Study design

A retrospective, longitudinal population-level study design was conducted to fulfill the purposes of our study. We also divided the study period into 4 sub-periods: 2002–2005, 2006–2009, 2010–2013, 2014–2017.

Data source and study population

We used the 2001–2017 NHI claims data, Taiwan’s Death Registry, and the Taiwan Cancer Registry in this study. The Death Registry was used to identified the study population; the Taiwan Cancer Registry was used to identify if a deceased person had had cancer or not; and the NHI claims data were used to identify the utilization of hospice care in the previous 12 months before a patient’s death.

People who died between 2002 and 2017 were included in this study. We identified their death events and the causes of death by consulting the National Death Registry database first, and we accessed the NHI claims data and the Taiwan Cancer Registry to retrieve the patients’ medical records. We used NHI claims data to identify whether a patient had received hospice care in the previous 12 months before that patient’s death, and we also used that data to retrieve each patient’s demographic characteristics. In addition, we used Taiwan Cancer Registry to retrieve a patient’s cancer-related information.

Exclusion criteria

Deceased people who were under 18 years old, and deceased people for whom the information on gender, marital status and poverty status was unknown were excluded in this study.

Definition of variables of interest.

a.

Dependent variables.

There are 2 dependent variables used in this study: hospice care use, and the Initiation time of 1st hospice care use. The definitions of them are as follows:

Hospice care use

Hospice care use was defined as receiving any types of hospice care at least once in the last 12 months before death.

Initiation time of 1st hospice care use

The duration between 1st hospice care use and the date of death in the last year was used for calculation the initiation time.

b.

Variables for stratified analysis.

We used demographic characteristics and health status for stratified analysis. Demographic characteristics include age of death, gender, marital status, poverty status, and urbanization of residence. Health status includes: history of having severe disease, and having the experience in receiving CPR in the last year. The details are following:

Age of death

Age of death was classified as young adult (18 to 40 years old), middle-aged (41 to 65 years old), old (66 to 85 years old), and oldest old (over 85 years old).

Marital status

Marital status was retrieved from the death registry, and there are 4 types of marital status: single, married, divorced, and widowed.

Poverty status

As for poverty, the registry for beneficiaries were used to distinguish patients in the low-income group from other populations. In Taiwan, the NHI Scheme classifies the insured into six insured categories according to the insured’s occupation. Households below the poverty line belong to classification 5. We used this information in the National Health Insurance Research Database as a criterion to identify the poverty status.

Urbanization of residence

A person’s residential area was linked to the urbanization level. Following Chang et al., [11] the actual location of each person is assumed in this study to be where an individual has the most outpatient and pharmacy visits. The location of each clinic and pharmacy is recognized as either urban or rural according to the definition of urbanization published by Taiwan’s National Health Research Institutes. All 365 townships in Taiwan are classified into 7 clusters based on the following indicators: population density (people/km2), proportion of people with a college degree or above, proportion of elderly people over 65 years of age, proportion of people who are agricultural workers, and the number of physicians per 100,000 people [12]. Residential areas located in clusters 1 to 3 were categorized as urban, others as rural.

History of having severe disease

Taiwan NHI’s catastrophic illness and rare disease certification, and cancer registry were used for identifying the history of having severe disease. If a person who can be identified in any of these two lists, then he/she was classified as having the history of severe disease.

Receiving CPR within a year before death

If a person who had ever received CPR within a year before his/her last month of life, then he/she was classified as receiving CPR within a year before death.

Statistical analysis

In statistical testing, two-sided P-value ≤ 0.05 was considered statistically significant. The distributional properties of continuous variables were expressed by mean plus or minus standard deviation (SD), whereas categorical variables were presented by frequency and percentage. A chi-square test and two-sample t-tests were conducted to test the differences in hospice care utilization with respect to various people’s demographic characteristics and health status. All statistical analyses were performed using SAS (version 9.4, SAS Institution Inc., Cary, NC, USA).