End-of-life discussions reduce the utilization of life-sustaining treatments during the last three months of life in cancer patients

Studies to examine the impact of end-of-life (EOL) discussions on the utilization of life-sustaining treatments near death are limited and have inconsistent findings. This nationwide population-based cohort study determined the impact of EOL discussions on the utilization of life-sustaining treatments in the last three months of life in Taiwanese cancer patients. From 2012 to 2018, this cohort study included adult cancer patients, which were confirmed by pathohistological reports. Life-sustaining treatments during the last three months of life included cardiopulmonary resuscitation, intubation, and defibrillation. EOL discussions in cancer patients were confirmed by their medical records. Association of EOL discussions with utilization of life-sustaining treatments were assessed using multiple logistic regression. Of 381,207 patients, the mean age was 70.5 years and 19.4% of the subjects received life-sustaining treatments during the last three months of life. After adjusting for other covariates, those who underwent EOL discussions were less likely to receive life-sustaining treatments during the last three months of life compared to those who did not (Adjusted odds ratio [AOR] 0.87; 95% confidence interval [CI] 0.85–0.89). Considering the type of treatments, EOL discussions correlated with a lower likelihood of receiving cardiopulmonary resuscitation (AOR = 0.45, 95% CI 0.43–0.47), endotracheal intubation (AOR = 0.92, 95%CI 0.90–0.95), and defibrillation (AOR = 0.54, 95%CI 0.49–0.59). Since EOL discussions are associated with less aggressive care, our study supports the importance of providing these discussions to cancer patients during the EOL treatment.

www.nature.com/scientificreports/ Advance care planning (ACP) is the process of discussing patient preferences regarding life-sustaining treatments and place of death during EOL care, and involves healthcare professionals, patients, and their family 5 . It is considered a means for helping patients die in their preferred treatment and is emphasised as an end-of-life care strategy 5 . A previous report showed that patients engaging in ACP have more stable EOL preferences compared to those who do not 6 .
Comprehensive EOL discussions with physicians during the ACP consultation could allow cancer patients to confront the limitations of medical treatments and consider their preferences regarding life-sustaining treatments through the EOL treatment. A meta-analysis review showed that these discussions predicted an increased use of hospice care and better quality of death 7 . Although cancer patients may avoid discussing their preference of life-sustaining treatments during EOL discussions 8,9 , healthcare providers are allowed-or morally obligated-to decide whether life-sustaining treatments are viable for terminal cancer patients 10 .
Although EOL discussions offer patients the opportunity to define their expectations for life-sustaining treatments that they want to receive during the EOL treatment, the impact of EOL discussions on the use of lifesustaining treatments near death has not been extensively studied. A US cohort study followed up 332 patients with advanced cancer and found that patients who participated in EOL discussions were less likely to receive ventilation (adjusted odds ratio [AOR] 0.26; 95% confidence interval [CI] 0.08-0.83) and resuscitation (AOR 0.16, 95% CI 0.03-0.80) during the EOL treatment compared to those who did not 8 . In another observational study in the US involving 145 patients with advanced cancer, EOL discussions lowered the rates of ventilation and resuscitations during the EOL treatment 2 . Previous studies on the association between EOL discussions and the use of life-sustaining treatment, however, have used small sample sizes (n = 145) 2 and inadequately controlled for potential confounders such as dementia 2,8 .
EOL discussions serve to maximize patients' quality of life while considering their goals and expectations regarding EOL treatment. It is important to understand the impact of these discussions on the administration of life-sustaining treatments among terminally ill patients. We therefore conducted a nationwide population-based cohort study to determine the impact of EOL discussions on the utilization of life-sustaining treatments during the last three months of life among Taiwanese cancer patients, from 2012 to 2018.

Promotion of end-of-life discussions in terminally ill patients in Taiwan. The Taiwan National
Health Insurance Administration, since 2012, has initiated an EOL discussions program that encourages healthcare providers to provide these discussions to terminally ill patients 11 . While patients were admitted to the hospital, healthcare providers offered EOL discussions to those with life-limiting diseases 11 . During the discussions, healthcare providers discuss patients' goals and preferences concerning medical care and life-sustaining treatments towards the end of their lives 11 . Each EOL discussion with a patient awards the healthcare provider with the equivalent of US$75 11 . There should be a record of all discussion content in the patient's medical chart and each session must be at least an hour long 11 . Study population. This nationwide cohort study used the Taiwan National Health Insurance Research Database (NHIRD), which contains healthcare data from more than 99% of the population in Taiwan 12 . It is a large-scale database derived from the national health insurance system, which consists of registration files and original claims data. The database de-identified and scrambled the patients' identification codes before releasing the data for research purposes 13 .
This cohort study selected subjects aged 18 years or older who had received a cancer diagnosis and died between January 1, 2012 and December 31, 2018. We identified these patients from the Registry for Catastrophic Illness. In Taiwan, the Registry for Catastrophic Illness requires a peer review of pathohistological reports before a cancer diagnosis is reported 14 . This registry was also linked to Taiwan's death certificate database to confirm the demise of cancer patients 15 . The Institutional Review Board of Taipei City Hospital (no. TCHIRB-10709107-W) approved this study. The informed consents for study participants were waived by the Institutional Review Board of Taipei City Hospital. All methods in this study were performed in accordance with relevant guidelines and regulations. This study is also in accordance with the Declaration of Helsinki.
Outcome variable. The outcome was life-sustaining treatments during the last three months of life in cancer patients. Life-sustaining treatments included cardiopulmonary resuscitation, intubation, and defibrillation 16 . Main explanatory variable. The main explanatory variable was EOL discussions with physicians, which was determined by patients' medical records.
Controlling variables. The controlling variables included sociodemographic characteristics, type of cancer, and comorbidities. Sociodemographic factors included age, sex, urbanization, and income level. Urbanization described whether subjects resided in urban, suburban, or rural areas 17 . We calculated the income level from the average monthly income of the insured person and grouped it into three categories: low (≤ 19,200 New Taiwan Dollars [NTD]), intermediate (19,201 NTD to < 40,000 NTD), and high (≥ 40,000 NTD). Type of cancer and the comorbidities in the study subjects were determined according to the International Classification of Diseases, Ninth and Tenth Revision, Clinical Modification (ICD-9-CM and ICD-10-CM) code. The type of cancer included solid tumors and hematologic malignancies. Subjects' comorbidities include diabetes, chronic kidney disease, congestive heart failure, coronary heart disease, liver cirrhosis, chronic obstructive pulmonary disease, dementia, cerebrovascular disease, and depressive disorder (Supplementary Table S1). A person was considered to have a comorbidity only if the condition occurred in an inpatient setting or in three or more outpatient visits 18  www.nature.com/scientificreports/ Statistical analysis. First, we analyzed the subjects' demographic data. We then analyzed categorical data using the Pearson χ 2 test where appropriate. We presented continuous data as mean ± standard deviation (SD), and conducted a two-sample t-test to compare outcomes between patients who underwent EOL discussions and those who did not.
We assessed the crude associations of EOL discussions and other covariates with the outcome (utilization of life-sustaining treatments during the last three months of life) by computing the odds ratios (ORs) and corresponding 95% confidence intervals (CIs). We then performed a multivariate logistic regression to estimate the association between EOL discussions and the utilization of life-sustaining treatments after adjusting for potential confounders. A variable with p < 0.05 was defined as a significant factor associated with the utilization of lifesustaining treatments in the multivariate analysis. Adjusted odds ratios (AOR) with 95% confidence intervals (CI) indicated the strength and direction of these associations.
We conducted subgroup and sensitivity analyses to examine the associations between EOL discussions and the utilization of life-sustaining treatments, after stratifying participants by age and sex. To examine the robustness of the main findings, propensity score analysis was conducted to evaluate the associations between hospice care services and utilization of life-sustaining treatments during the last three months of life. Logistic regression was conducted to calculate the probability of enrolling in a hospice care program in cancer patients according to their age, sex, income level, type of cancer, and comorbidities. A multivariate logistic regression was used to estimate the association between hospice care services and utilization of life-sustaining treatments after adjusting for patients' propensity score. We performed all data management and analyses using the SAS® v9.4 statistical software package (SAS Institute, Cary, NC, USA).

Results
Participant selection. This cohort study involved the raw data of 383,128 deceased cancer patients from 1 January 2012 to 31 December 2018. After excluding those younger than 18 years (n = 850) and those with missing data (n = 1071), 381,207 patients were eligible for the purposes of the study. The overall mean (SD) age was 70.5 (14.3) years; 65.6% of the subjects were male; and 14.1% had EOL discussions with healthcare providers towards the end of their lives. The median duration from the EOL discussion to death for the 53,783 patients completing EOL discussions with healthcare providers was 34 days (interquartile range 16-79 days); of these patients, 7382 (13.7%) had such discussions after utilizing life-sustaining treatment. Table 1 shows the characteristics of cancer patients with and without EOL discussions before death. The patients who participated in EOL discussions tended to be younger and were more likely to be female. The proportion of patients completing EOL discussions with healthcare providers significantly increased from 3.  Table 2 shows the univariate and multivariate analyses of factors associated with life-sustaining treatments during the last three months of life among the cancer patients. After adjusting for the sociodemographic factors and comorbidities, those who underwent EOL discussions were less likely to receive life-sustaining treatments during the last three months of life, compared to those who did not (AOR = 0.87, 95% CI 0.85-0.89). Table 3 shows the multivariate analyses for the association of EOL discussions with cardiopulmonary resuscitation, endotracheal intubation, and defibrillation in cancer patients. After adjusting for the sociodemographic factors and co-morbidities, patients who underwent EOL discussions were less likely to receive cardiopulmonary resuscitation (AOR = 0.45, 95% CI 0.43-0.47), endotracheal intubation (AOR = 0.92, 95% CI 0.90-0.95), and defibrillation (AOR = 0.54, 95% CI 0.49-0.59) during their last three months of life, compared to those who did not. Table 4 shows the results of subgroup analyses of the association between EOL discussions and life-sustaining treatments after stratifying participants by age and sex. In all patient subgroups, the discussions significantly predicted a lower likelihood of receiving life-sustaining treatments towards the end-of-life.

Subgroup and sensitivity analysis for the association between end-of-life discussions and life-sustaining treatments.
Propensity score analysis for the association between end-of-life discussions and utilization of life-sustaining treatments. Propensity score analysis showed that cancer patients with EOL discussions were less likely to receive life-sustaining treatments during the last three months of life than those without the discussion (AOR = 0.87, 95% CI 0.85-0.89), including cardiopulmonary resuscitation (AOR = 0.45, 95% CI 0.43-0.47), endotracheal intubation (AOR = 0.92, 95% CI 0.90-0.94), and defibrillation (AOR = 0.54, 95% CI 0.49-0.59). Our study revealed robust associations between EOL discussions and life-sustaining treatments after stratifying patients by age and sex. In all patient subgroups, the discussions significantly lowered the likelihood of receiving life-sustaining treatments near death.
This study found that 14.1% of cancer patients in Taiwan participated in EOL discussions, which was lower than 37.0% among cancer patients in the US 8 . The culture difference regarding death-related issues between Asia and Western countries may explain the lower rate of completing EOL discussions among Taiwanese cancer patients. In traditional Chinese culture, death-related topics are thought to be taboo, and mentioning it is sacrilegious and to be avoided 19 . A previous survey in Hong Kong reported that 73% of elderly patients with chronic diseases have never received discussions regarding their EOL treatment 19 . Since EOL discussions provide patients  www.nature.com/scientificreports/ the opportunity to define their goals and expectations regarding the EOL treatment 20 , our study suggests that it is imperative to provide such discussions for cancer patients nearing the end of their lives. This study found that 19.4% of cancer patients in Taiwan received life-sustaining treatments in the last three months of life. A previous report showed that the proportion of cancer patients receiving intensive care in the last six months of life was 40.3% in the US, 17.5% in Belgium, 15.2% in Canada, and 8.2% in Germany 21 . Since intensive care in patients with terminal diseases is associated with lower quality of life 3 , our study suggests that healthcare professionals should proactively provide EOL discussions for terminally ill patients to discuss their preferred treatment during EOL care.
The present cohort study found that cancer patients who underwent EOL discussions had 18% lower rates of receiving life-sustaining treatments in the last three months of life, than those who did not. EOL discussions could allow terminally ill patients to consider their preferences regarding life-sustaining treatments during the EOL care. A previous report showed that EOL discussions significantly increased the rates of do-not-resuscitate order completion among advanced cancer patients 22 . Another cohort study in the US found that EOL discussions correlated with lower rates of ventilation and resuscitation near death 8 . Moreover, a study in the US analyzed 145 cancer patients and found that patients who had undergone EOL discussions had lower rates of receiving ventilation and resuscitations in the last week of life 2 . Our cohort study followed up 381,207 cancer patients and found that EOL discussions associated with a lower utilization of life-sustaining treatments during the last three months of life. With regard to treatment types, cancer patients who had EOL discussions had a lower likelihood of receiving cardiopulmonary resuscitation, endotracheal intubation, and defibrillation during EOL care. Thus the findings of our study suggest that EOL discussions could reduce the utilization of life-sustaining treatments in cancer patients near death.
The EOL conversations with physicians improve the patients' understanding of the futility of aggressive therapies during EOL care 7 . This may account for the lower rate of receiving life-sustaining treatments at that stage. While cancer patients had EOL discussions with physicians, the benefit and harm of life-sustaining treatments during the EOL care would be discussed and emphasized. Through the EOL discussions with healthcare providers, cancer patients could consider and document their treatment preferences during EOL care. A prior study showed that EOL discussions significantly improved patients' understanding of their terminal illness and subsequently reduced the likelihood of receiving aggressive treatments during EOL care 23 . As EOL discussions are associated with less aggressive care 7 , the findings of our study suggest that it is important to provide them for cancer patients during EOL treatment.
This study showed that cancer patients receiving life-sustaining treatments during EOL care had higher medical expenditures than those not. A previous study involving 314 veterans in the US showed that aggressive EOL treatment was associated with significantly higher inpatient costs during a terminal hospitalization 24 . Another US study using administrative data found that EOL discussion was associated with significant hospital cost savings among patients during EOL care 25 . Since aggressive EOL treatment is associated with higher health care cost and poor quality of EOL care in terminally ill patients 2 , our study suggests that EOL discussions should be provided for cancer patients to discuss their treatment preference during EOL care.
This nationwide cohort study has several strengths. First, this study is the largest cohort study to determine the impact of EOL discussions on the utilization of life-sustaining treatments during the last three months of life among cancer patients. The research design enhances the validity of our findings by including unbiased subject selection and strict cancer diagnostic criteria. Moreover, since the Taiwan National Health Insurance covers all medical care of cancer patients, this nationwide population-based study traced all cancer patients, thus minimizing referral bias. Furthermore, the study's large sample size sufficiently detected the real, albeit subtle, difference between cancer patients who had and those who had not received EOL discussions towards the end of their lives.
This cohort study has several limitations. First, there may be important factors (e.g., patients' religion, functional status, and educational level) associated with the decision of receiving life-sustaining treatments, which the National Health Insurance Research Database did not record. Second, of the 53,783 patients completing EOL discussions with healthcare providers, 7382 (13.7%) had such discussions after utilizing life-sustaining treatment, which would underestimate the impact of EOL discussions on the utilization of life-sustaining treatments during the last three months of life among cancer patients. Finally, since all our enrollees were Taiwanese, the external validity of our findings may be a concern. Therefore, the generalizability of our results to other non-Asian ethnic groups requires further verification. However, our findings suggest new avenues for future research.
In summary, this nationwide, cohort study found that cancer patients who had EOL discussions were less likely to receive life-sustaining treatments during the last three months of life compared to those who did not have EOL discussions. With regard to treatment types, EOL discussions correlated with a lower likelihood of receiving cardiopulmonary resuscitation, endotracheal intubation, and defibrillation during the last three months of life. As end-of-life discussions are associated with less aggressive EOL care, it is important to provide them for cancer patients during EOL treatment.