Abstract
Introduction:
Stroke remains a leading cause of death worldwide. Stroke in young adults is an important issue, gaining extra attention in recent years. This study aims to investigate the mortality after stroke in young adults in Taiwan.
Patients and methods:
This is a registry- and population-based study in Taiwan of patients aged 20–50 years with first-ever stroke between 1999 and 2012, with follow-up until January 1, 2022. Patients and mortalities were identified through Taiwan National Health Insurance database.
Results:
The study population included 65,097 patients with stroke (mean age, 42.6 ± 6.6 years; 30.5% woman). There were 23,481 (36.1%) intracranial hemorrhage, 37,522 (57.6%) ischemic stroke, and 4094 (6.3%) stroke not otherwise specified. At the end of follow-up, a total of 18,248 deaths (28.0%) occurred during a median follow-up of 9.8 years (interquartile range, 6.4–13.7 years).
Conclusion:
Taiwan young adults who were 30-day survivors of first-ever stroke have significantly higher long-term mortality rates when compared to other population-based studies.
Introduction
Stroke is a major cause of death and disability worldwide. Although stroke incidence has decreased by around 40% in high-income countries in recent decades, 1 it is still a serious problem especially in Asia, which comprised more than 60% of the world’s population. Stroke mortality is higher in Asia than in Western Europe, the Americas, or Australia and New Zealand regions, apart from some countries such as Japan. 2 Although mortality rates in young adults with stroke are lower compared to older patients, they still have higher mortality rates compared to the general young population. 3 Furthermore, the socio-economic burden is very high. 4
Previous large observational studies of long-term mortality after stroke in young adults were either restricted to European and Northern American populations 5 or focused solely on ischemic stroke.3,6 –11 Epidemiological studies discussing different subtypes of stroke, other than ischemic type, in various ethnicities do exist,4,12,13 but none focused on a young population to the best of our knowledge.
Ekker et al. 14 published a retrospective cohort study consisting of 15,527 stroke patients aged 18–49 years with a median duration of 9.3 years follow-up to explore the issue of long-term mortality in young adult stroke survivors in the Netherlands. In reciprocation, we conducted a similar study to investigate the epidemiology of young stroke in Taiwan. Besides evaluating the cumulative mortality over time in this Asian population, the morality rates would be compared to that of the Dutch to provide a basic visual comparison between East-Asia and Western-Europe on this issue.
Subject and methods
Data sources
Taiwan launched the National Health Insurance (NHI) program in 1995 to finance health care for all citizens, with a coverage rate of up to 99%. 15 The NHI is a mandatory universal health insurance program offering comprehensive medical care in Taiwan. All medical records were maintained by the Taiwan Ministry of Health and Welfare, and de-identified secondary data were released for research purposes in the year 2000. 15 This study was approved by the institutional review board of E-Da Hospital.
Study populations
A retrospective cohort study was conducted, which enrolled stroke inpatients (431.X, 433.X, 434.X, and 346.X) utilizing the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) from January 1, 1999 to December 31, 2012. Patients aged below 20 or above 50 years were excluded. The stroke patients were classified into intracerebral hemorrhage (ICH), ischemic stroke, and stroke not otherwise specified cohorts. The first day of stroke diagnosis was defined as the index day in these three cohorts. Comorbidities were classified as those existing prior to the index day and then analyzed using the Charlson Comorbidity Index (CCI). Young adults were targeted to minimize the significant contributory effects of comorbidities on stroke mortality. 13 Furthermore, different risk factors such as hypertension, diabetes, and cardiac diseases, contribute to different pathophysiological mechanisms leading to different distributions in stroke subtypes. 12 The follow-up time period for the analyses was defined as the time between the index stroke and the day of NHI enrollment termination, death, or the end of this study period (1 January 2022). Figure 1 shows the flow chart of this study.
Flow chart of study design.
Outcomes
We divided our cohort into “20–29 years,” “30–39 years,” “40–49years” age groups. The primary outcome was all-cause cumulative mortality at the end of follow-up, stratified for age, sex, and stroke subtypes in 30-day survivors. The secondary outcomes were cumulative annual mortalities in the 30-day survivors. Deaths were identified through the NHI. Under this mandatory system with near maximal coverage, missing data were kept to a minimum.
Statistical analyses
The observed deaths per 1000 person-years in all strokes, ICH, ischemic stroke and stroke not otherwise specified cohorts were analyzed. The Cox proportional hazards models were used in calculating the hazard ratios (HRs) between men and women in all strokes, ICH, ischemic stroke and stroke not otherwise specified cohorts. The Kaplan–Meier method was used in determining survival probability in men and women in the four cohorts, and differences between men and women were tested using the log–rank test. All data management and HRs calculations were done using the Statistical Analysis System (SAS) software for Windows (version 9.4; SAS Institute, Cary, NC). Statistical significance was defined as a p-value of less than 0.05.
The Dutch data
Ekker et al. 14 published a registry-based study in the Netherlands, providing relevant data from Western Europe on stroke in young adults. The observed deaths per 1000 person-years in Taiwan were compared to corresponding data from the Netherlands, stratified by age, sex, and stroke cohorts, to provide a basic visual comparison between East-Asia and Western-Europe regarding young stroke mortality rates.
The rationale for this comparison is that our study, reciprocating the Dutch study, 14 focused on adults with age below 50 years and spanned the first decade of the 2000s. More importantly, several similarities can be appreciated between these two nations. In 2017, which marked the end of the Dutch study follow-up, 14 23.67 million (median age 40.3 years) lived in Taiwan with a population density of 669 per km2 over a total land area of 32,260 km2, with Han Chinese making up more than 98% of the people, 16 and the Heath Care Index was 81.14. 17 In that year, 17.02 million (median age 42.3 years) lived in the Netherlands with a population density of 414 km2 over a total land area of 41,540 km2, with Dutch making up more than 75% of the people, 17 and the Heath Care Index was 82.50. 18 In view of the general similarities between the two countries other than ethnicity, this study attempt to present a basic visual comparison on young stroke mortality rates between East-Asia and Western-Europe.
Results
Table 1 shows the baseline data of the 65,097 stroke patients identified from Taiwan NHI database over 14 years, after exclusion of patients less than 20 years old, more than 50 years old, death within 30 days after index stroke, and missing data. They were classified into any stroke (N = 65,097), ICH (N = 23,481), ischemic stroke (N = 37,522), and stroke not otherwise specified (N = 4094) subgroups. The distribution of ICH was 36.1%, ischemic stroke was 57.6%, and stroke not otherwise specified was 6.3%. Table 1 showed the baseline characteristics of all the young stroke patients and their subtypes, including demographics such as sex, age distributions and comorbidities. Patients diagnosed with ischemic stroke were older (43.1 ± 6.3 years) and associated with more comorbidities, especially cerebrovascular disease (29.1%).
Demographics including age, sex, Charlson comorbidity index (CCI), duration of follow-up and overall deaths in stroke survivors aged 20–50 years.
CCI score ranges from 0 to 6, with higher scores indicating more comorbidity.
Primary outcome
Majority of the study population (88.6%) had a CCI less than or equals to 3 with the ICH cohort having the lowest proportion of a CCI more than 3 (8.0%). The ICH man cohort had the highest mortality rate (34.0%). Table 2 shows the subgroup analysis of mortality within different age groups and sexes. For any stroke, the highest mortality lies in the 40-–49-year-old man group (observed deaths = 34.7 per 1000 person-year). The number of deaths in men was greater than that in women in all four stroke cohorts and all age groups. The greatest difference in mortality occurred in the 40–49-year-old ICH cohort, with 40.6 observed man deaths per 1000 person-years compared to 27.4 observed woman deaths per 1000 person-years.
All-cause mortality rates in stroke survivors, stratified in age groups.
Secondary outcome
Cumulative survival for all stroke subtypes (Supplementary Table 1) was visualized using Kaplan–Meier diagrams in Figure 2. Survival probability was higher in women across the board, most significant for stroke not otherwise specified, followed by ICH, then ischemic stroke. Different patterns in the prognoses of ischemic stroke and ICH cohorts were observed. For ischemic stroke cohort, both sexes had more than 90% cumulative survival at the third year. The difference in cumulative survival became apparent from fourth year onward. For ICH cohort, the cumulative survival was already different at the second year, 90.3% for woman and 89.4% for man. Then, the difference increased over the years. At the 12th year, the cumulative survival was 74.9% for woman and 65.7% for man.
Cumulative survival over time in stroke survivors: (a) any stroke, (b) intracranial hemorrhage, (c) ischemic stroke, and (d) stroke not otherwise specified.
Comparison of Taiwan and Dutch Study
Prominent findings emerged when data from East-Asia and Western-Europe are placed side by side, shown in Figure 3. The greatest difference in mortality was observed in ICH of man in 40–49 years age group, where Taiwan was 67.9 and the Dutch was 22.3, a difference of 45.6 observed deaths per 1000 person-years. The smallest difference in mortality was observed in stroke not otherwise specified of woman in 30–39 years age group, where Taiwan was 16.6 and the Dutch was 7.0, a difference of 9.6 observed deaths per 1000 person-years. The mortality rates were much higher in Taiwan than that in the Netherlands regardless of stroke subtypes and age groups across the board.
Mortality rates in Taiwan stroke survivors compared with the Dutch stroke survivors.
Discussion
When compared to other similar studies in Western populations, the mortality rates of young survivors after a stroke event in Taiwan were higher, regardless of stroke subtypes, sexes, and age groups. Stroke in young adults is receiving more and more attention in recent years. However, most studies on stroke in young adults are hospital-based,5,19 –21 focused on single stroke pathology5,21,22 or with follow-up until half-decade only.10,23 To the best of our knowledge, this study is the first registry- and population-based study that focused on long-term stroke mortality rates of young adults in an Asian population. Main findings include that the cumulative survival was higher in women than men for all stroke subtypes; the proportion of ischemic stroke was higher than ICH; and the survival in ischemic stroke was significantly higher compared with ICH. These general trends concurred with previous literature.4,12,14 It is demonstrated in other studies that stroke, both ischemic stroke and ICH, is more common among men than women, except for subarachnoid hemorrhage (SAH) where women have higher incidence. 24 The pathophysiology and presentation differ among sexes as well. 25 Literature have shown strokes to be more severe in women with higher case fatality at 1 month. 24 However, our study focused on long-term outcome of stroke in young adults, excluding mortality cases within 30 days, and a few noticeable findings emerged.
The distribution of stroke subtypes varies in different ethnicities. Asians were reported to have two times higher incidence rate of ICH than other ethnic groups. 26 The reasons for the difference in distribution of main stroke subtypes between East-Asia and Western-Europe are not fully understood. Our study showed a very high proportion of ICH (36.1%). This is understandable since a large proportion of ischemic stroke patients would encounter their index stroke beyond 50 years old. 27 Without the upper age limit of 50 years, the reported stroke subtypes frequencies in Taiwan adults were ischemic stroke 74%, ICH 16.1%, transient ischemic attack 6.7%, SAH 2.8%, cerebral venous thrombosis 0.2%, and stroke not otherwise specified 0.2%. 27 This epidemiology data was published by Hsieh and Chiou using the Taiwan Stroke registry to collect clinical data between 2006and 2008. 27
The mortality rates after a stroke event in Taiwan young adults were higher than anticipated, especially when examined over time. The first-year ischemic stroke mortality rates, which focused on young adults, all done in Western populations, varied from 4.5% to 6.3%.3,6 –11 Ours is 4.2% (Supplemental Table 1). While the average mortality rates varied between 0.6% and 1.8% during subsequent years in those studies,3,6 –11 Taiwan’s mortality rates increased at an average of 1.9% annually (Supplementary Table 1), significantly higher than any Western population. The long-term ischemic stroke mortality rates in young adults were reported at 16% (mean follow-up: 16 years) and 19.4% (mean follow-up: 11 years), 9 compared to our 25.4% (mean follow-up: 9.8 years) (Table 1). These figures from Western populations were lower than ours despite slight variations in the chosen upper age limits (range: 45–50 years) and lower age limits (range: 16–20 years).3,6 –11 In our study, the overall mortality of young ischemic stroke patients is higher and did not decrease as much over subsequent years under long-term follow-up. This is different from the findings in Western populations.3,6 –11
To generate a basic comparison between East-Asia and Western-Europe regarding stroke mortality, the Dutch study by Ekker et al. 14 was incorporated. As shown in Figure 3, all-cause stroke mortality rates were two to three times higher in Taiwan, regardless of the stroke subtypes, sexes, or specific age groups. Referring to the stroke not otherwise specified of woman in 30–39 years age group, where the difference of mortality is the least between the two countries, Taiwan has more than two times the observed deaths than the Dutch already, let alone other stroke subtypes and age groups where the difference in observed deaths can exceed three times. Although stroke-related mortality has been decreasing in East-Asian countries, including Taiwan, 28 ethnicity might play a pertinent role in determining the patient’s long-term survival after a stroke event.
Hypothesis
In recent years, increasing studies have raised concerns over dyslipidemia and intense lipid-lowering therapy, with ICH being one of the proposed adverse effects. 29 Low-density lipoprotein (LDL)-cholesterol had been shown to bear a J-shaped association with stroke events in statin-naïve young adults. 30 The lowest risk for stroke events seems to fall in the middle LDL-cholesterol range of 84–101 mg/dL. 30 Furthermore, large cohort studies have shown that lower serum LDL-cholesterol independently predicts higher mortality after acute ICH. 31
From a genetic perspective, Western populations have higher lipid profile than Han Chinese, due to the central role played by Apolipoprotein E (APOE) gene in lipoprotein metabolism and lipid transportation. The frequency of APOE*4, the ancestral allele, was 0.163 in the Dutch and 0.071 in Chinese. 32 APOE*4 is shown to be closely associated with elevated LDL-cholesterol. 33
Another major difference between Taiwanese and the Dutch, other than genetics, is the diet. Large observational epidemiological studies have shown that different diets around the world greatly impact chronic disease epidemiology and public health. 34 The Tzu Chi Health Study estimated that 30% (1484 out of 4384) of Taiwanese are vegetarians. 35 On the other hand, the Netherlands Cohort Study showed that only 0.01% (1150 out of 120,852) are self-reported vegetarians. 36
The compound effects of genetics and lifestyle would likely place Han Chinese in Taiwan and Dutch in the Netherlands at opposite ends of the lipid profile spectrum, providing a plausible explanation to the phenomenon observed in this epidemiology population-based study.
Limitations
Firstly, the most obvious limitation is the inherent flaw of a registry-based study. Relevant details such as medical images and comprehensive patient history were inaccessible. Thus, it is unable to adjust for possible confounders, like stroke severity, family history, smoking history, and medication use. Individual comorbidities could only be gauged with the CCI, although shown to be a reliable method, important information about actual risk factors, underlying etiologies, disease durations and treatments administrated were lacking. 37 There are the possible effects of underestimation of the CCI score prior to the index stroke in young adults and overestimation of the CCI score when comorbidities decrease after the index stroke until death. 14 However, this possible error is expected to be negligible because young adults are less likely to have major comorbidity compared to the elderlies. 26
Secondly, strokes between 1999 and 2012 were defined as index stroke if no earlier admissions for stroke were recorded in the NHI database. Since the NHI was initiated in 1995, 15 this is a relatively short look back time to verify history of prior stroke. This can surely lead to misclassification of some events as first-ever-stroke even if they might have had remote stroke history. However, given that this study focused on young adults, where evidence have shown only a 2% yearly risk of recurrent stroke 3 years from the index stroke, this risk of such misclassification is extremely low. 38
Thirdly, recent advancements in ischemic stroke managements, such as mechanical thrombectomy, could significantly decrease mortality after a stroke event.39,40 All these improvements to stroke diagnosis, prevention and treatments can lead to the presented data not entirely generalizable to reflect stroke outcomes today. 41
Fourthly, we were unable to generate a strict reciprocation of the Dutch study due to setbacks encountered during data extraction. The lower age limit is 20 years in this study, while the lower age limit of the Dutch study is 18 years, because the age of majority is declared to be 20-year-old under the current constitutional law in Taiwan. Thus, extracting data in the 18–20 years age group from the NHIRD would require filing a separate study notion, since they are not lawfully defined as adults. Additionally, as much as we would like to compare the mortality in our study with healthy matched individuals from the Taiwan general population to gain an overview of the baseline mortality risk, as performed in the Dutch study, we were regretful to be informed that provision of specific ICD codes is the essential first step when filing a study notion with the NHIRD. We were unable to provide a comprehensive presentation of the expected matched mortality rates in Taiwan stroke-naïve individuals.
Finally, the data extracted from the Dutch study by Ekker et al. served only to provide a crude visualization of the impact of ethnicities on young adult mortalities after a stroke event. Individual-patient level analysis that adjust for important differences between people in the two countries are not available in these registry-based data, thus, unable to draw extensive comparative conclusions.
Conclusions
This epidemiology study on nationwide scale authenticated the fact that age, sex, stroke subtypes, and ethnicities played pertinent roles in long-term mortality of young adults after a stroke incidence. Han Chinese in Taiwan have a very high risk of mortality after experiencing a first stroke event. This risk does not diminish over the years. Taiwanese have higher long-term mortality when compared to similar Western Europe study. More cross-ethnicity studies on stroke populations are warranted to shed more light on this matter.
Supplemental Material
sj-docx-1-eso-10.1177_23969873221115268 – Supplemental material for Long-term mortality rates of young stroke in Taiwan: A decade-long epidemiology population-based study
Supplemental material, sj-docx-1-eso-10.1177_23969873221115268 for Long-term mortality rates of young stroke in Taiwan: A decade-long epidemiology population-based study by Yu-ying Wu, Po-Yuan Chen, Cheng-Chun Wu, Han-Jung Chen, Cheng-Loong Liang, Yi-Che Lee, Chi-Wei Lin, Chao-Ming Hung, I-Fan Lin and Hao-Kuang Wang in European Stroke Journal
Footnotes
Acknowledgements
This study was supported by Center for Database Research, E-DA HEALTHCARE GROUP. We are also grateful to Giant Data Technology for providing administrative and technical support.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported in part by grants from the E-Da hospital, Kaohsiung, Taiwan (EDAD-2020-024).
Ethics approval and consent to participate
We analyzed administrative claims data obtained from Taiwan’s National Health Insurance Research Database (NHIRD). The details of database generation, monitoring, and maintenance are published online by the Taiwan NHRI (
). Because the National Health Insurance (NHI) program is the sole payer in Taiwan, the data represents the original medical claims for all islanders covered by the NHI program, and are distributed by sex, age, or amount of average payroll-related insurance payments. The subjects of this study were recruited by reviewing monthly patient discharge data released by NHIRD, and data in the NHIRD that could be used to identify patients is scrambled before being sent to the NHRI for database construction. Because the analysis was limited to aggregate secondary data that could not be used to identify the patients, the study was approved by the NHRI Ethics Review Committee, Taiwan. The study protocol conformed to the ethical standards established by the Declaration of Helsinki (1964), which do not require written or verbal consent for data-linkage studies.
Guarantor
HKW
Contributorship
YYW, PYC, IFL, and HKW conceptualized the study. YYW, IFL, and HKW analyzed the data. PYC, CCW, HJC, CLL, YCL, CWL, and CMH interpreted the data. CCW had a major role in acquisition of the data. YYW wrote the first draft of the manuscript. All authors reviewed and edited the manuscript and approved the final version of the manuscript.
Data sharing statement
This retrospective, population-based cohorts were selected from among patients registered in the National Health Research Institute (NHRI) Database, which was released for research purposes in 2008. The details of database generation, monitoring, and maintenance are published online by the Taiwan NHRI ().
Supplemental material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
