Survival Duration among Patients with a Noncancer Diagnosis Admitted to a Palliative Care Unit: A Retrospective Study

Abstract

Background:

Palliative care unit (PCU) beds are a limited resource in Canada, so PCU admission is restricted to patients with a short prognosis. Anecdotally, PCUs further restrict admission of patients with noncancer diagnoses out of fear that they will “oversurvive” and reduce bed availability. This raises concerns that noncancer patients have unequal access to PCU resources.

Purpose/Methods:

To clarify survival duration of patients with a noncancer diagnosis, we conducted a retrospective review of all admissions to four PCUs in Toronto, Canada, over a 1-year period. We measured associations between demographic data, prognosis, Palliative Performance Score (PPS), length of stay (LOS), and waiting time.

Results:

We collected data for 1000 patients, of whom 21% had noncancer diagnoses. Noncancer patients were older, with shorter prognoses and lower PPS scores on admission. Noncancer patients had shorter LOS (14 versus 24, p<0.001) than cancer patients and a similar likelihood of being discharged alive to cancer patients. Noncancer patients had a trend to lower LOS across a broad range of demographic, diagnostic, prognostic, and PPS categories. Multivariable analysis showed that LOS was not associated with the diagnosis of cancer (p=0.36).

Discussion/Conclusion:

Noncancer patients have a shorter LOS than cancer patients and a similar likelihood of being discharged alive from a PCU than cancer patients, and the diagnosis of cancer did not correlate with survival in our study population. Our findings demonstrate that noncancer patients are not “oversurviving,” and that referring physicians and PCUs should not reject or restrict noncancer referrals out of concern that these patients are having a detrimental impact on PCU bed availability.

Introduction

Many dying patients are admitted to inpatient palliative care units (PCUs) prior to their death, either because of symptom or support needs, or a patient/family preference not to die at home. The timing of admission is important—a longer stay in a PCU has been associated with a higher degree of satisfaction,¹ but PCU beds are a limited resource that cannot accommodate a large number of prolonged admissions.

In Ontario, Canada, inpatient PCUs generally restrict admission to patients with an anticipated prognosis of less than 3 months. This restriction is understandable but sometimes problematic, as physicians are notoriously inaccurate when it comes to prognostication,^2–4 and noncancer diagnoses in particular often defy prognosis.^2,5,6 Anecdotally, concerns about noncancer patients “oversurviving” their prognosis has led to a practice of restricting admissions for such patients until very late in the disease course. This raises concerns that patients with noncancer diagnoses may have unequal access to palliative care resources.⁷

In the present study, we conducted a retrospective review of admissions to multiple PCUs in Toronto, Canada, over a 1-year period to determine whether noncancer patients are “oversurviving” their prognosis.

Methods

Study design and setting

We conducted a retrospective chart review of all patients admitted to four inpatient PCUs in Toronto, Canada, from April 1, 2009 to March 31, 2010. The four PCUs were located in various settings: Toronto Grace Hospital is a nonacute care hospital specializing in chronic disease management, Toronto East General Hospital is a community academic hospital with a regional PCU on site, Sunnybrook Health Sciences Centre is a tertiary academic hospital with a regional PCU located on its campus, and St. Michael's Hospital is a tertiary academic hospital with a regional PCU in the main hospital building.

Most provincially funded PCUs in Ontario (including the four in our study) limit admission to patients with an anticipated prognosis of less than 3 months. Patient prognosis is determined by the referring physician, either based on the Palliative Performance Score (PPS) or by subjective judgment. Occasionally, PCUs will contact the referring physician to clarify prognosis or care needs before deciding whether or not to admit a patient.

Patients

We included all patients admitted to one of the four participating PCUs during the study period. There were no exclusion criteria. This study was approved by the Research Ethics Board of all participating facilities.

Data collection and analysis

In Toronto, all PCUs require referring physicians to complete a specially designed Palliative Care Common Referral Form (CRF) for prospective admissions. The PCU staff then reviews the CRF and decides whether or not the patient is appropriate for admission, or whether more information is needed. The PCU staff may also need to triage admissions based on urgency if there is a waiting list for admission.

For the present study, we collected admission and discharge/death dates, referral dates, demographic data, admitting diagnosis, PPS, estimated prognosis, comorbidities, source of admission and disposition following discharge from the CRF, the medical record, or from local administrative databases (where available) at each PCU. If more than one CRF was sent, we used the latest one. We excluded the source of referral for 158 patients at one facility because the data could not be interpreted.

Prognostic information is provided in the CRF at the time of referral by the referring physician. Physicians are asked to select one of several categories: <1 month, <3 months, <6 months, <1 year, or indeterminate.

We compared demographic data, length of stay (LOS), prognosis, and PPS between cancer and noncancer patients using Welch's t test. We measured correlations between LOS and patient characteristics using Pearson's test. To perform multivariable regression analyses, we grouped PPS scores into low (10–20), medium (30–40) and high (50+).⁸ We constructed a multivariable regression model for LOS that included age and gender, as well as all factors that correlated with LOS on bivariate analysis with a p<0.05 (the PPS and prognostic categories as well as a cancer diagnosis). We also constructed a multivariable regression model for waiting time that included all factors associated with waiting time on bivariate analysis with a p<0.05 (age; gender; prognosis <90 days, <180 days, and <365 days; PPS 30–40; PPS 50+; cancer; and the location of the patient at the time of referral [e.g., inpatient ward versus home, emergency department, or other medical facility]). All statistical tests were performed using SPSS version 15.0 (SPSS Inc., Chicago, IL).

Results

In total, we collected data for 1000 patients at four sites. Demographic data are presented in Table 1. Compared with cancer patients, noncancer patients were older, with a lower PPS and a nonsignificant trend toward a shorter prognosis. Noncancer patients were also less likely to survive until discharge or be discharged home from a PCU, although the difference was nonsignificant. Only 2% of noncancer patients were admitted to a PCU for more than 3 months, compared with 5% of cancer patients (p=0.1; data not shown).

Table 1.

Demographics

	Overall	Noncancer	Cancer
Number of cases	1000	209 (21%)	791 (79%)
Age (mean)	73	81^a	71^a
Gender (% male)	48	44	49
Admitted from
Home	87 (9%)	4 (2%)	83 (10%)
Emergency Department	25 (5%)	15 (7%)	37 (5%)
Inpatient Ward (same facility)	380 (38%)	99 (47%)	281 (36%)
Other Healthcare Facility	221 (22%)	29 (14%)	192 (24%)
Other/Not Available	260 (26%)	62 (30%)	197 (25%)
Prognosis available
<30 days (n, %)		74 (46%)	73 (16%)
30–90 days (n, %)		85 (53%)	334 (73%)
91–180 days (n, %)		1 (1%)	41 (9%)
181–365 days (n, %)		0	9 (2%)
Not Available (n)		49	334
Palliative Performance Score (Mean)	35	21^a	40^a
Discharge disposition
Deceased	909 (91%)	199 (95%)	708 (90%)
Alive	93 (9%)	10 (5%)	83 (10%)
Home	35 (4%)	3 (1%)	32 (4%)
Other Healthcare Facility	50 (5%)	7 (3%)	43 (5%)
Other/Not Available	8 (1%)	0 (0%)	8 (1%)

p<0.0001 for comparison between cancer and noncancer patients.

The mean LOS for noncancer and cancer patients is shown in Table 2, and displayed by subgroup according to various characteristics. Overall, noncancer patients had a significantly shorter LOS than cancer patients (14 versus 24 days, p<0.001). A nonsignificant trend to shorter LOS among noncancer patients was evident across all age groups 40 years and older, in both genders, from every admission source, and across all four participating sites (data not shown). Noncancer patients had a significantly shorter LOS than cancer patients with a prognosis <3 months; there was a nonsignificant trend toward shorter LOS among noncancer patients across all PPS scores (Table 3). The mean LOS for cancer and noncancer patients discharged alive was 40 and 53 days, respectively.

Table 2.

Mean Length of Stay in Days by Demographic, Admission, and Discharge Characteristics

Characteristic	Cases (noncancer/cancer)	Overall	Noncancer	Cancer
Entire cohort	211/825	22	14^a	24^a
25^th percentile		4	2	5
50^th percentile		10	6	13
90^th percentile		58	25	66
Age
<40	3/15	28	32	27
40–59	11/133	21	7	22
60–79	53/389	22	12	24
80–89	99/210	23	15	27
>89	43/42	19	13	25
Gender
Male	93/391	20^b	9^b	22^c
Female	116/399	25^b	18^b	27^c
Admitted from
Home	4/83	23	10	24
Emergency Department	28/61	18	14	20
Inpatient Ward	99/281	19	9	23
Other Healthcare Facility	29/192	25	18	26
Other/Not Available	3/61	26	17	27
Discharge disposition
Deceased	199/708	20	12^a	22^a
Alive	10/83	42	53	40
Home	3/32	35	35	35
Other Healthcare Facility	7/43	49	60	47
Other/Not Available	0/8	28	–	28

p<0.001 for comparison between cancer and noncancer patients.

p<0.005 for comparison between male and female.

p<0.02 for comparison between male and femal

Table 3.

Mean Length of Stay in Days by Prognosis and Palliative Performance Score

Characteristic	Cases (noncancer/cancer)	Overall LOS	Noncancer LOS	Cancer LOS	P value (noncancer versus cancer
Prognostic category
<30 days	73/74	9	7	11	0.17
30–90 days	83/330	22	12	24	<0.0001
91–180 days	2/35	34	52	33	0.78
181–365 days	0/9	17	–	17	–
Not available	51/343	26	25	26	–
Palliative Performance Score (PPS)
PPS 10%	53/21	5	5	7	0.47
PPS 20%	20/21	9	8	9	0.86
PPS 30%	22/79	19	15	20	0.21
PPS 40%	6/102	21	16	21	0.41
PPS 50%	2/60	24	10	24	0.36
PPS 60%	0/27	28	–	28	–
PPS 70%	0/8	27	–	27	–
PPS 80%	0/1	22	–	22	–
PPS 90%	1/0	8	8	–	–
PPS 100%	1/0	19	19	–	–
Not available	104/472	25	19	2

LOS, length of stay.

Table 4 shows the LOS, prognosis, and PPS for patients by disease group. Aside from heart failure, the mean LOS for every noncancer diagnosis was shorter than the mean LOS for every cancer diagnosis (all nonsignificant). Mean PPS scores were lowest for brain injury and stroke; other noncancer diagnoses had broadly similar mean PPS scores. All cancer diagnoses had higher mean PPS scores than all noncancer diagnoses, and cancer PPS scores were broadly similar to one another.

Table 4.

Mean Length of Stay, Prognosis, and Palliative Performance Score by Diagnosis

	Length of stay (d)			Prognosis (d)			PPS
Diagnosis	Cases	Mean	Median	Cases	Mean	Median	Cases	Mean	Median
Noncancer	209	14^a	6	160	63^a	90	108	20	20
Heart Disease	45 (22%)	25	6	32	64	90	25	27	30
Dementia	31 (15%)	10	8	20	69	90	9	18	20
Brain Injury	26 (12%)	8	6	23	59	30	19	12	10
Respiratory Disease	22 (11%)	15	5	17	76	90	10	29	25
Stroke	22 (11%)	11	6	19	57	30	14	13	10
Other	63 (30%)	11	6	47	60	30	29	22	20
Cancer	791	24^a	13	448	93^a	90	318	39	40
Lung	192 (24%)	29	15	120	88	90	76	40	40
Colorectal	73 (9%)	22	10	46	99	90	39	37	40
Brain	58 (7%)	35	23	27	80	90	22	30	30
Breast	54 (7%)	24	12	34	107	90	23	40	40
Hematologic	54 (7%)	26	13	27	106	90	18	37	40
Head and Neck	50 (6%)	21	8	23	95	90	14	44	40
Pancreas	47 (6%)	19	12	25	78	90	21	36	40
Liver/Biliary Tract	46 (6%)	24	14	25	79	90	18	41	40
Urologic	46 (6%)	28	16	27	93	90	17	35	30
Prostate	42 (5%)	21	10	19	108	90	14	46	50
Other	129 (16%)	16	10	75	95	90	56	39	40

p<0.01 for camparison between cancer and noncancer patients.

PPS, Palliative Performance Score.

Table 5 shows the mean and median waiting time (defined as the time between the date of initial referral to PCU and the date of admission to that PCU). The mean waiting time for noncancer patients was significantly shorter than that of cancer patients.

Table 5.

Mean Waiting Time in Days from Application to Admission

	Overall	Noncancer	Cancer
Mean Waiting Time (SD)	8 (19)	3^a (5)	9^a (21)
Median Waiting Time	3	2	4

p<0.0001 for comparison between noncancer and cancer.

SD, standard deviation.

Paired prognosis and PPS scores were not available for all patients. For patients for whom both data were recorded, we used Spearman's test to look for a correlation between LOS, prognosis, and PPS. LOS was weakly correlated with prognosis in both the noncancer and cancer populations (n=158 and n=448; R=0.24 and 0.26, respectively), and with the PPS score for cancer patients (n=318; R=0.24). LOS was more strongly correlated with PPS score in the noncancer population (n=106; R=0.37).

Multivariable regression revealed that LOS correlated with a prognosis of <6 months (p<0.02) and a PPS of 30 to 40 (p<0.02) and 50+(p<0.01), but not the diagnosis of cancer (p=0.36). Waiting time was correlated with gender (p<0.01), a prognosis of <1 year (p<0.01), and the location of the patient (p<0.01) but not with the diagnosis of cancer (p=0.75) on multivariable analysis.

In a post hoc analysis, we looked for differences in waiting time according to the location of the patient at the time of referral. Patients referred from another medical facility waited an average of 4.66 (SE 1.01) days longer than patients referred from an inpatient ward in the same facility (p<0.01). Even in the multivariable model, patients referred from other medical facilities waited 4.6 (standard error [SE] 1.3) days longer than those on an inpatient ward. Furthermore, adding the referral source to the multivariable model eliminated the significant relationship between PPS score and waiting time, as well as the relationship between a prognosis of <6 months and waiting time; the relationship between prognosis of <1 year and waiting time remained significant.

Discussion

In our study, we found that patients admitted to PCU with noncancer diagnoses were older, with lower PPS scores and shorter prognoses. Surprisingly, we also found that the LOS for noncancer patients was shorter than for cancer patients, even within the same demographic, prognostic, and PPS categories, and that noncancer patients were equally likely to be discharged alive from a PCU. Multivariable analysis further demonstrated that a noncancer diagnosis was not associated with either LOS or waiting time.

Similar to other studies, we found that clinician estimates of prognosis generally overestimate survival.^2–4 We also found a correlation between PPS and survival.^5,9–12 It was not surprising, therefore, to find that noncancer patients had both a lower PPS and a lower survival. However, we did not expect to find a lower survival even when comparing patients with similar prognoses and PPS scores, or that the diagnosis of cancer was not correlated with survival on multivariable analysis. Christakis and Lamont found that noncancer patients referred for hospice care were more likely to outlive their prognosis than cancer patients,² whereas Lau and colleagues found that noncancer patients had a lower hazard ratio for mortality than cancer patients once they adjusted for PPS and other significant covariates.⁵ Fox and coworkers found that a majority of noncancer patients meeting hospice criteria survived beyond the expected 6-month prognosis.⁶ On the other hand, one recent study has found that cancer and noncancer patients referred to a palliative care service were equally likely to defy their prognosis,¹³ although this latter study used specialist palliative care physicians to assess prognosis.

The reason for the inconsistency between our findings and previous studies is unclear. The difference would not be explained by PCUs giving higher priority for admission to cancer patients (thus lengthening their stay in the PCU as opposed to their previous location), because noncancer patients actually had a shorter waiting time between referral and admission. If a selection bias was responsible for our results, it would have to be further upstream, either by physicians failing to refer noncancer patients for PCU admission until very late in their illness, or by PCU directors preferentially rejecting many of them. We cannot exclude the possibility that end-stage cancer and noncancer patients have similar prognoses (all other factors being equal), but we cannot confirm this conclusion using data from such a selected population.

Anecdotally, many PCU directors in Ontario are reluctant to accept noncancer patients because they fear that the latter will stabilize and survive beyond their PCU's 3-month “limit.” Such patients may be difficult to transfer because they are often unwilling to be placed in a long-term care facility and too ill to discharge home. As a result, these patients occupy a “short-term” PCU bed for longer than anticipated, denying it to other patients. Our study should reassure such directors and referring physicians, because it shows that noncancer patients have a shorter LOS and are less frequently discharged alive than cancer patients. Directors should also be reassured by the multivariable analysis and the finding that patients who are discharged alive generally leave long before the 3-month “limit,” suggesting that the impact of noncancer patients in general and oversurvival in particular on bed utilization is small. However, our population was highly selected and included a relatively small subset of noncancer patients with a PPS score >30. If PCUs start admitting a larger number of noncancer patients with a PPS of 30+, then their effect on LOS may rise.

Our study showed that the PPS was a better correlate of survival than the clinician's estimated prognosis. This is not surprising,¹⁴ particularly given previous studies of the accuracy of PPS for predicting survival^4,5,15 and the fact that the documented prognosis may be strongly influenced by nonclinical, pragmatic considerations (e.g., PCU admission criteria).^16,17 However, we did not expect to find that PPS was more strongly correlated with survival in noncancer patients than cancer patients. This has been noted previously,¹⁰ and may be due to the fact that noncancer patients had lower PPS scores overall; lower scores are more closely associated with survival than higher scores.^4,5,15 Our findings further support the idea that PPS is a more rational means than prognosis for determining PCU eligibility.

Of course, there are more accurate prognostic tools available for clinicians, including the Palliative Prognostic Score and the Palliative Prognostic Index.^18,19 These tools may be particularly accurate when combined with other clinical information.²⁰ However, even the most precise prognosis would have a limited role in determining when a patient needs admission to a PCU. This decision is complicated,²¹ and would ideally be based on symptom and support needs rather than life expectancy alone. PCU directors must therefore find a balance between maximizing bed availability and meeting the needs of the palliative patient population.

We found that patients referred from other medical facilities waited almost 5 days longer than patients from an inpatient ward in the same facility to be admitted to a PCU, and that this difference remained even in a multivariable model. Ideally, PCUs should not prioritize admissions on the basis of location, even though many hospitals face pressure to improve patient “flow” and reduce length of admission on their acute wards. All stakeholders must recognize that this is unfair to dying patients. Many previous authors have suggested that patients face inequitable access to palliative care resources based on age, location, socioeconomic status, gender, ethnicity, education, and diagnosis.^7,22–24

The limitations of our study include the fact that the study population was selected both by the referring physician and the accepting PCU, which limits our ability to make broad conclusions about the effects of diagnosis, PPS, and prognosis on survival. We do not have data for dying patients who were rejected by the PCUs, or who were never referred for PCU admission. In addition, the prognostic and PPS data were provided by the referring physician, who was often not a specialist palliative care physician. The data were generally not independently verified prior to admission, so we cannot confirm the accuracy of the information. However, this is often the case when patients are referred to PCUs, so this “limitation” may actually improve the generalizability of our findings. In Canada, we have universal health insurance that covers almost all PCU-related costs for Canadians. Therefore, we could not determine the relationship between health insurance status and LOS and waiting times.

Conclusion

We found that noncancer patients had a shorter LOS and a similar likelihood of being discharged alive to cancer patients, and that a cancer diagnosis was not associated with LOS on multivariable analysis. Referring physicians and PCUs should not reject or restrict end-stage noncancer referrals out of concern that these patients are “oversurviving” and having a detrimental impact on PCU bed utilization.

Footnotes

Acknowledgments

The authors thank Anna Marie Danielson for her help with statistical analysis of the data, as well as Dr. Lawrence Librach, Karina Wulf, Catherine Lacombe, and Ron Lirette for their assistance with this study.

Author Disclosure Statement

No competing financial interests exist.

References

Rickerson

, Harrold

, Kapo

, Carroll

, Casarett

. Timing of hospice referral and families' perceptions of services: Are earlier hospice referrals better? J Am Geriatr Soc, 2005; 53:819–823.

Christakis

, Lamont

. Extent and determinants of error in doctors' prognoses in terminally ill patients: Prospective cohort study. BMJ, 2000; 320:469–472.

Glare

, Virik

, Jones

, Hudson

, Eychmuller

, Simes

, Christakis

. A systematic review of physicians' survival predictions in terminally ill cancer patients. BMJ, 2003; 327:195–198.

Glare

, Sinclair

, Downing

, Stone

, Maltoni

, Vigano

. Predicting survival in patients with advanced disease. Eur J Cancer May, 2008; 44:1146–1156.

Lau

, Maida

, Downing

, Lesperance

, Karlson

, Kuziemsky

. Use of the Palliative Performance Scale (PPS) for end-of-life prognostication in a palliative medicine consultation service. J Pain Sympt Manage, 2009; 37:965–972.

Fox

, Landrum-McNiff

, Zhong

, Dawson

, Wu

, Lynn

. Evaluation of prognostic criteria for determining hospice eligibility in patients with advanced lung, heart, or liver disease. SUPPORT Investigators. Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments. JAMA, 1999; 282:1638–1645.

Walshe

, Todd

, Caress

, Chew-Graham

. Patterns of access to community palliative care services: A literature review. J Pain Sympt Manage, 2009; 37:884–912.

Head

, Ritchie

, Smoot

. Prognostication in hospice care: Can the palliative performance scale help? J Palliat Med, 8:492–502.

Vigano

, Dorgan

, Buckingham

, Bruera

, Suarez-Almazor

. Survival prediction in terminal cancer patients: A systematic review of the medical literature. Palliat Med, 000; 14:363–374.

10.

Harrold

, Rickerson

, Carroll

, McGrath

, Morales

, Kapo

, Casarett

. Is the palliative performance scale a useful predictor of mortality in a heterogeneous hospice population? J Palliat Med, 2005; 8:503–509.

11.

Olajide

, Hanson

, Usher

, Qaqish

, Schwartz

, Bernard

. Validation of the palliative performance scale in the acute tertiary care hospital setting. J Palliat Med, 2007; 10:111–117.

12.

Downing

, Lau

, Lesperance

, Karlson

, Shaw

, Kuziemsky

, Bernard

, Hanson

, Olajide

, Head

, Ritchie

, Harrold

, Casarett

. Meta-analysis of survival prediction with Palliative Performance Scale. J Palliative Care, 2007; 23:245–252discussion 252–244.

13.

Selby

, Chakraborty

, Lilien

, Stacey

, Zhang

, Myers

. Clinician accuracy when estimating survival duration: The role of the patient's performance status and time-based prognostic categories. J Pain Sympt Manage, 2011; 42:578–588.

14.

Dawes

, Faust

, Meehl

. Clinical versus actuarial judgment. Science, 1989; 243:1668–1674.

15.

Chow

, Harth

, Hruby

, Finkelstein

, Wu

, Danjoux

. How accurate are physicians' clinical predictions of survival and the available prognostic tools in estimating survival times in terminally ill cancer patients? A systematic review. Clin Oncol, 2001; 13:209–218.

16.

Schonwetter

, Teasdale

, Storey

, Luchi

. Estimation of survival time in terminal cancer patients: An impedance to hospice admissions? Hospice J, 1990; 6:65–79.

17.

Kinzbrunner

. Ethical dilemmas in hospice and palliative care. Support Care Cancer, 1995; 3:28–36.

18.

Morita

, Tsunoda

, Inoue

, Chihara

. Improved accuracy of physicians' survival prediction for terminally ill cancer patients using the Palliative Prognostic Index. Palliative Med, 2001; 15:419–424.

19.

Maltoni

, Caraceni

, Brunelli

, Broeckaert

, Christakis

, Eychmueller

, Glare

, Nabal

, Viganò

, Larkin

, De Conno

, Hanks

, Kaasa

. Steering Committee of the European Association for Palliative Care: Prognostic factors in advanced cancer patients: Evidence-based clinical recommendations—a study by the Steering Committee of the European Association for Palliative Care. J Clin Oncol, 2005; 23:6240–6248.

20.

Tarumi

, Watanabe

, Lau

, Yang

, Quan

, Sawchuk

, deMoissac

, Wolch

, Thai

. Evaluation of the Palliative Prognostic Score (PaP) and routinely collected clinical data in prognostication of survival for patients referred to a palliative care consultation service in an acute care hospital. J Pain Sympt Manage, 2011; 42:419–431.

21.

Kapo

, Harrold

, Carroll

, Rickerson

, Casarett

. Are we referring patients to hospice too late? Patients' and families' opinions. J Palliat Med, 2005; 8:521–527.

22.

Beccaro

, Costantini

, Merlo

. Inequity in the provision of and access to palliative care for cancer patients. Results from the Italian survey of the dying of cancer (ISDOC) BMC Public Health, 2007; 7:66.

23.

Burge

, Lawson

, Johnston

, Grunfeld

. A population-based study of age inequalities in access to palliative care among cancer patients. Med Care, 2008; 46:1203–1211.

24.

Maddison

, Asada

, Urquhart

. Inequity in access to cancer care: A review of the Canadian literature. Cancer Causes Control, 2011; 22:359–366.