Catastrophic Illness

In this study, we identified 854 pancreatic cancer patients who were admitted and died in-hospital between January 2003 and December 2006. Pancreatic cancer was defined using an ICD-9-CM compatible code (157) from the Registry for Catastrophic Illness Patient database, which is a separate subsection of the NHI database. Most patients who are diagnosed with cancer sign up for a Catastrophic Illness Card in Taiwan, and these cardholders are exempt from cost-sharing under the NHI program. The effects of comorbidities were estimated using the Charlson comorbidity index [14 (link)]. Propensity score matching was performed to minimize the potential influence of selection bias, whereby logistic regression was used to create a propensity score for the inpatient palliative care and acute hospital care groups [15 (link)]. The covariates that we examined included patient characteristics (age and sex), Charlson comorbidity index, hospital location, and hospital accreditation level. One-to-one matching was subsequently performed for the two patients groups, using the nearest-neighbor matching method.

A retrospective fixed cohort population was used as the original study population. Every claimant of the NHI Program at any time during 2005 was included in the population (22,717,053 people) selected for random sampling. Using a random number generator to produce 1,073,891 random numbers, the original claims data of 1,000,000 randomly-sampled beneficiaries from the year 2005 was the original study population of the present study. The majority of the sample population of 1,000,000 subjects were aged from 20 to 65 years, and the male and female subpopulations were of a similar size (49.6% vs. 50.4%, respectively). There were no significant differences in the gender distribution, age distribution or average insured payroll-related amount between the claimants in the 2005 sampled data and the original NHIRD [7 (link),8 (link)]. The registration and claims data of these 1,000,000 individuals collected by the NHI Program could be followed until 2008. In 2008, all the subjects aged 65 years or older were selected from the original study population to be the sampled population for a further analysis.
If a patient’s ailment is diagnosed by a physician as a “catastrophic illness” under Department of Health guidelines, the patient can submit related information and apply for a catastrophic illness certificate. The application will be formally reviewed, and if approved, the information is entered into his or her IC card [10 ]. Therefore, the 2008 registered catastrophic illness dataset of Taiwan (including 30 categories, presented in the result section) was used to verify the target cases from the sampled population.
Moreover, emergency cases served by the NHI system are recorded within the dataset of “ambulatory care expenditure by visit”, and all cases classified as a “medical emergency” could be identified. Finally, to study the difference in ED utilization between Taiwan’s elderly with a catastrophic illness and those without, subjects aged 65 years or older in 2008 were selected and their medical records regarding ambulatory services in 2008 were analyzed.

In 1996, the Taiwan government organized a national-wide and single-payer health insurance program called the “Taiwan National Health Insurance Program” (Taiwan NHIP). Since 1998, the NHIP had covered >99% of the 23 million Taiwanese.^{[13 ]} Consequently, the National Health Research Institutes (NHRI) established the National Health Insurance Research Database (NHIRD) composed of the annual reimbursement claims data from the NHIP.
The present study used the Longitudinal Health Insurance Database (LHID), a subset of NHIRD, to construct the study population. The LHID contained the annual reimbursement claims data from 1 million insured people randomly selected in 1996 to 2000. Based on the NHRI report, the distribution of age and sex ratios between LHID and NHIRD was not different. The NHRI provided a scrambled and anonymous identification number for each insured person to link each claim data, including registry for beneficiaries, out-patient and in-patient visits, and other medical services records. The Institutional Review Board of China Medical University in central Taiwan exempted this study from the informed consent requirement (CMU-REC-101-012).
The disease record in NHIRD was defined by the disease according to the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). Cancer diagnosis database was collected from the catastrophic illness registry and the history of other diseases was gathered from in-patient and out-patient files. The Registry of Catastrophic Illness Database is a subpart of the NHIRD. In Taiwan, insured people with major diseases can apply for a catastrophic illness certificate that grants exemption from copayment. The issuance of catastrophic illness certificates was validated by at least 2 specialists, based on careful evaluation of the medical records, laboratory studies, and imaging studies. Individuals would be issued a catastrophic illness certificate when they meet the diagnostic criteria for major diseases. Cancers, cirrhosis with esophageal varices bleeding, uremia on hemodialysis, and organ failure are statutorily included in the catastrophic illness category.

The outcome of interest in this study was newly diagnosed autoimmune diseases after the index dates. In Taiwan, people with several autoimmune diseases can apply for catastrophic illness certification to receive a copayment exemption from the NHI program, and all the applications are reviewed by rheumatologists. These autoimmune diseases, including SLE, systemic sclerosis, Sjögren’s syndrome, inflammatory myopathy, rheumatoid arthritis, Behcet’s syndrome, systemic vasculitis, type I diabetes mellitus, multiple sclerosis, myasthenia gravis, inflammatory bowel diseases, autoimmune hemolytic anemia, and pemphigus, were identified from the Registry for Catastrophic Illness, and thus the diagnoses should be valid. For autoimmune diseases that did not fulfill the requirements to be considered catastrophic illnesses in Taiwan, including ankylosing spondylitis, post-infectious arthritis, uveitis, psoriasis, autoimmune thyroid disease, primary adrenocortical insufficiency, Guillain–Barré syndrome, autoimmune encephalomyelitis, and celiac disease, one hospital admission or at least three outpatient visits with relevant ICD-9-CM or ICD-10-CM codes were regarded as the confirmed diagnosis (S1 Table). The follow-up of study participants started on the index dates and continued until the confirmation of study outcomes, death, or August 31, 2018, whichever occurred first.

Head and neck cancer cases between 1980 and 2019 were obtained from the national Taiwan Cancer Registration (TCR) database (https://twcr.tw/). The Taiwan Cancer Registration (TCR) database has collected newly diagnosed cancer cases from hospitals with 50 or more beds in Taiwan since 1979 [9 (link), 10 (link)]. In addition, the completeness of the TCR, the percentage of cases with death certificates, and the percentage of morphological verification in 2016 were respectively 98.4, 0.9, and 93%, while the completeness was measured by all registered cancer cases divided by all potential cancer cases from profiles of death certificate, NHI catastrophic illnesses, and four major cancer screening programs [9 (link)]. All cases in this analysis were classified based on the International Classification of Diseases for Oncology, third edition (ICD-O-03) [11 ]. Head and neck cancer cases were categorized into oral cancer (C00, C02, C03, C04, C050, C058, C059, and C06, excluding C024), oropharyngeal cancer (C01, C024, C051, C052, C09, C10, C142, and C148), hypopharyngeal cancer (C12, C13, and C140), and laryngeal cancer (C32).
Based on the 2000 World Health Organization standard population, the age-adjusted incidence rate in men from 1980 to 2019 was only analyzed due to the low incidence in women in Taiwan. For the analysis of long-term trends, the age-specific incidence rate from 1980 to 2019 was calculated for specific age groups, time periods, and birth cohorts. The age-specific incidence rate was classified into eighteen 5-year age groups (0–4, 5–9, 10–14, 15–19, 20–24, 25–29, 30–34, 35–39, 40–44, 45–49, 50–54, 55–59, 60–64, 65–69, 70–74, 75–79, 80–84, and 85+) and eight 5-year time periods (1980–1984, 1985–1989, 1990–1994, 1995–1999, 2000–2004, 2005–2009, 2010–2014, and 2015–2019). In addition, the birth cohort was divided into eleven birth groups (1930–1934, 1935–1939, 1940–1944, 1945–1949, 1950–1954, 1955–1959, 1960–1964, 1965–1969, 1970–1974, 1975–1979, and 1980–1984) and twelve 5-year age groups (30–34, 35–39, 40–44, 45–49, 50–54, 55–59, 60–64, 65–69, 70–74, 75–79, 80–84, and 85+). Moreover, to describe the linear change in the age-adjusted incidence rate from 1980 to 2019, a join point regression model was utilized to detect the change point and calculate the average annual percent change (AAPC) and annual percent change (APC) [10 (link)]. In addition, the 95% confidence intervals of the average annual percent change (AAPC) and annual percent change (APC) were analyzed. 95% confidence interval indicated 95% would fall between the upper limit and the lower limit, while 95% confidence interval including 0 showed statistically nonsignificant. The research protocol was approved by the Institutional Review Board of Fu-Jen Catholic University (No. C104014).

In this retrospective cohort study, we analyzed database obtained from the Registry for Catastrophic Illness Patients (CIPR), which consists of disease categories including ESRD, with verifying chronic dialysis and the irreversible status by two nephrologists. The data represented a subset of the National Health Insurance Research Database (NHIRD) covering a period from 1 January 2000, to 31 December 2016 (National Science Council—Taiwan, RRID: SCR_011434). The NHIRD, with comprehensive records of claims relating to outpatient and inpatient medical care, enrolled data from 23 million Taiwanese people as in 2015 [16 ,17 ]. The RCID comprises information on the registry of beneficiaries, diagnostic codes, medical procedures, and prescription data. All data extracted from the NHIRD were deidentified to protect privacy.
Definitions of diseases are based on the International Classification of Diseases, 9th Revision, Clinical Modification for cases diagnosed before 1 January 2016 and the International Classification of Diseases, 10th Revision, Clinical Modification for cases diagnosed afterwards. The list of ICD-9 and ICD-10 codes were used to define the inclusion criteria for patients on dialysis, outcome events, and comorbidities as shown in Supplemental File S1: Table S1.