ORIGINAL ARTICLE

Timing of code status documentation and end-of-life outcomesin patients admitted to an oncology ward

Amanda Caissie & Nanor Kevork & Breffni Hannon &

Lisa W. Le & Camilla Zimmermann

Received: 4 June 2013 /Accepted: 11 September 2013 /Published online: 28 September 2013# Springer-Verlag Berlin Heidelberg 2013

AbstractPurpose Guidelines recommend documentation of care pref-erences for patients with advanced cancer upon hospital ad-mission. We assessed end-of-life outcomes for patients whodid or did not have code status (CS) documented within 48 hof admission.Methods This was a retrospective cohort study of patientswho died on an inpatient oncology ward between January2004 and February 2009. Primary end-of-life outcomes were“code blues” and cardiopulmonary resuscitation (CPR) at-tempts; secondary outcomes included unsuccessful CPR at-tempts, intensive care unit (ICU), consultations, and ICUadmissions. Using logistic regression, outcomes were com-pared between those with and without CS documentation≤48 h from admission (full code or do-not-resuscitate), con-trolling for significant confounders.

Results The 336 patients had a median age of 61 years; 97 %had advanced cancer. The median time from admission todeath was 12 days (range <1–197 days); 151 patients (45 %)had CS documentation ≤48 h from admission. Controlling forconfounders of reason for admission and marital status, pa-tients with CS documentation ≤48 h from admission hadfewer “code blues” (2 vs. 15 %; adjusted odds ratio (AOR)0.12, 95 % confidence interval (CI) 0.02–0.43), CPR attempts(1 vs. 11 %; AOR 0.12, 95 % CI 0.01–0.51), unsuccessfulCPR attempts (0 vs. 11 %), ICU consultations (9 vs. 30 %;AOR 0.19, 95 % CI 0.08–0.40) and ICU admissions (2 vs.5 %; AOR 0.18, 95 %CI 0.02–0.85).Conclusions In patients who died on an oncology ward, CSdocumentation within 48 h of admission was associated withless aggressive end-of-life care, regardless of the reason foradmission.

Keywords Cancer . Code status . Intensive care unit .

Cardiopulmonary resuscitation . Advance directives

Introduction

When first introduced around 1960, cardiopulmonary resus-citation (CPR) was used mainly intra-operatively [1]. In the1970s, The American Medical Association recommendeddocumentation of code status in the hospital chart, and hospi-tal policies made CPR the default unless do-not-resuscitate(DNR) orders were written [2]. Decades later, this policyremains in place, yet there are low rates of code status docu-mentation in hospitalized patients [3–5] Such documentationis important, given the low rates of CPR success in hospitalinpatients. Although approximately 4 out of 10 patients have areturn of spontaneous circulation, only 10–20 % survives tohospital discharge [6, 7].

A. CaissieDepartment of Radiation Oncology, Dalhousie University, Halifax,NS, Canada

B. Hannon :C. ZimmermannDivision of Medical Oncology and Hematology, Department ofMedicine, University of Toronto, Toronto, Canada

N. Kevork : B. Hannon :C. Zimmermann (*)Department of Psychosocial Oncology and Palliative Care, PrincessMargaret Hospital, Toronto, University Health Network, 610University Ave., 16-712, M5G 2M9 Toronto, ON, Canadae-mail: camilla.zimmermann@uhn.on.ca

C. ZimmermannCampbell Family Cancer Research Institute, Ontario CancerInstitute, Princess Margaret Hospital, Toronto, University HealthNetwork, Toronto, Canada

L. W. LeDepartment of Biostatistics, Princess Margaret Hospital, Toronto,University Health Network, Toronto, Canada

Support Care Cancer (2014) 22:375–381DOI 10.1007/s00520-013-1983-4

The success rate of CPR is even lower in patients withcancer, and is further reduced by metastatic disease, poorperformance status, advanced age, and acute illnesses includ-ing infection [6, 8, 9]. A meta-analysis of 42 studies from1966 to 2005, specifically in hospitalized patients with cancer,reported an overall survival rate to discharge of 6.2 % (9.5 %in patients with localized disease and 5.6 % in patients withmetastatic disease) [9]. For patients whose cardiac arrest wasanticipated, rather than occurring unexpectedly, the survivalrate to discharge after CPR was 0 % [8].

Due to these exceedingly low rates of success for CPR inhospitalized patients with cancer, guidelines recommend thatfor patients with advanced cancer, goals of care should bereviewed and documented within 48 h of any admission to ahospital [10]. While the latter publication does not explicitlydefine advanced cancer, it has been defined elsewhere ascancer that is incurable [11]. These recommendations forgoals of care discussions are in line with the Patient SelfDetermination Act, which requires that all institutions in theUSA that are receiving Medicare/Medicaid funds inform pa-tients, upon admission, of their right to accept or refusetreatment and their right to an advance directive, regardlessof whether or not their cancer is in the advanced stages [12].However, the guideline of documenting goals of care within48 h is founded only from level III evidence (textbooks,opinions, and descriptive studies) [10]. Scant evidence existsregarding both the frequency with which code status is actu-ally documented within 48 h of admission, and whether thisserves to reduce aggressive end-of-life care [13].

Early code status documentation may decrease the numberof patients subjected to aggressive interventions not appropri-ately aligned with goals of care. The aims of the current studywere (1) to assess the timing of code status documentation in allpatients who died after admission to an oncology ward and (2)to examine whether code status documentation within 48 h wasassociated with less aggressive end-of-life care. We hypothe-sized that code status documentation within 48 h would beassociated with fewer “code blues” and CPR attempts. Second-ary outcomes included unsuccessful CPR attempts and inten-sive care unit (ICU) consultations and admissions before death.

Materials and methods

Study site and sample

The study took place at Princess Margaret Hospital (PMH), acomprehensive cancer center and member of the UniversityHealth Network (UHN) in Toronto, Canada. In addition toPMH, the UHN also includes two general hospitals: TorontoGeneral Hospital and Toronto Western Hospital. The studysample included consecutive inpatients that died between Janu-ary 2004 and February 2009, inclusive, on the two general

oncology (medical or radiation oncology) wards at PMH, whichhave a total of 40 beds. Patients admitted to these wards all havesolid tumors, with the exception of a small minority of patientswith lymphoma or multiple myeloma who are admitted forpalliative radiation. PMH also has 69 beds allocated for patientswith hematologic malignancies and a 12-bed palliative care unit(PCU); patients who died on these units were not included in thestudy. PMH has no emergency department; however, patientshave access to the emergency departments of the other UHNhospitals. Patients may be admitted either directly from ambula-tory care; transferred from general medical or surgical wards atanother hospital; or transferred after presenting at the emergencydepartment at another hospital.While PMH has no ICU, patientshave access to ICU care at the adjacentMount Sinai Hospital viaa bridge connecting the two hospitals.

This research was conducted following approval of theUHN Research Ethics Board. Two investigators (AC andNK) conducted a thorough retrospective chart review, using astandardized abstraction spreadsheet. Data abstracted includedinformation on patient demographics, cancer diagnosis andstage, admission date, reason for admission, admitting service,status of the physician admitting the patient (patients may beadmitted onto the service of a staff physician by this physicianpersonally or by a hospitalist, fellow, or resident), code statusdocumentation, ICU consultations and admissions, CPR at-tempts (and outcome thereof), and date and cause of death.

Data collection was facilitated by the availability of data-bases and standardized forms. The reason for admission isspecified by the admitting physician at the time that the patientis placed on the waiting list for admission, and is then enteredinto the standardized database by the admissions coordinator(a registered nurse) using a drop-down menu of differentpotential reasons for admission. When there are multiplereasons for admission, the primary reason is recorded. For thisstudy, the 28 reasons for admission listed in the database weregrouped into three categories: symptom control, cancer treat-ment or investigation, and palliative planning (Table 1). Thelast category was taken unaltered from the database, and refersto patients whose condition is deteriorating at home, and whoare admitted with the anticipation of transfer to hospice orPCU or discharge home with further support.

Code status documentation and ICU consultations andadmissions were abstracted from the physician’s orders andfrom the chart notes. Code status documentation was takenfrom the inpatient medical orders, because according to UHNpolicy, this is where DNR documentation must be, to befollowed for inpatients. Code status was classified as “fullcode” or “DNR”. A DNR order precludes the interventionsof chest compressions, defibrillation, or intubation in the eventof a cardiopulmonary arrest, but does not preclude an ICUconsultation or admission. Details of code blues and CPRattempts and outcomes were abstracted from standardizedforms completed by the code blue team. The date and cause

376 Support Care Cancer (2014) 22:375–381

of death was obtained from the death certificate, a copy ofwhich was included in each chart.

A “code blue” was defined as any medical emergencywhere there was documented involvement of the code blueteam (cardiopulmonary arrest or near-cardiopulmonary ar-rest). CPR attempts were defined as instances where therewere documented attempts at resuscitation, including chestcompressions, defibrillation, and/or intubation. CPR was de-fined as “successful” if the patient was revived and survivedfor one or more days and “unsuccessful” if the patient was notrevived or survived for less than 1 day. All ICU admissions atPMH are preceded by an ICU consultation, but an ICUconsultation does not necessarily result in an admission tothe ICU. All code blues result in an ICU consultation; inaddition, the ICU staff may provide consultations on thePMH oncology wards for medical emergencies that have notyet progressed to the severity of a code blue. ICU consulta-tions may also be requested pre-emptively, to provide inputregarding whether or not CPR and ICU admission would beappropriate for a particular patient. These pre-emptive consul-tations are rare and supplement rather than replace the moreusual situation of a code status discussion with the primaryoncologist, admitting oncology team, or palliative care team.

A UHN policy is in place to guide the medical team throughdecision making with respect to life support interventions andcode status discussions. During the study period, there was noinstitutional policy or guideline regarding the actual timing ofsuch code status discussions, nor is there one currently.

Statistical analysis

Associations between timing of initial code status documen-tation (≤48 h from admission, vs. longer or not at all) andpatient characteristics were assessed using chi-square, Fisher’sexact, and t tests, as appropriate. Associations between timing

of code status documentation and end-of-life outcomes (codeblue, CPR attempt, unsuccessful CPR attempt, ICU consulta-tion, ICU admission) were then assessed using multivariablelogistic regression analyses, adjusting for significant con-founders. Confounders were defined as patient characteristicsassociated with both timing of code status documentation andany end-of-life outcome (p <0.05). Marital status and reasonfor admission met criteria as confounders and were includedin the multivariable analysis. Among the reasons for admis-sion, inclusion of “palliative planning” caused separation ofthe data, because a large majority of patients admitted forpalliative planning also had code status assessed within 48 hof admission. Therefore, patients admitted for palliative plan-ning were excluded from the multivariable analysis.

Results

Table 2 shows the timing of relevant end-of-life events for the336 patients who died on the oncology wards during the studyperiod. Of 336 patients, only 151 (45 %) had code statusdocumentation ≤48 h from admission (141 as DNR and 10as full code) and 28 (8 %) had no documentation of their codestatus before death. Of note, 16 patients had a change in theircode status: for 12 patients (whom had code status documen-tation within 48 h), an initial “full code” status was subse-quently changed to DNR; for 4 patients, (whom had codestatus documentation within 48 h) an initial DNR status waschanged to full code, and then back to DNR prior to death.Twenty-eight patients died within 48 h of admission; 23/28had code status documented, and 5/28 did not. Of the 23 whodid have code status documented, one (documented as full

Table 2 End-of-life milestones

End-of-life events All patientsa

Code status documented before death 308 (92)

Code status documentation within 48 h of admission 151 (45)

Code status documentation within 48 h of death 88 (26)

Time from admission to death; median (range), in days 12 (1–197)

Time from admission to initial documentation ofcode statusb; median (range), in days

3 (1–178)

Time from initial code status documentation to deathb;median (range), in days

6 (1–97)

Time from initial code status to final code statusc;median (range), in days

3 (1–39)

Time from final code status to deathc; median (range),in days

6 (1–34)

a Unless specified, units for all characteristics are number (%), and n =336b n=308 patients who had their code status documentedc n=16 patients who had a change in code status

Table 1 Categories of reasons for admission listed in oncology database

Category Reasons for admission listed in oncologydatabase

Symptom control Pain control, dehydration, neurological changes,biochemical abnormality, hematologicalabnormality, bleeding, bowel obstruction,nausea/vomiting, distal vein thrombosis/pulmonary embolus, superior vena cavaobstruction, seizures, hypercalcemia, fever notyet diagnosed, off-treatment deterioration,febrile neutropenia, infection, pleural effusion,respiratory problems, other medical problem

Cancer treatment orinvestigation

Radiation therapy, radiation chemotherapy,chemotherapy, investigation, cordcompression, staging, gastrostomy tube, on-treatment deterioration

Palliative planning Palliative planning

Support Care Cancer (2014) 22:375–381 377

code) was transferred to the ICU after a code blue and success-ful CPR. Of the five who did not have code status documented,all had unsuccessful CPR after a code blue. All of these patientshad advanced illness and deteriorating status on admission.

The demographics of all 336 patients are presented in Table 3,both for the entire study population and according to timing ofcode status documentation. Of all patients, (295/336) had stageIV disease on admission; of those with non-stage IV disease, allbut 10 (3 % of the whole sample) had advanced, incurabledisease at admission (for example advanced glioblastoma).Ninety-six percent died of their cancer, with the remaining 13patients dying either from complications of the disease or fromsevere medical comorbidities. Code status documentation ≤48 hafter admission was associated with being married and admis-sion for palliative planning and negatively associated with ad-mission for cancer treatment or investigation. Specifically, 29/34(85 %) of patients admitted for palliative planning, 87/195(45 %) admitted for symptom control, and 35/107 (33 %) ofpatients admitted for cancer treatment or investigation, had codestatus documentation within 48 h.

After excluding the 34 patients who were admitted forpalliative planning, and adjusting for other reason for admis-sion and marital status, patients with code status documenta-tion ≤48 h from admission had significantly fewer codeblues, CPR attempts, and ICU consultations and admis-sions (Table 4). All 20 CPR attempts in the group that did nothave code status documented within 48 h after admission wereunsuccessful (for 18 attempts the patient was not revived; for2, the patient survived less than 24 h), whereas the 2 in the ≤48group (both “full code” documentations) were successful inreviving the patient. In both of these cases with successfulCPR attempts, the patient’s code status was changed to DNRupon transfer back to the oncology ward (3 to 11 days post-CPR), and cancer-related death occurred within 3 weeks ofinitial resuscitation.

We repeated the analyses discounting the patients who diedwithin 48 h of admission, and the results remained significantfor all outcomes (p <0.05).

Discussion

Current recommendations for patients with advanced cancerare to discuss and document goals and preferences for carewithin 48 h of admission to hospital [10]. This 48-h windowallows for appropriate code status documentation, while leav-ing time for the oncology team to gather information that mayaid in the discussion, including the input of the primarytreating oncologist. The results of our study support documen-tation of code status within this time frame.

The large majority of patients who died on the inpatientoncology service during the 5-year study period had incurablecancer. Although code status was documented in more than

90 % before they died, documentation was completed within48 h of admission in only 45 % of the overall sample, and didnot take place consistently even in patients admitted for pal-liative planning. In those patients who had code status docu-mented within 48 h, there was less aggressive end-of-life carefor all measured outcomes, even after adjusting for confound-ing variables of reason for admission and marital status.

Other studies have demonstrated the effect of end-of-lifediscussions on less aggressive medical care near death, dem-onstrating lower rates of ventilation, resuscitation, and ICUadmissions [14, 15]. In a recent study of patients with ad-vanced lung or colorectal cancer, those who had end-of-lifediscussions before the last 30 days of life were less likely toreceive aggressive measures at the end-of-life, including che-motherapy and acute hospital-based care [16]. As well, in aretrospective study of 118 terminally ill oncology inpatients,earlier recognition that the patient was dying was associatedwith timelier establishment of goals of care, including earlierDNR code status documentation and discontinuation of anti-cancer therapy [13]. Our study adds to this body of literatureencouraging appropriate discussion of end-of-life planning,by specifically supporting the documentation of code status inpatients with advanced cancer within 48 h of hospitaladmission.

It is increasingly accepted that aggressive end-of-life carefor patients with advanced cancer is not only cost-ineffectivebut also represents poor quality care [17–20]. CPR is a highlycostly intervention [21], and the very small chance that pa-tients with advanced cancer have of surviving to dischargeafter a cardiopulmonary arrest [9] approaches zero when thearrest is anticipated [8]. In the current study, 11 % of thosewho had their code status documented greater than 48 h afteradmission died after an unsuccessful CPR attempt, comparedto none in the ≤48 h group. This is an unpleasant way to dieand can have adverse effects for all involved. Aggressive carein the final week of life has been associated with poor patientquality of life, as well as with a higher risk ofmajor depressionin bereaved caregivers [15]. Unsuccessful CPR attempts arealso highly disturbing for those performing it; generally, phy-sicians in training [22].

Code status discussions are complex and challenging,and physicians may feel reluctant to discuss such a sensi-tive topic [23]. Patients generally have a poor understand-ing of CPR and its low success rate in patients withadvanced illnesses [24] and this is important to convey ina clear and sensitive manner [25]. Evidence suggests thatpatients are receptive to discussions about code status andother advance directives on admission to hospital [26, 27],and that code status discussions at admission do not affectpatient or surrogate satisfaction with care [28]. Indeed,having a DNR order at the time of death has been associ-ated with higher quality of end-of-life care ratings byfamily members [29]. Ideally, these discussions take place

378 Support Care Cancer (2014) 22:375–381

Table 3 Patient characteristics according to timing of code status

Characteristic All patients (n =336) Code status documented within 48 h of admission Univariable

Numbers (%) Yes (n =151) No (n =185) OR (95 % CI) p Value

Age (years)

median (range) 61 (23–94) 63 (23–87) 61 (27–94) 1.00 (0.98-1.01) 0.71

Sex

Female 186 (55) 81 (54) 105 (57) 0.88 (0.57–1.36) 0.57

Male 150 (45) 70 (46) 80 (43) –

Marital status

Married/common law 229 (69) 112 (76) 117 (64) 1.76 (1.08–2.84) 0.02

Not marrieda 102 (31) 36 (24) 66 (36) –

Unknown 5 3 2 –

Interpreter needed

Yes 37 (11) 20 (87) 17 (9) 1.51 (0.76–3.00) 0.24

No 299 (89) 131 (13) 168 (91) –

Religious affiliation

Yes 136 (41) 65 (43) 71 (38) 1.21 (0.78–1.88) 0.39

No 200 (59) 86 (57) 114 (62) –

Primary cancer site 0.16

Gastrointestinal 74 (22) 32 (21) 42 (23) 1.27 (0.62–2.58)

Lung 64 (19) 37 (25) 27 (15) 2.28 (1.10–4.76)

Gynecological 55 (16) 22 (15) 33 (18) 14.11 (0.52–2.39)

Breast 30 (9) 13 (9) 17 (9) 1.28 (0.52–3.14)

Genitourinary 30 (9) 17 (11) 13 (7) 2.18 (0.88–5.37)

Head and neck 27 (8) 9 (6) 18 (10) 0.83 (0.32–2.19)

Other 56 (17) 21 (14) 35 (19) –

Stage at admission

Stage IV 295 (88) 134 (89) 161 (87) 1.18 (0.61–2.28) 0.63

Other 41 (12) 17 (11) 24 (13) –

Cause of death

Cancer 323 (96) 147 (97) 176 (95) 1.88 (0.57–6.23) 0.30

Other 13 (4) 4(3) 9 (5) –

Admitting service

Medical oncology 249 (74) 119 (79) 130 (70) 1.57 (0.95–2.60) 0.08

Radiation oncology 87 (26) 32 (21) 55 (30) –

Admitting physician

Staff/hospitalist 155 (46) 68 (45) 87 (47) 0.92 (0.60–1.42) 0.72

Otherb 181 (54) 83 (55) 98 (53) –

Reason for admission <0.0001

Palliative planning 34 (10) 29 (19) 5 (3) 7.20 (2.67–19.37)

Cancer treatment or investigation 107 (32) 35 (23) 72 (39) 0.60 (0.37–0.99)

Symptom Control 195 (58) 87 (58) 108 (58) –

Time from diagnosis to admission 0.27

>1 year 103 (31) 53 (35) 50 (27) 1.54 (0.88–2.69)

>1 month to ≤1 year 135 (40) 58 (38) 77 (42) 1.09 (0.64–1.85)

≤1 month 98 (29) 40 (26) 58 (31) –

a Not married includes single, separated, divorced or widowedbOther includes fellows, residents and unknown

Support Care Cancer (2014) 22:375–381 379

with the primary oncologist prior to admission as a part ofadvance care planning, in which case, the status just needsto be reverified and documented on admission to a hospital.

Patients and family members may be reassured byknowing that a DNR status does not preclude otheractive treatments directed either at the cancer itself or atcomplications such as infections [30]. As well, not all codestatus discussions necessarily result in a DNR order. Interest-ingly, in the current study, all those patients who initiallyhad “full code” status documentation (whether before orafter 48 h from admission) either had successful codes(n =2), or had their code status subsequently changed toDNR before a code blue was called (n =10). This mayindicate that early code status discussions resulting in fullcode orders may also allow patients and substitute-decisionmakers the time to consider less aggressive end-of life careand reconsider DNR orders.

In the present study, only 33 % of those admitted forcancer treatment or investigation had documentation ofcode status within 48 h. However, the large majority ofthe latter group of patients had incurable disease, and thetreatments offered were thus of palliative intent. Patientsundergoing such treatments may have misconceptions re-garding the purpose of palliative treatment or of the poorprognosis associated with their disease [31]. Of note, therewas an increased rate of DNR discussions ≤48 h fromadmission in patients with a spouse/partner, which is sim-ilar to results of another study documenting lower rates ofCPR in married individuals [32]. These results may reflectan opportunity for discussions about advance directivesamongst partners and/or the desire of patients not to placethe burden of end-of-life decisions on their spouse in the caseof incompetence to make decisions.

Our study has strengths and limitations. Unlike otherstudies [4, 5, 13], we used as our main variable timing ofany code status documentation, including “full code” aswell as DNR. We felt that what was important was thediscussion regarding code status and the documentationof this decision, regardless of whether it was full code orDNR. It is possible that additional code status discus-sions took place, for which the outcome was not docu-mented as an order, particularly as a full code is impliedunless a DNR order is written. Other limitations of thisstudy include its retrospective nature and the collectionof data from a single tertiary cancer care center. Thecohort of patients examined all died on the oncologywards; thus we do not have information on patientswho were admitted and survived, or who were admittedand subsequently died in a different setting after beingtransferred. In particular, this likely resulted in an under-estimation of ICU admissions, because patients who weretransferred to the ICU and died there were not includedin the study.

Routine discussion and documentation of code statuson admission in all patients with advanced cancer canavoid misunderstandings regarding patient wishes andprevent aggressive end-of-life care. It has been recom-mended that all healthcare institutions should have a “donot attempt resuscitation” policy [33]. Many health carefacilities, including our own, have established policies toguide clinical-decision making with respect to wishesregarding CPR as a treatment and establishing limitswhen CPR is deemed almost certainly of no benefit toa patient. The results of this study provide support for theinclusion within such policies of guidelines for routine docu-mentation of code status within 48 h.

Table 4 Timing of code status and end-of-life outcomes

Code status ≤48 h OR (95 % CI) Exact p Value AOR (95 % CI) Exact p Value

Yes (n =151) No (n =185)

Code blue 3 (2 %) 28 (15 %) 0.11 (0.03–0.38) 0.0004 0.12 (0.02–0.43) 0.0001

CPR attempt 2 (1 %) 20 (11 %) 0.11 (0.03–0.48) 0.0005 0.12 (0.01–0.51) 0.001

Unsuccessful CPR attempt 0 (0 %) 20 (11 %) NE NE NE NE

ICU consultation 13 (9 %) 56 (30 %) 0.22 (0.11–0.42) <0.0001 0.19 (0.08–0.40) <0.0001

ICU admission 3 (2 %) 11 (5 %) 0.32 (0.09–1.17) 0.10 0.18 (0.02–0.85) 0.03

AOR adjusted odds ratio (adjusting for reason for admission and marital status), CPR cardio-pulmonary resuscitation, ICU intensive care unit, NE notevaluable (due to zero value for outcome with code status ≤48 h)

380 Support Care Cancer (2014) 22:375–381

Disclosures There are no financial disclosures from any of the authors.This research was funded by the Canadian Cancer Society (grant #700862;CZ), the Bluma Appel Research Fund and the Ontario Ministry of Healthand Long Term Care. Dr. Zimmermann is supported by the Rose Chair inSupportive Care, Faculty of Medicine, University of Toronto.

References

1. Kouwenhoven WB, Jude JR, Knickerbocker GG (1960) Closed-chest cardiac massage. JAMA 173:1064–1067

2. Loertscher L, Reed DA, Bannon MP, Mueller PS (2010) Cardiopul-monary resuscitation and do-not-resuscitate orders: a guide for clini-cians. Am J Med 123(1):4–9

3. Fairchild A, Kelly KL, Balogh A (2005) In pursuit of an artful death:discussion of resuscitation status on an inpatient radiation oncologyservice. Support Care Cancer 13(10):842–849

4. Middlewood S, Gardner G, Gardner A (2001) Dying in hospital:medical failure or natural outcome? J Pain Symptom Manage 22(6):1035–1041

5. Becker G, Sarhatlic R, OlschewskiM,Xander C,MommF, BlumHE(2007) End-of-life care in hospital: current practice and potentials forimprovement. J Pain Symptom Manage 33(6):711–719

6. Ebell MH, Afonso AM (2011) Pre-arrest predictors of failure tosurvive after in-hospital cardiopulmonary resuscitation: a meta-analysis. Fam Pract 28(5):505–515

7. Peberdy MA, Kaye W, Ornato JP, Larkin GL, Nadkarni V, ManciniME et al (2003) Cardiopulmonary resuscitation of adults in thehospital: a report of 14720 cardiac arrests from the National Registryof Cardiopulmonary Resuscitation. Resuscitation 58(3):297–308

8. Ewer MS, Kish SK, Martin CG, Price KJ, Feeley TW (2001) Char-acteristics of cardiac arrest in cancer patients as a predictor of survivalafter cardiopulmonary resuscitation. Cancer 92(7):1905–1912

9. Reisfield GM, Wallace SK, Munsell MF, Webb FJ, Alvarez ER,Wilson GR (2006) Survival in cancer patients undergoing in-hospital cardiopulmonary resuscitation: a meta-analysis. Resuscita-tion 71(2):152–160

10. Walling A, Lorenz KA, Dy SM, Naeim A, Sanati H, Asch SM et al(2008) Evidence-based recommendations for information and careplanning in cancer care. J Clin Oncol 26(23):3896–3902

11. Peppercorn JM, Smith TJ, Helft PR, Debono DJ, Berry SR, WollinsDS et al (2011) American Society of Clinical Oncology statement:toward individualized care for patients with advanced cancer. J ClinOncol 29(6):755–760

12. Library of Congress. Bill Text, 101st Congress (1989–1990),H.R.4449.IH. Patient Self Determination Act of 1990 . http://thomas.loc.gov/cgi-bin/query/z?c101:H.R.4449.IH. Accessed 20 May 2013

13. Hui D, Con A, Christie G, Hawley PH (2009) Goals of care and end-of-life decisionmaking for hospitalized patients at a Canadian tertiarycare cancer center. J Pain Symptom Manage 38(6):871–881

14. Bradley EH, Hallemeier AG, Fried TR, Johnson-Hurzeler R, CherlinEJ, Kasl SV et al (2001) Documentation of discussions about prog-nosis with terminally ill patients. Am J Med 111(3):218–223

15. Wright AA, Zhang B, Ray A, Mack JW, Trice E, Balboni T et al(2008) Associations between end-of-life discussions, patient mentalhealth, medical care near death, and caregiver bereavement adjust-ment. JAMA 300(14):1665–1673

16. Mack JW, Cronin A, Keating NL, Taback N, Huskamp HA, Malin JLet al (2012) Associations between end-of-life discussion characteristics

and care received near death: a prospective cohort study. J Clin Oncol30(35):4387–4395

17. Earle CC, Park ER, Lai B, Weeks JC, Ayanian JZ, Block S (2003)Identifying potential indicators of the quality of end-of-life cancercare from administrative data. J Clin Oncol 21(6):1133–1138

18. Earle CC, LandrumMB, Souza JM, Neville BA, Weeks JC, AyanianJZ (2008) Aggressiveness of cancer care near the end of life: is it aquality-of-care issue? J Clin Oncol 26(23):3860–3866

19. Grunfeld E, Lethbridge L, Dewar R, Lawson B, Paszat LF, JohnstonG et al (2006) Towards using administrative databases to measurepopulation-based indicators of quality of end-of-life care: testing themethodology. Palliat Med 20(8):769–777

20. Barbera L, Paszat L, Chartier C (2006) Indicators of poor quality end-of-life cancer care in Ontario. J Palliat Care 22(1):12–17

21. Lee KH, Angus DC, Abramson NS (1996) Cardiopulmonary resus-citation: what cost to cheat death? Crit Care Med 24(12):2046–2052

22. Menezes BF, Morgan R (2008) Attitudes of doctors in training tocardiopulmonary resuscitation. Clin Med 8(2):149–151

23. Sulmasy DP, Sood JR, Ury WA (2008) Physicians’ confidence indiscussing do not resuscitate orders with patients and surrogates. JMed Ethics 34(2):96–101

24. Heyland DK, Frank C, Groll D, Pichora D, Dodek P, Rocker G et al(2006) Understanding cardiopulmonary resuscitation decision mak-ing: perspectives of seriously ill hospitalized patients and familymembers. Chest 130(2):419–428

25. Volandes AE, Paasche-Orlow MK, Mitchell SL, El Jawahri A, DavisAD, Barry MJ et al (2013) Randomized controlled trial of a videodecision support tool for cardiopulmonary resuscitation decisionmaking in advanced cancer. J Clin Oncol 31(3):380–386

26. Reilly BM, Magnussen CR, Ross J, Ash J, Papa L, Wagner M (1994)Can we talk? Inpatient discussions about advance directives in acommunity hospital. Attending physicians’ attitudes, their inpatients’wishes, and reported experience. Arch Intern Med 154(20):2299–2308

27. Reilly BM,Wagner M, Ross J, Magnussen CR, Papa L, Ash J (1995)Promoting completion of health care proxies following hospitaliza-tion. A randomized controlled trial in a community hospital. ArchIntern Med 155(20):2202–2206

28. Anderson WG, Pantilat SZ, Meltzer D, Schnipper J, Kaboli P,Wetterneck TB et al (2011) Code status discussions at hospitaladmission are not associated with patient and surrogate satisfactionwith hospital care: results from the multicenter hospitalist study. AmJ Hosp Palliat Care 28(2):102–108

29. Finlay E, Shreve S, Casarett D (2008) Nationwide veterans affairsquality measure for cancer: the family assessment of treatment at endof life. J Clin Oncol 26(23):3838–3844

30. Smith CB, Bunch OL (2008) Do not resuscitate does not mean do nottreat: how palliative care and other modalities can help facilitatecommunication about goals of care in advanced illness. Mt Sinai JMed 75(5):460–465

31. Guo Y, Palmer JL, Bianty J, Konzen B, Shin K, Bruera E (2010)Advance directives and do-not-resuscitate orders in patients withcancer with metastatic spinal cord compression: advanced care plan-ning implications. J Palliat Med 13(5):513–517

32. Chen JS, Wang HM, Wu SC, Liu TW, Hung YN, Tang ST (2009) Apopulation-based study on the prevalence and determinants of car-diopulmonary resuscitation in the last month of life for Taiwanesecancer decedents, 2001–2006. Resuscitation 80(12):1388–1393

33. Deakin CD (2005) New standards for cardiopulmonary resuscitation.BMJ 330(7493):685–686

Support Care Cancer (2014) 22:375–381 381