing of referral. Inthat study, only 8% of respondents believedthat their loved ones were referred too lateto hospice.27 Our work expands upon thatstudy’s findings by investigating the associationof perception of too late referral and a numberof outcome measures based on the FamilyPerceptions of Hospice Care survey, developedby Brown University and currently used bythe National Hospice and Palliative Care

Organization. Our results show that percep-tion of late referral is associated with lower sat-isfaction with hospice care overall, as well aswith coordination of care, family self-efficacyor confidence in providing care for the dyingperson at home, emotional support, and infor-mation regarding patient condition. Elicitingfamily members’ perspective has allowed usto examine their unique viewpoint on barriersto timely hospice referral. In contrast to earlierstudies of health professionals, which have sug-gested patient and family readiness among thecommonest barriers,4--7 the respondents herenoted physician-centered reasons for later re-ferral more frequently than patient- or family-related barriers.

Our results suggest several interventionsthat might be effective in helping to overcomethese factors. First, both research and educa-tion are needed to help physicians to formu-late more accurate prognoses. Prognosticmodels have had persistent shortcomings in in-dividual cases,37 although recent attempts tointegrate variables such as performance statusand clinical signs through regression analysisinto an index of likely survival time have pre-dicted a 3 week survival with sensitivity andspecificity, each of approximately 85%.38 An-other way of arriving at more accurate progno-ses may be to consult independent physicians,who may not be as closely connected emotion-ally to the patient, for independent survivalestimates.39

More widespread awareness among physi-cians of the apparently endemic tendency tooverestimate survival might help to counteractit. Education concerning the art of prognosticdisclosure could also help physicians overcomethe tendency to withhold or distort this infor-mation. Findings on patient preferences forcommunication,40 algorithms for deliveringprognosis,41 and primers on relational skillsfor physicians42 are widely available.

Finally, more and earlier referrals to hospicemay be facilitated by using interventions de-signed to capture patients in the preterminalstages of illness, and to provide education tothem and their families about disease processand prognosis, treatment options, and advancecare planning.43,44

Our previous work describing dying inAmerica attests to the benefits of hospiceservices for those who die at home.30

406 Vol. 30 No. 5 November 2005Schockett et al.

Unfortunately, hospice referral is made toolate in at least one in seven cases and, as thisstudy demonstrates, a ‘‘too late’’ referral canlead to bereaved family members’ greater dis-satisfaction, unmet needs, and concerns withthe quality of hospice services. In light of thefamily members’ responses, efforts must bemade to ensure that physicians, the gatekeep-ers to hospice,45,46 are better trained in end-of-life care issues and better equipped to en-gage patients and their loved ones earlier inplanning for quality end-of-life care.

AcknowledgmentsThis research was made possible by funding

from the Robert Wood Johnson Foundation.

References1. What is hospice and palliative care? Web Page.

Available from. http://www.nhpco.org/i4a/pages/index.cfm?pageid53281. Accessed June 18, 2003.

2. Christakis NA, Iwashyna TJ. The impact of indi-vidual and market factors on the timing of initiationof hospice terminal care. Med Care 2000;38:528--541.

3. NHPCO facts and figures. Web Page. Availablefrom. http://www.nhpco.org/files/public/Hospice_Facts_110104.pdf. Accessed November 22, 2004.

4. Sanders BS, Burkett TL, Dickinson GE,Tournier RE. Hospice referral decisions: the roleof physicians. Am J Hosp Palliat Care 2004;21:196--202.

5. Ogle K, Mavis B, Wang T. Hospice and primarycare physicians: attitudes, knowledge, and barriers.Am J Hosp Palliat Care 2003;20:41--51.

6. Becker JE. Oncology social workers’ attitudestoward hospice care and referral behavior. HealthSoc Work 2004;29:36--45.

7. Friedman BT, Harwood MK, Shields M. Barriersand enablers to hospice referrals: an expert over-view. J Palliat Med 2002;5:73--84.

8. Bradley EH, Cramer LD, Bogardus ST Jr, et al.Physicians’ ratings of their knowledge, attitudes,and end-of-life-care practices. Acad Med 2002;77:305--311.

9. Massarotto A, Carter H, MacLeod R,Donaldson N. Hospital referrals to a hospice: timingof referrals, referrers’ expectations, and the natureof referral information. J Palliat Care 2000;16:22--29.

10. Born W, Greiner KA, Sylvia E, Butler J,Ahluwalia JS. Knowledge, attitudes, and beliefsabout end-of-life care among inner-city African

Americans and Latinos. J Palliat Med 2004;7:247--256.

11. Christakis NA. Timing of referral of terminallyill patients to an outpatient hospice. J Gen InternMed 1994;9:314--320.

12. Christakis NA. Predicting patient survival beforeand after hospice enrollment. Hosp J 1998;13:71--87.

13. Brickner L, Scannell K, Marquet S, Ackerson L.Barriers to hospice care and referrals: survey ofphysicians’ knowledge, attitudes, and perceptionsin a health maintenance organization. J PalliatMed 2004;7:411--418.

14. Crawley LM, Marshall PA, Lo B, Koenig BA.Strategies for culturally effective end-of-life care.Ann Intern Med 2002;136:673--679.

15. Miller SC, Weitzen S, Kinzbrunner B. Factorsassociated with the high prevalence of short hospicestays. J Palliat Med 2003;6:725--736.

16. Hospice and palliative care information: historyof hospice. Web Page. Available from. http://www.nhpco.org/i4a/pages/index.cfm?pageid53285 Oc-tober 7, 2004.

17. Naik A, DeHaven MJ. Short stays in hospice.A review & update. Caring 2001;20:10--13.

18. Frantz TT, Lawrence JC, Somov PG, Somova MJ.Factors in hospice patients’ length of stay. Am JHosp Palliat Care 1999;16:449--454.

19. Miller S, Mor V, Gozalo P. Hospice lengths ofstay in nursing homes and factors influencing shortstays. Gerontologist 2000;40(Special Issue 1):302.

20. Somova MJ, Somov PG, Lawrence JC, Frantz TT.Factors associated with length of stay in a mid-sized,urban hospice. Am J Hosp Palliat Care 2000;17:99--105.

21. Christakis NA, Iwashyna TJ. Impact of individu-al and market factors on the timing of initiation ofhospice terminal care. Med Care 2000;38:528--541.

22. Christakis NA, Escarce JJ. Survival of Medicarepatients after enrollment in hospice programs [seecomments]. N Engl J Med 1996;335:172--178.

23. Virnig BA, Persily NA, Morgan RO, DeVito CA.Do Medicare HMOs and Medicare FFS differ intheir use of the Medicare hospice benefit? Hosp J1999;14:1--12.

24. Casarett DJ. Differences between patients re-ferred to hospice from academic vs. non-academicsettings. J Pain Symptom Manage 2001;21:197--203.

25. Casarett DJ, Hirschman KB, Henry MR. Doeshospice have a role in nursing home care at theend of life? J Am Geriatr Soc 2001;49:1493--1498.

26. Virnig BA, Kind S, McBean M, Fisher E.Geographic variation in hospice use prior to death.J Am Geriatr Soc 2000;48:1117--1125.

27. Miceli PJ, Mylod DE. Satisfaction of familiesusing end-of-life care: current successes and

Vol. 30 No. 5 November 2005 407Late Referral to Hospice and Quality of Care

challenges in the hospice industry. Am J HospPalliat Care 2003;20:360--370.

28. Miceli PJ, Wojciechowski SL. Impacting familysatisfaction with hospice care. Caring 2003;22:14--18.

29. Teno JM, Clarridge B, Casey V, Edgman-Levitan S, Fowler J. Validation of Toolkit After--Death Bereaved Family Member Interview. J PainSymptom Manage 2001;22:752--758.

30. Teno JM, Clarridge BR, Casey V, et al. Familyperspectives on end-of-life care at the last place ofcare. JAMA 2004;291:88--93.

31. Vigano A, Dorgan M, Buckingham J, Bruera E,Suarez-Almazor ME. Survival prediction in terminalcancer patients: a systematic review of the medicalliterature. Palliat Med 2000;14:363--374.

32. Glare P, Virik K, Jones M, et al. A systematic re-view of physicians’ survival predictions in terminallyill cancer patients. BMJ 2003;327:195.

33. Christakis NA, Lamont EB. Extent and determi-nants of error in doctors’ prognoses in terminally illpatients: prospective cohort study. BMJ 2000;320:469--472.

34. Lamont EB, Christakis NA. Prognostic disclo-sure to patients with cancer near the end of life.Ann Intern Med 2001;134:1096--1105.

35. Weeks JC, Cook EF, O’Day SJ, et al. Relationshipbetween cancer patients’ predictions of prognosisand their treatment preferences. JAMA 1998;279:1709--1714.

36. McPherson C, Addington-Hall J. Judging thequality of care at the end of life: can proxies providereliable information. Soc Sci Med 2003;56:95--109.

37. Lynn J, Teno JM, Harrell FE Jr. Accurate prog-nostications of death. Opportunities and challengesfor clinicians. West J Med 1995;163:250--257.

38. Morita T, Tsunoda J, Inoue S, Chihara S. ThePalliative Prognostic Index: a scoring system forsurvival prediction of terminally ill cancer patients.Support Care Cancer 1999;7:128--133.

39. Poses RM, McClish DK, Bekes C, Scott WE,Morley JN. Ego bias, reverse ego bias, and physi-cians’ prognostic. Crit Care Med 1991;19:1533--1539.

40. Wenrich MD, Curtis JR, Shannon SE, et al.Communicating with dying patients within the spec-trum of medical care from terminal diagnosis todeath. Arch Intern Med 2001;161:868--874.

41. Lamont EB, Christakis NA. Complexities inprognostication in advanced cancer: ‘‘to help themlive their lives the way they want to.’’ JAMA 2003;290:98--104.

42. von Gunten CF, Ferris FD, Emanuel LL. Thepatient-physician relationship. Ensuring competen-cy in end-of-life care: communication and relationalskills. JAMA 2000;284:3051--3057.

43. Stuart B, D’Onofrio CN, Boatman S, Feigelman G.CHOICES: promoting early access to end-of-life carethrough home-based transition management. J PalliatMed 2003;6:671--683.

44. Brumley RD, Enguidanos S, Cherin DA. Effec-tiveness of a home-based palliative care programfor end-of-life. J Palliat Med 2003;6:715--724.

45. Kolbe RH, Dwyer FR. Diagnosing the physicianas gatekeeper in hospice marketing. J Health CareMark 1986;6:23--33.

46. Gordon AK. The physician gatekeeper: access tothe Medicare hospice benefit. Am J Hosp Care 1989;6:44--47.