Racial and Ethnic Differences in Hospice Enrollment Among Children With Cancer

Pediatr Blood Cancer 2013;60:1662 1666 Racial and Ethnic Differences in Hospice Enrollment Among Children With Cancer Rachel Thienprayoon, MD, 1,2 * Simon Craddock Lee, PhD, MPH, 3,4 David Leonard, PhD, 3 and Naomi Winick, MD 1,2,4 Background. Hospice is an important provider of end of life care. Adult minorities are less likely to enroll on hospice; little is known regarding the prevalence of pediatric hospice use or the characteristics of its users. Our primary objective was to determine whether race/ethnicity was associated with hospice enrollment in children with cancer. We hypothesized that minority (Latino) race/ethnicity is negatively associated with hospice enrollment in children with cancer. Procedure. In this single-center retrospective cohort study, inclusion criteria were patients who died of cancer or stem cell transplant between January 1, 2006 and December 31, 2010. The primary outcome variable was hospice enrollment and primary predictor was race/ethnicity. Results. Of the 202 patients initially identified, 114 met inclusion criteria, of whom 95 were enrolled on hospice. Patient race/ethnicity was significantly associated with hospice enrollment (P ¼ 0.02), the association remained significant (P ¼ 0.024) after controlling for payor status (P ¼ 0.995), patient diagnosis (P ¼ 0.007), or religion (P ¼ 0.921). Latinos enrolled on hospice significantly more often than patients of other races. Despite initial enrollment on hospice however, 34% of Latinos and 50% of non-latinos had withdrawn from hospice at the time of death (P ¼ 0.10). Race/ethnicity was not significantly associated with dying on hospice. Conclusions. These results indicate that race/ethnicity and diagnosis are likely to play a role in hospice enrollment during childhood. A striking number of patients of all race/ethnicities left hospice prior to death. More studies describing the impact of culture on end of life decision-making and the hospice experience in childhood are warranted. Pediatr Blood Cancer 2013;60:1662 1666. # 2013 Wiley Periodicals, Inc. Key words: pediatrics; oncology; hospice; palliative care; ethnic disparities INTRODUCTION Fifty thousand children die annually in the United States, many of them following a progressive or chronic illness. Cancer is the leading cause of death due to disease in childhood. Hospice services, designed to meet the needs of families and individuals with terminal illnesses, are an important component of end of life care. Studies examining rates of hospice enrollment in adults have consistently revealed differences between minorities and Caucasians, with minorities less likely to enroll on hospice [1 6]. In adult Latinos, the largest minority group in the United States, lower rates of hospice enrollment have been noted compared to non-latino Caucasians [7]. Whether similar disparities exist in the use of hospice for children has not been reported. Therefore, the primary aim of this study was to address, in a large cohort of deceased childhood cancer patients, whether race/ethnicity was associated with hospice enrollment. Specifically, we sought to learn whether Latinos were enrolled on hospice less often than non-latino whites. Other goals were to determine demographic factors associated with hospice enrollment and with death during hospice care. The primary hypothesis was that minority (Latino) race/ethnicity is negatively associated with hospice enrollment in children with cancer. METHODS Approval was obtained from the University of Texas Southwestern Medical Center Institutional Review Board prior to data collection. For this retrospective cohort study, the cancer registry and neuro-oncology databases of the Center for Cancer and Blood Disorders (CCBD) were queried to identify oncology patients who died between January 1, 2006 and December 31, 2010. Databases of Hospice of East Texas (HOET) and Community Hospice of Texas (CHOT), the two primary hospice organizations to which our program refers patients, were also queried for all children who died during the same time period and cross-referenced with referring oncologists in our program to identify any additional patients. CHOT serves patients of all ages in the Dallas-Fort Worth area; HOET is located in Tyler, Texas and serves patients of all ages who C 2013 Wiley Periodicals, Inc. DOI 10.1002/pbc.24590 Published online 3 June 2013 in Wiley Online Library (wileyonlinelibrary.com). reside in East Texas. The vast majority of children in our catchment area are served by one of these two hospice organizations. The CCBD follows approximately 300 children newly diagnosed with cancer per year and is the primary consultative and patient care resource in North Texas for pediatric patients with blood disorders and cancer. Approximately 45% of CCBD patients are Latino, about half of whom are primarily Spanish-speakers; 35% are Caucasian; and 20% are African-American or other races/ ethnicities. Inclusion criteria and exclusion criteria were as follows: Inclusion criteria: patients followed in our program for cancer or bone marrow transplant who died between January 1, 2006 and December 31, 2010. Exclusion criteria: (1) patients greater than 18 years at the time of cancer diagnosis, (2) death unrelated to cancer or complications of cancer treatment (i.e., death due to trauma in patients previously treated for cancer), (3) death in the hospital due to an acute process, in a child for whom a hospice referral would never have been appropriate, for example, a child with leukemia in first complete remission whose death was secondary to sepsis (4) enrollment on hospice prior to cancer diagnosis, (5) transfer of care to another non-hospice facility, (6) patients without a diagnosis of cancer, and (7) if oncology or bone 1 Center for Cancer and Blood Disorders, Division of Hematology- Oncology, Department of Pediatrics, Dallas, Texas; 2 Department of Clinical Sciences, Dallas, Texas; 3 Harold C. Simmons Cancer Center, University of Texas at Southwestern Medical Center, Dallas, Texas; 4 Children s Medical Center, Dallas, Texas Grant sponsor: St. Baldrick s Foundation Fellows Grant Conflict of interest: Nothing to declare. Prior presentations: Thienprayoon R, Leonard D, Lee SC, Winick NJ. Association of race and ethnicity with hospice use in pediatric oncology patients. J Clin Oncol 30, 2012 (suppl; abstr e19536). Correspondence to: Rachel Thienprayoon, MD, University of Texas at Southwestern Medical Center, 5523 Harry Hines Boulevard, Dallas, TX 75390-9063 E-mail: rachel.thienprayoon@utsouthwestern.edu Received 31 January 2013; Accepted 10 April 2013

Hospice Racial/Ethnic Disparities in Children 1663 marrow transplant was not the primary service, for example, a solid organ transplant patient with post-transplant lymphoproliferative disorder followed primarily by cardiology. Hospice enrollment was the primary outcome variable and death on hospice was the secondary outcome variable. Patient race/ethnicity was the primary predictor variable and secondary predictors included patient religion, age at death, and diagnosis. The primary hypothesis was that minority (Latino) race/ethnicity is negatively associated with hospice enrollment in children with cancer. Demographic data were collected from institutional medical records and hospice databases. Location of death and DNR status were collected in the same manner. Race/Ethnicity was coded as Latino ¼ Hispanic, Non-Latino-Caucasian ¼ Caucasian, and Other ¼ African-American, Asian/Pacific Islander, or Other. These data were available administratively for all patients. Hospice enrollment was defined as the patient having enrolled in hospice with either CHOT or HOET, and having received inpatient and/or outpatient services from these two organizations. If a hospice consult was documented in the institutional medical record but neither hospice organization enrolled the patient, the patient was not defined as having enrolled on hospice. Dates of hospice enrollment and discharge and reasons for discharge were collected directly from the hospices. Some patients were enrolled and discharged several times; if patient was discharged from his or her final hospice encounter for any reason other than death then he was defined as not dying on hospice. No eligible patients received hospice services from other hospice organizations. Two-way associations were tested using Fisher s exact statistics, and two-way associations controlling for additional factors were tested using multiple logistic regression with Firth s correction for small sample bias [8,9]. All tests were two-sided; P-values were reported without adjustment for multiple testing. SAS/STAT statistical software, version 9.3, was used for all analyses. RESULTS Two hundred two cancer or bone marrow transplant patients who died between January 1, 2006 and December 31, 2010 were initially identified. One hundred fourteen were considered evaluable and 88 patients were excluded. Reasons for exclusion are summarized in Table I. Among those excluded, the most common causes of death due to an acute process were sepsis (N ¼ 15), treatment complications (N ¼ 15), and complications related to primary disease (N ¼ 10). Of the 18 patients who transferred care to another facility, 11 (61%) were non-latino Caucasian, 4 (22%) were Latino and 3 (17%) were other race/ ethnicities. All seven patients lost to follow up were Caucasian. TABLE I. Reasons for Exclusion of Identified Patients Reason for exclusion Death due to an acute process 46 Transfer of care to another facility 18 Oncology/BMT not primary service 9 Lost to follow up but recorded as deceased 7 Enrolled on hospice at diagnosis 3 Death off treatment unrelated to cancer 2 Non-oncologic diagnosis 2 Age > 18 at diagnosis 1 Total 88 Abbreviations: BMT, bone marrow transplant. N Patient race/ethnicity was significantly associated with hospice enrollment (P ¼ 0.02, Table II), and the association remained significant (P ¼ 0.024) after controlling for payor status (P ¼ 0.995), patient diagnosis (P ¼ 0.007), or religion (P ¼ 0.921). Latino patients were significantly more likely to enroll on hospice versus non-latino Whites (OR 5.961, 95% CI [1.451, 31.037], Table III). Patient diagnosis was significantly associated with hospice enrollment (P ¼ 0.02, Table II); patients with leukemia, lymphoma, or stem cell transplant were significantly less likely to enroll on hospice versus patients with brain and solid tumors (OR 0.166, 95% CI [0.039,0.598], Table III). Of those patients who enrolled on hospice, only 50% of Caucasians and 66% of Latinos remained enrolled at the time of death (P ¼ 0.10). DISCUSSION This study was designed to determine whether for children with cancer, race/ethnicity was associated with hospice enrollment. We found that in children under our care, 83% of evaluable subjects enrolled on hospice and race/ethnicity was significantly associated with hospice enrollment. Latino patients enrolled on hospice significantly more often than non-latinos. Patient diagnosis was also associated with hospice enrollment; those with brain and other solid tumors enrolled significantly more often than patients with leukemia, lymphoma or following stem cell transplant. Patient age and religion were not significantly associated with hospice enrollment, and race/ethnicity was not significantly associated with dying while on hospice. These results differ significantly from patterns of hospice use cited in both the adult and pediatric literature. Studies examining the correlation between race/ethnicity and hospice enrollment have consistently found that minorities are less likely to enroll on hospice [1 6]. A review of Florida Medicaid records between 2003 and 2006 found that the probability of incurring any hospice expenditure was 42% less for Hispanic than white non-hispanic children [10]. However only 17% of the subjects were Latino [10], as opposed to our sample in which 44% were Latino. That Florida study also included children who died of all causes; nearly half died of external causes including accidents or trauma, few of whom were enrolled on hospice [10] thus possibly impacting their findings. Using administrative data, we were unable to directly measure acculturation retrospectively; however, limited English proficiency may be used as a surrogate for this [11,12]. This variable was also significantly associated with hospice enrollment, supporting our belief that in this case, Caucasians and Latinos represent two distinct racial/ethnic groups. To our knowledge, no other studies examining the association between race/ethnicity and hospice in childhood have been published. A racial disparity in hospice enrollment of children with cancer may be explained by differing expectations between Caucasians and Latinos who enroll on hospice. Caregivers of adult Latino patients enrolled on hospice have reported no knowledge of hospice, or more misconceptions about hospice, before the patient s illness [13]. In contrast, non-latino Caucasian caregivers have reported prior experience with and knowledge of hospice, as well as higher expectations for hospice, prior to the patient s illness [13]. Bereaved Mexican-American parents whose child died in the US have described difficulty in negotiating the medical system in the US, including language-related barriers and perceived discrimination by the healthcare team [14]. Latinos have also expressed a

1664 Thienprayoon et al. TABLE II. Demographic Characteristics of Eligible Patients and Hospice Enrollees Demographic or clinical characteristic All eligible, N (%) Enrolled on hospice, N (%) Not enrolled on hospice, N (%) P-value Gender Female 55 (48) 47 (49) 8 (42) 0.62 Race/ethnicity Hispanic 50 (44) 47 (49) 3 (16) 0.02 Non-Hispanic white 42 (37) 32 (34) 10 (53) Other 22 (19) 16 (17) 6 (32) Religion Catholic 28 (24) 26 (27) 2 (11) 0.10 Christian, NDS 23 (20) 17 (18) 6 (32) Protestant 10 (9) 8 (8) 2 (11) Other 11 (10) 7 (7) 4 (21) NR/unknown 42 (37) 37 (39) 5 (26) Payor status Medicaid 62 (54) 51 (54) 11 (58) 0.76 Private insurance/military 44 (39) 36 (38) 8 (42) Private pay/cshcn 7 (6) 7 (7) 0 (0) Other 1 (1) 1 (1) 0 (0) Diagnosis Brain tumors 41 (36) 37 (39) 4 (21) 0.02 Other solid tumors 38 (33) 34 (36) 4 (21) Non-AML and lymphoma 23 (20) 17 (18) 6 (32) AML 9 (8) 4 (4) 5 (26) Other/BMT 3 (3) 3 (3) 0 (0) Primary language English 86 (75) 68 (72) 18 (95) 0.07 Spanish 27 (24) 26 (27) 1 (5) Other 1 (1) 1 (1) 0 (0) Limited English proficiency No 83 (73) 65 (70) 18 (95) 0.02 Yes 31 (27) 30 (30) 1 (5) Location of death ICU 14 (12) 5 (5) 9 (47) <0.001 Inpatient floor 23 (20) 13 (15) 9 (47) Other 6 (5) 5 (5) 1 (5) Home 58 (51) 58 (61) 0 (0) NR/unknown 13 (11) 13 (14) 0 (0) DNR status Yes 55 (48) 43 (45) 12 (63) <0.001 No 43 (38) 41 (43) 2 (11) Withdrawal of care 7 (6) 2 (2) 5 (26) NR/unknown 9 (8) 9 (9) 0 (0) Abbreviations: NDS, no denomination specified; CSHCN, children with special healthcare needs; AML, acute myelocytic leukemia; BMT, bone marrow transplant; ICU, intensive care unit; NR, not recorded; DNR, do not resuscitate. TABLE III. Odds Ratios and 95% Confidence Intervals for the Association of Hospice Enrollment With Patient Race/Ethnicity, Diagnosis, Religion, and Payor Status Factor OR 95% CI Race/ethnicity Latinos versus non-latino whites 5.961 1.451, 31.037 Other minorities versus non-latino whites 0.652 0.184, 2.314 Diagnosis Leukemia, lymphoma and stem cell transplant versus brain and solid tumors 0.166 0.039, 0.598 Religion Catholic versus protestant, Christian, NDS 1.254 0.389, 3.977 Payor status Medicaid, private pay, CSHCN, or other versus private pay/military insurance 1.004 0.332, 2.966 Abbreviations: NDS, no denomination supplied; CSHCN, children with special healthcare needs.

Hospice Racial/Ethnic Disparities in Children 1665 preference for a more pro-active role on the physician s part in addressing end of life issues [15]. Thus, a possible explanation for the disparity in hospice enrollment reported here is that the Latino families, due to a preference for paternalism by the medical provider, may have been more likely to follow the advice of the physician to enroll on hospice. Fourteen of the 18 patients who transferred care to another facility were non-latino, raising the possibility that non-latino parents may have been more likely to voice disagreement and pursue further active or palliative care, in our program or elsewhere. It is also possible that the two hospice groups providing the majority of the pediatric palliative care for our institution have enhanced their profile with Latino families, but we were unable to identify an overt outreach program, the hospice intake nurses do not speak Spanish and the initial consults are completed with the aid of a hospital-based interpreter. The role of race/ethnicity and culture in the pediatric provider-parent dynamic during end of life discussions is not well described, especially among Latino community subgroups; this is an area that merits investigation. To our knowledge, it has also not been reported that children with brain and solid tumors are more likely to enroll on hospice. It is possible that the poor prognosis associated with some CNS malignancies (e.g., diffuse intrinsic pontine gliomas, rhabdoid tumors) and solid tumors (e.g., sarcomas metastatic at diagnosis) may impact the quality and timing of the discussions about hospice care. In addition, neuro-oncology patients were largely cared for by a neuro-oncology medical team that was constant during the period of study. A very high percentage of eligible patients of all race/ethnicities were enrolled on hospice, which may reflect the inclusion/exclusion criteria applied to the study population or a uniform and consistent approach to end of life care by childhood cancer providers at our institution. It should be noted that in our institution, the responsibility of end of life discussions and hospice referral in children with cancer continues to lie with the primary oncologist. A palliative care service became available in our institution in 2006 and the team was consulted on 19 of the patients in this cohort. However for all patients in this cohort who were enrolled on hospice, the referral to hospice was made by the primary oncologist. A striking number of patients who were enrolled on hospice did not die on hospice. We postulated above that a possible explanation for the disparity in hospice enrollment reported here is that the Latino families may have been more likely to follow the advice of the physician to enroll on hospice. Should there have been a discrepancy in the degree to which Latino families accepted medical opinion, it is possible that Latino parents enrolled their children on hospice without an adequate understanding of the services offered/not offered and then ultimately returned to the hospital. This does not explain, however, the proportion of Caucasians who also left hospice. The reasons for returning to the hospital are likely complex and heterogeneous, underscoring the need for increased research regarding the childhood hospice experience, and increased education regarding hospice services prior to hospice enrollment for both parents and oncology providers. It should also be noted that children who die in the hospital under the care of the oncology team may receive end of life care consistent with the principles offered by hospice. Future studies may address the relationship between race/ethnicity and preferences for aggressive interventions in end of life care in children, but this study was not designed specifically to do so. There are several limitations to this study. This is a singleinstitution analysis of cancer patients cared for by a relatively homogenous group of providers with respect to race/ethnicity (90% Caucasian). As little is known about the role that provider-patient race/ethnic concordance in the pediatric setting, it is difficult to speculate on how this may have impacted our findings; further studies are warranted to better understand this dynamic [16]. Due to the nature of the cancer experience, it may be difficult to generalize these results to patients without cancer in our own institution or elsewhere. Further, demographic information was obtained from administrative data, and it was not possible to evaluate their accuracy. Religion was not consistently recorded in the medical record; nearly 40% of patients did not have information regarding religious affiliation, limiting our ability to draw firm conclusions about this potentially important cultural factor. Likewise, specific location of death and DNR status were inconsistently recorded, particularly for those patients who died on hospice, limiting our ability to explore these factors. Strengths of this study are the high volume of patients treated in our program, including Latino minorities and, thus, the high number of patients enrolled on hospice over the 5-year period studied. To our knowledge, this is the first report to describe a relationship between hospice enrollment and patient race/ethnicity in children, and one of the first to describe a cohort of children enrolled on hospice. Our success in cross-referencing patient records between the two local hospice databases is a unique feature of our data collection, those data being distinct from that obtained from our hospital and clinic medical record. Since few hospices in our region accept children and no eligible patient enrolled with another hospice, we feel confident that we captured data for our entire cohort. We have not found evidence of other studies involving children which have combined inpatient and hospice data in this way and believe we have demonstrated a highly feasible model to study patterns of hospice use in these patients. CONCLUSIONS We found that race/ethnicity is associated with hospice enrollment in childhood cancer patients, with Latinos more likely to enroll on hospice than non-latinos. This finding was unexpected, as adult patient studies generally report lower use of hospice among minority families. Multi-institution studies are warranted to confirm these findings, as well as to explore cultural factors that may impact end of life decision making for children. Our findings suggest the importance of exploring such potential differences within and between minority groups, particularly if acculturation plays a role. That so many children enrolled on hospice yet left prior to death was unexpected and concerning from a quality of care standpoint. It is critical for providers to fully investigate explanations for this finding, in order to ultimately provide the best end of life care for each patient and his or her family. Future studies are warranted to identify and elucidate those cultural values that may be important to families when discussing hospice enrollment and to identify any targets amenable to both patient and provider education efforts. ACKNOWLEDGMENTS The authors wish to thank the Hospice of East Texas (Tyler, Texas) and Community Hospice of Texas (Dallas, Texas) for their

1666 Thienprayoon et al. assistance with acquisition of data from their respective organizations. Dr. Thienprayoon received 100% salary support for this project from the St. Baldrick s Foundation Fellows Grant. REFERENCES 1. Cohen LL. Racial/ethnic disparities in hospice care: A systematic review. J Palliat Med 2008;11:763 768. 2. Hardy D, Chan W, Liu CC, et al. Racial disparities in the use of hospice services according to geographic residence and socioeconomic status in an elderly cohort with nonsmall cell lung cancer. Cancer 2011;117:1506 1515. 3. Enguidanos S, Yip J, Wilber K. Ethnic variation in site of death of older adults dually eligible for Medicaid and Medicare. J Am Geriatr Soc 2005;53:1411 1416. 4. Ngo-Metzger Q, Phillips RS, McCarthy EP. Ethnic disparities in hospice use among Asian-American and Pacific Islander patients dying with cancer. J Am Geriatr Soc 2008;56:139 144. 5. Smith AK, Earle CC, McCarthy EP. Racial and ethnic differences in end-of-life care in fee-for-service Medicare beneficiaries with advanced cancer. J Am Geriatr Soc 2009;57:153 158. 6. Givens JL, Tjia J, Zhou C, et al. Racial and ethnic differences in hospice use among patients with heart failure. Arch Intern Med 2010;170:427 432. 7. Colon M, Lyke J. Comparison of hospice use and demographics among European Americans, African Americans, and Latinos. Am J Hosp Palliat Care 2003;20:182 190. 8. Firth D. Bias reduction of maximum-likelihood-estimates. Biometrika 1993;80:27 38. 9. Heinze G, Schemper M. A solution to the problem of separation in logistic regression. Stat Med 2002;21:2409 2419. 10. Knapp CA, Shenkman EA, Marcu MI, et al. Pediatric palliative care: Describing hospice users and identifying factors that affect hospice expenditures. J Palliat Med 2009;12:223 229. 11. Colon M. Acculturation and attitudes of Latinos toward hospice. J Soc Work End Life Palliat Care 2012;8:229 248. 12. Selsky C, Kreling B, Luta G, et al. Hospice knowledge and intentions among Latinos using safety-net clinics. J Palliat Med 2012;15:984 990. 13. Kreling B, Selsky C, Perret-Gentil M, et al. The worst thing about hospice is that they talk about death : Contrasting hospice decisions and experience among immigrant Central and South American Latinos with US-born White, non-latino cancer caregivers. Palliat Med 2010;24:427 434. 14. Contro N, Davies B, Larson J, et al. Away from home: Experiences of Mexican American families in pediatric palliative care. J Soc Work End Life Palliat Care 2010;6:185 204. 15. Braun UK, Beyth RJ, Ford ME, et al. Voices of African American, Caucasian, and Hispanic surrogates on the burdens of end-of-life decision making. J Gen Intern Med 2008;23:267 274. 16. Brown TN, Ueno K, Smith CL, et al. Communication patterns in medical encounters for the treatment of child psychosocial problems: Does pediatrician-parent concordance matter? Health Commun 2007; 21:247 256.