Association between palliative care and end-of-Life care for patients with hematological malignancies: A population-based study.

Introduction

The incidence of hematological malignancies (HMs) in Taiwan is markedly lower than that in Western countries, but it had a drastically increasing trend in recent decades. [ 1 ] In Taiwan, the percentage of HMs among all cancer deaths was 4.53% and HMs were the seventh most common cause of cancer-related deaths in 2012. [ 2 ] Despite advancements in diagnosis and treatment, mortality due to HMs is not decreased. A previous study reported that the mean number of symptoms and level of distress were comparable to those patients with metastatic non-hematological malignancy. [ 3 ] In addition, patients with HMs were often treated with intensive antineoplastic regimens until the last days of life. [ 4 ] Under some medical conditions such as infections, cytopenias, and coagulopathies, patients with HMs needed frequent hospitalizations, invasive investigations, monitoring and therapies. [ 5 , 6 ] A cohort study reported that patients with HMs received more inappropriate care during end-of-life (EOL) care. [ 7 ] Continued efforts are needed to improve the provision of quality EOL care for patients with HMs.

Palliative care is an interdisciplinary approach to symptom management, psychological support, and treatment decision-making for patients with serious illnesses and their family members. More evidence highlights that patients with cancer could benefit greatly from palliative care, which can reduce symptom burden, [ 8 ] improve quality of life and mood, [ 9 , 10 ] increase the likelihood of survival, [ 10 , 11 ] and improve outcomes for caregivers as well as for patients. [ 12 , 13 ] In addition, the palliative care services may assist hematologists in the management of their patients’ suffering and quality of life during the timing of increased symptom burden for patients with HMs. [ 3 ] In Taiwan, there are no residential facilities to provide hospice care. “Palliative care” encompasses much more than just EOL or hospice care. [ 13 ] In the current study, palliative care included hospital-based inpatient care, outpatient services, and home care.

Six QIs of EOL cancer care have been developed and are outlined as follows: undergoing chemotherapy during the last 2 weeks of life, having more than 1 emergency room (ER) visits in the final month of life, being admitted to a hospital at least twice in the final month of life, receiving intensive care unit (ICU) care in the final month of life, receiving cardiopulmonary resuscitation (CPR) in the final month of life, and dying in a hospital. [ 14 , 15 ] These QIs have been adopted in the United States, [ 16 ] Canada, [ 17 ] and Taiwan [ 18 ] and are considered as aggressive EOL cancer care. All indicators are considered to indicate poor quality care. More aggressive EOL care is considered inappropriate for the terminally ill patients. [ 17 ] Inappropriate EOL care was examined by a composite score adapted from Tang et al. [ 19 ] Therefore, measuring the score is crucial for evaluating the quality of palliative care programs. In this study, we used Taiwan's National Health Insurance Research Database (NHIRD) to evaluate the impact of palliative care on QIs of EOL cancer care and health care costs for patients with HMs in the final month of life.

Methods

Data source

Taiwan's National Health Insurance (NHI) program was implemented in March 1995; it is a single-payer program that covered as many as 99.9% of Taiwan's residents in 2012. [ 20 ] Taiwan's NHI has the unique characteristics of universal insurance coverage, comprehensive services (including medications, home care, even Chinese herbal medicine therapy) provided, and a single-payer system with the government as sole insurer. Patients have free access to any health care system and provider they choose. Health care systems are reimbursed for services provided, and copayment is waived for patients with examined catastrophic illness certificate (CIC), including malignancy. In the present study, patient data were linked to Taiwan's 2000 Longitudinal Health Insurance Database (LHID2000), a subset of the NHIRD. The LHID2000 contains all original claims data for 1 million individuals randomly sampled from the 2000 NHIRD Registry. All patients who had a diagnosis of hematological malignancies with matching CIC between January 1, 2000 and December 31, 2011 were included in our study. We followed patients with HMs until December 2012 by using the LHID2000. Claims data included medical records (inpatient care, outpatient records, and home care) of patients who had and had not received palliative care. Patients under 20 years old and those who had died within 1 month after HM diagnosis were excluded. The International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes (200–208) and A codes (A140, A141, and A149) were used to define HMs. To increase the validity of diagnoses of diabetes or hypertension, we defined patients with these conditions as those with 3 reported diagnoses of diabetes or three instances of hypertension in their medical claims data based on the ICD-9-CM or A codes for these disease entities. [ 21 , 22 ]

Variables

Patient characteristics included age, sex, age at death, median (range) survival in years after HM diagnosis, chemotherapy (the chemotherapy was assumed whenever there was an order for a reimbursement code of oral or intravenous chemotherapy during the period of study), geographic location, [ 23 ] level of urbanization, and whether diagnosis was made at a teaching hospital (Table 1 ). Comorbid conditions listed in the Charlson comorbidity index (CCI) and common comorbidities (e.g., diabetes, hypertension, stroke, and chronic kidney disease) were identified based on ICD-9-CM codes. [ 24 ]

Variable definitions

Inpatient palliative, home palliative, palliative, and nonpalliative care groups: We searched the claim data for the reimbursement codes of inpatient palliative care and home palliative care. Among the claim data, patients with the codes for inpatient palliative care and without the codes for home palliative care were classified as inpatient palliative (i-Pal) group. If patients were with the codes for home palliative care, they were classified as home palliative (h-Pal) group. Accordingly, these inpatient palliative units also served as the back-up system for patients receiving home palliative care when they experienced exacerbating symptoms that required further readmission to the palliative unit or whose families needed respite care from caregiving. Patients of i-Pal group and h-Pal group were combined into the Pal groups. Patients with HMs who had not received palliative care were categorized as non-palliative group (non-Pal group).

Charlson comorbidity index (CCI)

We calculated CCI scores by examining ICD-9-CM-based diagnoses and procedure codes recorded using the Deyo method. We subsequently applied calculated indices to inpatient and outpatient claims reported by Klabundle et al. [ 25 , 26 ]

Health care costs

We calculated each patient's health care costs by summing the inpatient service and outpatient service costs listed in his or her claims records. We converted these costs based on the average U.S. Dollar to New Taiwan Dollar exchange rate in 2006 (US$1.00 = NT$32.53).

Socioeconomic status (SES)

According to the procedures described in a previous study, [ 27 ] the income categories are generally representative of the 5 income groups in Taiwan in 2005. [ 28 ] In this study, we classified socioeconomic status (SES) as three groups: the low SES group comprised patients earning less than US$922 (NT$30,000) per month, the moderate SES group comprised patients earning between US$922 and US$3074 (NT$30,000–100,000) per month, and the high SES group comprised patients earning more than US$3074 (NT$100,000) per month.

Aggressive EOL care and composite scores

Aggressiveness of EOL care was examined using a composite measure adapted from Earle et al. [ 14 ] The following 6 QIs of EOL cancer care in the final month of life were employed: chemotherapy within the final 2 weeks of life; more than one ER visit, more than 1 hospitalization, ICU admission, and CPR during the final month of life; and dying in hospital. Instead of using this measure to determine the occurrence for any of the above 6 indicators, we scored 1 point per indicator per person. Composite scores ranged from 0 to 6, with a higher score indicating more aggressive EOL care, as adapted from Tang et al. [ 19 ] In these recent years, aggressive EOL care was considered as inappropriate EOL cancer care.

The protocol for this study was reviewed and approved by the Research Ethics Committee of Dalin Tzu Chi Hospital, Buddhist Tzu Chi Medical Foundation, Taiwan (No. B10301001). Because the analyzed NHIRD files contained only deidentified secondary data, the review board waived the requirement for informed consent.

Statistical analysis

All statistical analyses were performed using R version 3.0.2 (R Foundation for Statistical Computing, Vienna, Austria). A two-sided P value of ≤.05 was considered statistically significant. The distributional properties of continuous variables and categorical variables were expressed as the median (range) or frequency (percentage). Survival was defined as the time from HMs diagnosis until death. The Kaplan–Meier estimator was used to measure the survival probabilities of patients after HMs diagnosis and tested using the log-rank test. [ 29 ] Normality was examined using the Shapiro–Wilk test. In the univariate analysis, Wilcoxon rank-sum test, Kruskal-Wallis rank sum test, chi-squared test, and Fisher's exact test were conducted to examine differences in the distributions of continuous variables and categorical variables between 2 or 3 groups. We assessed patients’ demographic and clinical characteristics, including age, sex, CCI score, geographic area of residence, and treatment modality (Tables 1 and 2 ).

A multivariate analysis was conducted by fitting multiple logistic regression models with the stepwise variable selection procedure to determine vital predictors of QIs during the final month of life. Generalized additive models were fitted to detect the potential nonlinear effects of continuous covariates and determine appropriate cutoff points for discretizing continuous covariates if necessary, during stepwise variable selection.

We assessed the goodness of fit of the final logistic regression model based on the estimated area under the receiver operating characteristic curve (AUC) (also called the “ c -statistic”). In practice, a c -statistic value ( c = 0–1) of ≥0.7 suggests an acceptable level of discriminatory power. Statistical tools for regression diagnostics, including multicollinearity checking, were applied to ascertain any problems associated with the regression model or data.

Results

A total of 724 patients with HMs aged ≥20 years who died in 2000–2011 were analyzed. A total of 62 patients (8.6%) had received palliative care (Pal group) and 662 had not (non- Pal group). Among the Pal group, 43 (69.4%) had received only inpatient palliative care (i-Pal group), and 19 (30.6%) received home palliative care (h-Pal group). The median duration from the day of receiving palliative care until death was 40.3 (range: 0-369) days in the Pal group, with 15 patients (24.2%) receiving palliative care for less than 7 days, 22 patients (35.5%) receiving palliative care for more than 30 days, 7 patients (11.3%) receiving palliative care for more than 3 months, and 4 patients (6.5%) receiving palliative care for more than 6 months after enrollment. The study flowchart was shown in Figure 1 .

The median age at diagnosis of the patients in the Pal group was older than that of the patients in the non-Pal group (76.1 vs 68.4 years; P < .001). Compared with the non-Pal group, the Pal group had significantly higher proportions of patients with low SES ( P = .002) and those living in urban areas ( P = .024). Trend was still similar after separating the Pal group into the i-Pal and h-Pal groups (Table 1 ). Kaplan–Meier survival analysis revealed that the median probability of survival after diagnosis did not differ between the 2 groups (0.75 and 0.67, respectively; P = .377; Fig. 2 ), even the 3 groups ( P = .585). Six QIs of EOL cancer care and health care costs in the final month of life were compared between the Pal and non-Pal groups (Table 2 ). The median composite scores were not significantly different between Pal and non-Pal groups (2 vs 2, P = .758). Compared with the non-Pal group, the Pal group had lower proportions of patients receiving chemotherapy in the final 2 weeks of life (4.8% vs 35.2%; P < .001), those requiring ICU admission (4.8% vs 29.2%; P < .001), and those requiring CPR (9.7% vs 39.4%; P < .001). However, compared with the non-Pal group, the Pal group had significantly higher proportions of patients requiring more than 1 hospitalization (32.3% vs 16.5%; P = .005) and those who died in hospital (69.4% vs 54.4%; P = .024). No difference in patients requiring more than 1 ER visit was observed (56.5% vs 49.8%; P = .354). The median health care cost per person during the final month of life between the Pal group and the non-Pal group were not significantly different (US$3096 [0-13762] vs US$3900 [0-43783], P = .315). Similar results were found after separating the Pal group into i-Pal group and h-Pal group.

The significant factors related to the aforementioned 6 QIs of EOL care were explored through multivariate logistic regression (Table 3 ). The factors listed in Tables 1 and 2 were entered into these models. The i-Pal group had lower risk of receiving chemotherapy in the last 2 weeks (OR, 0.11; 95% CI, 0.02–0.63; P = .013), admitting to ICU (OR, 0.03; 95% CI, 0.004–0.30; P = .002), receiving CPR (OR, 0.13; 95% CI, 0.04–0.40; P < .001), but had higher risk of at least 2 hospitalization (OR, 2.44; 95% CI, 1.11–5.39; P = .027), higher risk of dying in hospital (OR, 2.49; 95% CI, 1.07–5.77; P = .034). The h-Pal group had the same trends as the i-Pal group but lower trend of dying in hospital (OR, 0.79; 95% CI, 0.25–2.54; P = .698). All AUCs were also calculated.

Discussion

The novel finding in this study was that patients with HMs receiving palliative care had less inappropriate EOL cancer care in the final month of life. The contributing factors might be associated with a significant decrease in the proportions of patients receiving chemotherapy during the final 2 weeks of life, those admitted to the ICU in the final month of life, and those receiving CPR in the final month of life. In addition, 35.5% patients of the Pal group survived for more than 30 days after receiving palliative care. Another novel finding of the present study was that the median probability of survival after diagnosis was not compromised in patients who received palliative care. However, palliative care use for patients with HMs was 8.6%, and patients receiving palliative care were more likely to be hospitalized or die in hospital.

Patients with HMs had more conditions, such as antineoplastic regimens, infections, cytopenias, and coagulopathies during the end of life, and these patients need more frequent ER visits and hospitalizations. However, the palliative care teams may assist hematologists in the management of their patients’ suffering and quality of life during the timing of increased symptoms burden. [ 3 ] In Taiwan, palliative care programs include both inpatient and home care models and have been available since 1990 for patients with serious illnesses without an absolute limitation of predicted survival duration. [ 30 ] Palliative care is covered by NHI, adopted palliative chemotherapy or radiotherapy, and included inpatient and home services. Thus, patients requiring inpatient palliative care were admitted to hospitals in Taiwan. Although most EOL quality measures were met by hematological oncologists, the quality of indicators, such as at least 2 hospitalizations and dying in hospital was not decreased in this model. We suggest that the QI of hospitalization during EOL care could be modified to days of hospital stay in the final month of life. In this study, we found that days of hospital stay in the final month of life did not differ between the Pal and non-Pal groups. One of the reasons of more hospital stays might be patients with HMs and those with solid tumors had significant symptom burden at the time of referral for palliative care, and patients with HMs exhibited more substantial drowsiness and tiredness than did those with solid tumors. [ 31 ] Previous study reported that patients receiving home-based palliative care was associated with a significant reduction of dying in a hospital. [ 32 ] In this study, we further separated the palliative group into h-Pal group and i-Pal group. We found that the h-Pal group had the similar trends as i-Pal group, but had lower trend of dying in hospital after adjustments. It might be due to the small sample size for h-Pal group to reach significant. In this study, we also found the median survival for patients with HMs was 0.74 years, which was different from previous report. The 5-year survival rate was 51.1% for patients with chronic lymphocytic lymphoma during 1990 to 2004, in Taiwan. [ 33 ] On account that those patients with HMs being alive at the end of this study were excluded, the selection bias might exist.

Another issue in care of patients with HMs is how to increase the proportion of patients with HMs who receive palliative care. Although the patients with HMs in this study who received palliative care benefited from it during EOL care, we found that palliative care use was 8.6% among patients with HMs. A cohort study reported that 8.0% of the patients with HMs had received inpatient palliative care in the United States. [ 7 ] A previous study reported that 19.9% patients with lung cancer received inpatient palliative care in Taiwan. [ 34 ] An integrative systemic review study reported that palliative care for patients with HMs is often limited to the EOL phase with late referral to palliative care. [ 35 ] Previous studies have reported that HMs were associated with inappropriate cancer-directed care during EOL care and underuse of palliative care. [ 16 , 36 , 37 ] Possible reasons for lesser use of care on palliative wards are

patients maintaining strong relationships with their oncologists and not wishing their care to become fragmented,

lower severity of symptoms among patient with HMs, and

oncologists tending toward optimism in their prognostication for patients with advanced cancers. [ 38 ]

Identification of when the EOL period begins in patients with HMs is crucial for hematological oncologists. In a previous study conducted as a series of focus groups with hematological oncologists, researchers reported that the factors influencing initiation of EOL care for patients with HMs were age, comorbidities, and performance status; the researchers also found that disease-directed treatments were causing a significant decline in patient quality of life. [ 39 ] Other barriers included hematologic oncologists’ attitudes and beliefs toward EOL care and patients’ and their family members’ preferences. [ 40 ]

There are limited data regarding the quality of EOL care for patients with HMs. Previous study reported that translating evidence into action improve chronic illness care. [ 41 ] The successful approaches included provider-oriented components, such as continuing education or physician feedback, information systems changes, and patient-oriented interventions. [ 41 ] We could learn from this model of improving chronic illness care. A potential method for increasing palliative care use is the timing of integrating palliative care. In 2012, the American Society of Clinical Oncology offered a guideline update on the integration of palliative care into standard oncologic care. [ 42 ] The guideline update recommended that inpatients and outpatients with advanced cancers should receive dedicated palliative care as early as possible in the disease course alongside standard oncologic care. [ 43 ] One study reported that patients with HMs had considerable physical and psychological symptom burden and the most appropriate time to introduce palliative care might be during increased symptom burden. [ 3 ] Prospective studies evaluating earlier implementation of palliative care with standard care of HMs are warranted. Another potential method was to reinforce the criteria (the indicators of EOL cancer care) that the health care team should follow to define the final decisions to continue or discontinue treatment in the EOL cancer care. Further studies should also look into patient reported quality of life outcomes.

Limitations

HMs were defined as incurable diseases at presentation or relapsed/refractory status. [ 7 ] The information about the staging of HMs was not obtained in the claimed data, and it is a major limitation in current study. We classified HMs as leukemia, lymphoma, and multiple myeloma. The numbers of these 3 subgroups would be too small to analyze. Additionally, this was another major limitation in current study. This study had other limitations. First, our cohort being restricted to adult patients might have limited the generalizability of our findings to people younger than 20 years of age. Second, misclassification bias may have occurred because of the inaccuracy of some of the variables used, including calculations of comorbidity scores. Third, the patients included in this study were not randomized to Pal (i-Pal, h-Pal), and non-Pal groups for comparison, and it might have selection bias. Fourth, the risk factors related to each QI (e.g., clinical symptoms and signs, patients’ or family members’ preferences, physician recommendations, and do-not-resuscitate designations) were not recorded in the administrative database. Patients’ and family members’ preferences may have influenced some outcomes. Fifth, the care of HMs has been improved over time. Alongside, the claimed data which had been used in the current study might have limitations of out-of-date care. Previous studies reported that clinical trials in HMs have been growing rapidly since 2010. Sixth, we used the insurance system exit dates as our proxy for death. The proxy date might be a couple of days later than the real death date, as it sparsely occurred. In addition, patient-centered outcome measurements, such as quality of life, health care utilization and functional capacity were incorporated in a small number of trials. [ 44 ] Finally, only 35.5% of the patients survived for more than 30 days after receiving palliative care, and the inappropriate EOL care score might be overestimated and the health care costs might be underestimated.

Conclusion

Patients with HMs who receive palliative care could benefit from less inappropriate EOL cancer care in the final month of life. However, we found that palliative care was received by 8.6% of patients with HMs. The related factors of more hospitalizations and dying in hospital warrant further investigation.

Author contributions

Conceptualization: Jui-Kun Chiang, Yee-Hsin Kao.

Formal analysis: Jui-Kun Chiang, Yee-Hsin Kao.

Investigation: Jui-Kun Chiang, Yang-Cheng Lee, Yee-Hsin Kao.

Methodology: Jui-Kun Chiang, Yang-Cheng Lee, Yee-Hsin Kao.

Resources: Yee-Hsin Kao.

Software: Jui-Kun Chiang.

Validation: Jui-Kun Chiang, Yang-Cheng Lee, Yee-Hsin Kao.

Visualization: Yee-Hsin Kao.

Writing – original draft: Yee-Hsin Kao.

Writing – review & editing: Jui-Kun Chiang, Yang-Cheng Lee.

Abbreviations: AUC = area under the receiver operating characteristic curve, CCI = Charlson comorbidity index, CIC = catastrophic illness certificate, CKD = chronic kidney disease, CPR = cardiopulmonary resuscitation, DNR = do not resuscitate, EOL = end-of-life, ER = emergency room, ICU = intensive care unit, NHI = National Health Insurance, NHIRD = National Health Insurance Research Database, QI = quality indicator.

How to cite this article: Chiang JK, Lee YC, Kao YH. Association between palliative care and end-of-Life care for patients with hematological malignancies. Medicine . 2019;98:40(e17395).

This study was based in part on data from the NHIRD provided by the Bureau of NHI, Department of Health, and managed by the National Health Research Institutes. The interpretation and conclusions contained herein do not represent those of the Bureau of NHI, Department of Health, or National Health Research Institutes. JK Chiang received research grants from Dalin Tzu Chi Hospital, Buddhist Tzu Chi Medical Foundation (DTCRD 103(2)-E-03 and 105(2)-E-22).

The authors have no conflicts of interests to disclose.

References

[1] Wu SJ Huang SY Lin CT The incidence of chronic lymphocytic leukemia in Taiwan, 1986–2005: a distinct increasing trend with birth-cohort effect . Blood 2010 ; 116 : 4430 – 5 . 20713960 10.1182/blood-2010-05-285221

[2] Ministry of Health and Welfare . Available at: http://www.mohw.gov.tw/cht/DOS/Statistic.aspx?f_list_no=312&fod_list_no=2747 Accessed November 12, 2018 .

[3] Manitta V Zordan R Cole-Sinclair M The symptom burden of patients with hematological malignancy: a cross-sectional observational study . J Pain Symptom Manage 2011 ; 42 : 432 – 42 . 21477979 10.1016/j.jpainsymman.2010.12.008

[4] Hui D Elsayem A Li Z Antineoplastic therapy use in patients with advanced cancer admitted to an acute palliative care unit at a comprehensive cancer center: a simultaneous care model . Cancer 2010 ; 116 : 2036 – 43 . 20162701 10.1002/cncr.24942 PMC2854875

[5] McGrath P Holewa H Special considerations for haematology patients in relation to end-of-life care: Australian findings . Eur J Cancer Care (Engl) 2007 ; 16 : 164 – 71 . 17371426 10.1111/j.1365-2354.2006.00745.x

[6] Franchini M Frattini F Crestani S Bleeding complications in patients with hematologic malignancies . Semin Thromb Hemost 2013 ; 39 : 94 – 100 . 23247655 10.1055/s-0032-1331154

[7] Hui D Didwaniya N Vidal M Quality of end-of-life care in patients with hematologic malignancies: a retrospective cohort study . Cancer 2014 ; 120 : 1572 – 8 . 24549743 10.1002/cncr.28614 PMC4013204

[8] Jacobsen J Jackson V Dahlin C Components of early outpatient palliative care consultation in patients with metastatic nonsmall cell lung cancer . J Palliat Med 2011 ; 14 : 459 – 64 . 21417739 10.1089/jpm.2010.0382

[9] Bakitas M Lyons KD Hegel MT Effects of a palliative care intervention on clinical outcomes in patients with advanced cancer: the project ENABLE II randomized controlled trial . JAMA 2009 ; 302 : 741 – 9 . 19690306 10.1001/jama.2009.1198 PMC3657724

[10] Temel JS Greer JA Muzikansky A Early palliative care for patients with metastatic non-small-cell lung cancer . N Engl J Med 2010 ; 363 : 733 – 42 . 20818875 10.1056/NEJMoa1000678

[11] Bakitas MA Tosteson TD Li Z Early versus delayed initiation of concurrent palliative oncology care: patient outcomes in the ENABLE III randomized controlled trial . J Clin Oncol 2015 ; 33 : 1438 – 45 . 25800768 10.1200/JCO.2014.58.6362 PMC4404422

[12] Dionne-Odom JN Azuero A Lyons KD Benefits of early versus delayed palliative care to informal family caregivers of patients with advanced cancer: outcomes from the ENABLE III randomized controlled trial . J Clin Oncol 2015 ; 33 : 1446 – 52 . 25800762 10.1200/JCO.2014.58.7824 PMC4404423

[13] LeBlanc TW El-Jawahri A When and why should patients with hematologic malignancies see a palliative care specialist? Hematology Am Soc Hematol Educ Program 2015 ; 2015 : 471 – 8 . 26637760 10.1182/asheducation-2015.1.471

[14] Earle CC Neville BA Landrum MB Trends in the aggressiveness of cancer care near the end of life . J Clin Oncol 2004 ; 22 : 315 – 21 . 14722041 10.1200/JCO.2004.08.136

[15] Earle CC Neville BA Landrum MB Evaluating claims-based indicators of the intensity of end-of-life cancer care . Int J Qual Health Care 2005 ; 17 : 505 – 9 . 15985505 10.1093/intqhc/mzi061

[16] Earle CC Landrum MB Souza JM Aggressiveness of cancer care near the end of life: is it a quality-of-care issue? J Clin Oncol 2008 ; 26 : 3860 – 6 . 18688053 10.1200/JCO.2007.15.8253 PMC2654813

[17] Barbera L Paszat L Chartier C Indicators of poor quality end-of-life cancer care in Ontario . J Palliat Care 2006 ; 22 : 12 – 7 . 16689410

[18] Tang ST Wu SC Hung YN Trends in quality of end-of-life care for Taiwanese cancer patients who died in 2000–2006 . Ann Oncol 2009 ; 20 : 343 – 8 . 18765460 10.1093/annonc/mdn602

[19] Tang ST Wu SC Hung YN Determinants of aggressive end-of-life care for Taiwanese cancer decedents, 2001 to 2006 . J Clin Oncol 2009 ; 27 : 4613 – 8 . 19704067 10.1200/JCO.2008.20.5096

[20] National Health Insurance Research Database (NHIRD) Taiwan . Available at: http://nhird.nhri.org.tw/date_01.html Accessed November 20, 2018 .

[21] Lin CC Lai MS Syu CY Accuracy of diabetes diagnosis in health insurance claims data in Taiwan . J Formos Med Assoc 2005 ; 104 : 157 – 63 . 15818428

[22] Yu KH Kuo CF Luo SF Risk of end-stage renal disease associated with gout: a nationwide population study . Arthritis Res Ther 2012 ; 14 : R83 . 22513212 10.1186/ar3806 PMC3446457

[23] Kreng VB Yang CT The equality of resource allocation in health care under the National Health Insurance System in Taiwan . Health Policy 2011 ; 100 : 203 – 10 . 21112116 10.1016/j.healthpol.2010.08.003

[24] Marcus MW Chen Y Duffy SW Impact of comorbidity on lung cancer mortality - a report from the Liverpool Lung Project . Oncol Lett 2015 ; 9 : 1902 – 6 . 25789065 10.3892/ol.2015.2916 PMC4356386

[25] Deyo RA Cherkin DC Ciol MA Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases . J Clin Epidemiol 1992 ; 45 : 613 – 9 . 1607900 10.1016/0895-4356(92)90133-8

[26] Klabunde CN Potosky AL Legler JM Development of a comorbidity index using physician claims data . J Clin Epidemiol 2000 ; 53 : 1258 – 67 . 11146273 10.1016/s0895-4356(00)00256-0

[27] Wu WC Wu JC Chiang TL Variation in the association between socioeconomic status and breastfeeding practices by immigration status in Taiwan: a population based birth cohort study . BMC Pregnancy Childbirth 2015 ; 15 : 298 . 26572970 10.1186/s12884-015-0732-8 PMC4647632

[28] Directorate-General of Budget, Accounting and Statistics, Executive Yuan, Republic of China (Taiwan) . Report on the survey of family income and expenditure, 2005. Available at: https://win.dgbas.gov.tw/fies/doc/result/106.pdf Accessed November 20, 2018 .

[29] Goel MK Khanna P Kishore J Understanding survival analysis: Kaplan–Meier estimate . Int J Ayurveda Res 2010 ; 1 : 274 – 8 . 21455458 10.4103/0974-7788.76794 PMC3059453

[30] Lai YL Su WH Palliative medicine and the hospice movement in Taiwan . Support Care Cancer 1997 ; 5 : 348 – 50 . 9322344 10.1007/s005200050090

[31] Hochman MJ Yu Y Wolf SP Comparing the palliative care needs of patients with hematologic and solid malignancies . J Pain Symptom Manage 2018 ; 55 : 82 – 8 . e1 . 28887271 10.1016/j.jpainsymman.2017.08.030 PMC5921902

[32] Tanuseputro P Beach S Chalifoux M Associations between physician home visits for the dying and place of death: a population-based retrospective cohort study . PLoS One 2018 ; 13 : e0191322 . 29447291 10.1371/journal.pone.0191322 PMC5813907

[33] Wu S-J Chiang C-J Lin C-T Improving but inferior survival in patients with chronic lymphocytic leukemia in Taiwan: a population-based study . PLoS One 2013 ; 8 : e62930 . 23638168 10.1371/journal.pone.0062930 PMC3634739

[34] Wang HM Koong SL Hsiao SC Impact of availability of an inpatient hospice unit on the parent hospital's quality of palliative care for Taiwanese cancer decedents . J Pain Symptom Manage 2011 ; 42 : 400 – 9 . 21458221 10.1016/j.jpainsymman.2010.12.011

[35] Moreno-Alonso D Porta-Sales J Monforte-Royo C Palliative care in patients with haematological neoplasms: an integrative systematic review . Palliat Med 2018 ; 32 : 79 – 105 . 29130387 10.1177/0269216317735246

[36] Ho TH Barbera L Saskin R Trends in the aggressiveness of end-of-life cancer care in the universal health care system of Ontario, Canada . J Clin Oncol 2011 ; 29 : 1587 – 91 . 21402603 10.1200/JCO.2010.31.9897 PMC3082976

[37] Sexauer A Cheng MJ Knight L Patterns of hospice use in patients dying from hematologic malignancies . J Palliat Med 2014 ; 17 : 195 – 9 . 24383458 10.1089/jpm.2013.0250 PMC3997144

[38] Kao SC Butow P Bray V Patient and oncologist estimates of survival in advanced cancer patients . Psychooncology 2011 ; 20 : 213 – 8 . 20878829 10.1002/pon.1727

[39] Odejide OO Salas Coronado DY Watts CD End-of-life care for blood cancers: a series of focus groups with hematologic oncologists . J Oncol Pract 2014 ; 10 : e396 – 403 . 25294393 10.1200/JOP.2014.001537 PMC5706135

[40] Hui D Bansal S Park M Differences in attitudes and beliefs toward end-of-life care between hematologic and solid tumor oncology specialists . Ann Oncol 2015 ; 26 : 1440 – 6 . 26041765 10.1093/annonc/mdv028 PMC4855240

[41] Wagner EH Austin BT Davis C Improving chronic illness care: translating evidence into action . Health Aff (Millwood) 2001 ; 20 : 64 – 78 . 11816692 10.1377/hlthaff.20.6.64

[42] Smith TJ Temin S Alesi ER American society of clinical oncology provisional clinical opinion: the integration of palliative care into standard oncology care . J Clin Oncol 2012 ; 30 : 880 – 7 . 22312101 10.1200/JCO.2011.38.5161

[43] Ferrell BR Temel JS Temin S Integration of palliative care into standard oncology care: American Society of Clinical Oncology Clinical Practice Guideline update . J Clin Oncol 2017 ; 35 : 96 – 112 . 28034065 10.1200/JCO.2016.70.1474

[44] Hamaker ME Stauder R van Munster BC On-going clinical trials for elderly patients with a hematological malignancy: are we addressing the right end points? Ann Oncol 2014 ; 25 : 675 – 81 . 24458474 10.1093/annonc/mdt592 PMC4433524