Review
Registry Studies of Stroke in Japan
2023 Volume 30 Issue 9 Pages 1095-1103
Details
2023 Volume 30 Issue 9 Pages 1095-1103
Recently, the Cerebrovascular and Cardiovascular Disease Control Act was enacted, for which it was necessary to establish a comprehensive and accurate nationwide database and promote rational and economical stroke countermeasures in Japan, thus serving the public interest. Among the many studies on stroke registries, the Fukuoka Stroke Registry, a regional cohort, provides highly accurate information, and the Japanese Stroke Data Bank, a nationwide cohort, is highly comprehensive. The findings of these studies have contributed to the construction of evidence and the establishment of guidelines for stroke management. In the Nationwide survey of Acute Stroke care capacity for Proper dEsignation of Comprehensive stroke CenTer in Japan, research on improving the quality of medical care to close the gap between guidelines and clinical practice was performed using electronic medical records. This has enabled the recommendation of medical policies in Japan by visualizing medical care. In the era of healthcare big data and the Internet of Things, plenty of healthcare information is automatically recorded electronically and incorporated into databases. Thus, the establishment of stroke registries with the effective utilization of these electronic records can contribute to the development of stroke care.
Stroke is the second leading cause of death and the third most common cause of disability and death combined worldwide, according to WHO’s 2019 Global Health Estimates. In Japan, it is the fourth leading cause of death and the second most common disease that requires nursing care; its incidence is higher in Japan than in Western countries1). Stroke management involves extensive healthcare resources and manpower, including acute inpatient treatment, treatment for prevention of stroke recurrence after discharge, rehabilitation, recovery, and social support for sequelae. The burden of stroke is significant, and its impact on the society and economy is substantial.
In December 2018, the “Cerebrovascular and Cardiovascular Disease Promotion Act” was established to develop preventive measures, disseminate accurate information, and provide service systems related to medical care, health, and welfare for patients with cerebrovascular and cardiovascular diseases. One of these initiatives is promoting research on cerebrovascular and cardiovascular diseases to establish a comprehensive and accurate nationwide database and to promote rational and cost-effective stroke strategies for public interest. Recent developments in information processing technology and the widespread use of electronic medical records have facilitated the statistical analysis of large-scale clinical data and have enabled the establishment of large-scale disease registration studies in Japan2). However, certain concerns remain, such as the operation of complex systems and the reliability of privacy and funding. In this article, we review studies on stroke registries conducted in Japan.
What is a Stroke Registry?A clinical registry is an observational database focusing on clinical conditions, procedures, therapies, or populations. A stroke registry is defined as “an organized system for the collection, storage, retrieval, analysis, and dissemination of information about individuals who have experienced a stroke”3, 4), and it aims to determine the course and prognosis of the disease.
We conduct stroke registration studies to understand the pathophysiology of stroke, its epidemiology, and the effectiveness of treatment by collecting and analyzing observational data, such as patients’ clinical information and outcomes, in order to promote medical advancements. With the development of epidemiology, multicenter registry studies have increased and evidence has contributed to establishing guidelines for stroke management and practicing evidence-based medicine (EBM) worldwide. Recently, studies on healthcare quality improvement have been conducted to determine whether adherence to these guidelines through EBM practices contributes to improving medical care quality. Adherence to these guidelines should improve the quality of medical care and benefit the patients. Clinical registration plays a central role in improving patient care3-5) (Fig.1). Thus, a stroke registry helps in improving patient outcomes, stroke care quality, and healthcare policies through the publication and communication of results6). Ideally, a stroke registry should be nationwide and include patients from as many hospitals as possible to increase representativeness and avoid selection bias6).
In Japan, these studies exist at both the regional and national levels. Based on a systematic review by Sato et al.7), a list of studies in Japan is presented in Table 1. This list includes studies limited to patients with stroke who underwent thrombolytic therapy or thrombectomy, such as the Stroke Acute Management with Urgent Risk-factor Assessment and Improvement study8), Kanagawa Intravenous and Endovascular Treatment for Acute Ischemic Stroke registry9), and Japanese Registry of NeuroEndovascular Treatment10).
| All stroke patients | |
|---|---|
| regional | Akita Stroke Registry44) |
| Fukuoka Stroke Registry (FSR)14) | |
| Hypertension, cerebral Amyloid, aGe Associated Known neuroimaging markers of cerebral small vessel disease Undertaken with stroke Registry prospective observational study (HAGAKURE)45) | |
| Iwate Stroke Registry46) | |
| Kyoto Stroke Registry47) | |
| Kyurashiki Stroke Registry48) | |
| Progressive Linkage Utilizing Kumamoto Seamless sTroke REferral Associates for CVD aMelioration database (K-PLUS)49) | |
| Shiga Stroke Registry50) | |
| South Tochigi Acute Ischemic Stroke Registry51) | |
| Stroke Registry in Miyagi Prefecture52) | |
| Stroke Registry in Shibata City, Niigata Prefecture53) | |
| Stroke Registry in Yamagata Prefecture54) | |
| STrOke Registry of Yokohama (STORY)55) | |
| Takashima Stroke Registry56) | |
| Toyama Stroke Registry (TOY STORE)57) | |
| national | Japan Multicenter Stroke Investigators’ Collaboration (J-MUSIC)58) |
| Japanese Standard Stroke Registry Study (JSSRS)59) | |
| Japan Stroke Data Bank (JSDB)27) | |
| Nationwide survey of Acute Stroke care capacity for Proper dEsignation of Comprehensive stroke CenTer in Japan (J-ASPECT)2) | |
| Limited patients with acute reperfusion therapy | |
| rt-PA | Stroke Acute Management with Urgent Risk-factor Assessment and Improvement (SAMURAI) rt-PA Registry8) |
| EVT | Kanagawa intravenous and endovascular treatment for acute ischemic stroke registry (K-NET)9) |
| Japanese Registry of Neuroendovascular Therapy (JR-NET)10) | |
| Recovery by Endovascular Salvage for Cerebral Ultra-acute Embolism (RESCUE)- JAPAN registry60) |
EVT: endovascular treatment.
In most studies, data were collected through individual questionnaires. Although this design allowed researchers to set specific items according to the study objectives and enabled detailed data collection, some problems in data input effort and cleaning were observed. In small-scale regional registration studies, the accuracy of data input may increase; however, generalizability might be limited. In contrast, in large-scale national registration studies, data input is delegated to each facility, resulting in a significant burden on input operators and the need for the quality control of input data. Therefore, appropriate data structures and governance policies are required to maintain a sustainable and well-operating nationwide stroke registry4).
Recently, advancements in information technology have led to the electronification of medical information, enabling the automatic extraction of data from existing databases, such as public insurance ones. Although the use of existing data reduces the flexibility of research, combining both approaches ensures generalizable and highly accurate registration studies. Notably, the Riks-Stroke Registry in Sweden11), launched in 1994, has covered all hospitals in the country that have been accepting patients with acute strokes since 1998. It has registered over 90% (approximately 25,000 patients per year) of stroke cases and collected information using individual questionnaires. The follow-up rate within the Riks-Stroke group was approximately 90% at 3 months and 75%–80% at 1 year. By linking data to official death registrations using individual identification numbers, information on mortality and causes of death was collected for all patients. Such linkage has been established not only in Sweden but also in other countries such as Finland, Denmark, and Korea.
Fukuoka Stroke Registry (FSR)The FSR is a regional cohort, multicenter prospective registration study comprising seven stroke centers in Fukuoka Prefecture. Between 2007 and 2019, approximately 20,000 patients hospitalized within 7 days from the onset of an acute stroke were registered. One notable feature was the accuracy of the data: the items were standardized across the participating facilities and data were input by specialized clinical research coordinators who had received training. Therefore, there were minimal missing data in the basic dataset. Additionally, the consent rate for study participation, which is crucial for the validity of cohort studies, was high (89%). Another distinguishable feature was the collection of long-term follow-up information through telephone surveys, with a follow-up rate of 85%. Furthermore, serum and genomic samples were collected from most participants.
The FSR study has reported numerous research findings, including independent factors associated with clinical outcomes 3 months after stroke onset, such as the presence of high blood pressure12), blood pressure variability after admission13), prestroke poor glycemic control14), chronic kidney disease15), prestroke statin use16), insulin resistance17), impaired insulin secretion18), high-sensitivity C-reactive protein19), and smoking20). Moreover, the impact of the time from symptom onset to hospital arrival on the functional outcomes 3 months after stroke onset was investigated. Compared with patients who arrived within 24 h from onset, those who arrived within 6 h showed a significantly increased odds ratio (OR) (95% confidence interval [CI]) for favorable outcomes even after multivariate adjustment (within 1 h: 2.68 [2.05–3.49]; 1–2 h: 2.10 [1.60–2.77]; 2–3 h: 1.53 [1.15–2.03]; and 3–6 h: 1.31 [1.05–1.64]). Early hospital arrival within 6 h from onset was associated with favorable outcomes after an ischemic stroke (IS), regardless of reperfusion treatment or stroke severity.
Whether kidney dysfunction (decreased estimated glomerular filtration rate [eGFR]), kidney damage (proteinuria), or both are associated with long-term outcomes after an IS was investigated21). During the median follow-up of 4.3 years (interquartile range: 2.1–7.3 years), 2481 patients experienced recurrent stroke (48.0/1000 patient-years) and 4032 patients died (67.3/1000 patient-years). Chronic kidney disease was independently associated with an increased risk of stroke recurrence and all-cause death even after adjusting for multiple confounding factors, including traditional cardiovascular risk factors. Both eGFR and proteinuria were independently associated with increased risks of stroke recurrence (multivariate-adjusted hazard ratio [95% CI], eGFR ＜45 mL/min/1.73 m²: 1.22 [1.09–1.37] vs. eGFR ≥ 60 mL/min/1.73 m²; proteinuria ≥ 2+: 1.25 [1.07–1.46] vs. proteinuria negative) and death (eGFR ＜45 mL/min/1.73 m²: 1.45 [1.33–1.57] vs. eGFR ≥ 60 mL/min/1.73 m²; proteinuria ≥ 2+: 1.62 [1.45–1.81] vs. proteinuria negative). Kidney dysfunction and damage were independent but associated with an increased risk of recurrent stroke and all-cause death.
Moreover, sex differences between the groups were assessed. Although women were found to have poor functional outcomes at discharge (OR [95% CI]: 1.30 [1.08–1.57])22), they also had a lower risk of 30-day mortality after stroke (hazard ratio [95% CI]: 0.76 [0.62–0.92])23). Furthermore, women had a higher risk of functional dependence 5 years after stroke (multivariate-adjusted OR [95% CI] vs. that of men: 1.56 [1.26–1.93])24). Therefore, sex-specific strategies must be considered.
In addition, the association between stroke onset and environmental factors was investigated25, 26). Windblown sand dust originating from mineral soil in the deserts of China and Mongolia was significantly associated with the incidence of atherothrombotic brain infarction after adjusting for expected confounders, including meteorological variables and other air pollutants. Moreover, short-term exposure to suspended fine particles matter 2.5 was significantly associated with stroke occurrence.
Although the findings of this regional cohort might have limited generalizability, the use of highly accurate data provides novel insights and answers to clinical questions.
Japan Stroke Data Bank (JSDB)The JSDB was initiated as the Japanese Standard Stroke Registry Study in 1999. It was the first nationwide study to unify disease classification, severity scales, prognostic evaluation, and other parameters related to stroke. It collects clinical data using individual patient information such as patient characteristics, examinations, treatments, and stroke outcomes. The results are regularly published as highly representative national-level data. In July 2021, 132 hospitals participated in the JSDB, and approximately 240,000 patients with acute IS (AIS), intracerebral hemorrhage (ICH), subarachnoid hemorrhage (SAH), and transient ischemic attack have been enrolled27).
Recently, a study has reported changes in stroke severity and outcomes within 7 days from onset, including IS and hemorrhagic stroke, based on 20-year data from the JSDB between January 2000 and December 2019 28). This study included 183,082 patients (135,268 with IS, 36,014 with ICH, and 11,800 with SAH). Multivariate analysis revealed that over the past 20 years, the median age at stroke onset increased for all three stroke subtypes, whereas the National Institutes of Health Stroke Scale scores for IS and ICH and the World Federation of Neurosurgical Societies scores for SAH decreased, indicating a trend toward milder cases over time. The proportion of good outcomes increased over the years after adjusting for age in patients with AIS; however, it did not increase over time in patients with ICH and SAH. This suggests that the effects of reperfusion therapy are reflected in the outcomes of patients with AIS, whereas the lack of effective treatments for ICH and SAH is equivalent to the lack of reperfusion therapy for AIS. The proportion of patients with hemorrhagic stroke may not have decreased because of the widespread use of anticoagulants prior to stroke onset. Thus, the long study duration and large population size of the JSDB are useful for clarifying the trends of stroke outcomes in Japan.
Another study has reported the differential impact of renal dysfunction on stroke outcomes depending on clinical subtypes29). A low GFR (＜40 mL/min/1.73 m²) was significantly associated with poor outcomes in small vessel occlusion and cardioembolic stroke; however, in cardioembolic stroke, higher eGFR was significantly associated with poor outcome. Additionally, proteinuria (dipstick proteinuria ≥ 1) was independently associated with poor outcomes in both the small vessel occlusion and cardioembolic stroke subtypes. Therefore, renal impairment may have different predictive values for outcomes in patients with both subtypes.
Nationwide Survey of Acute Stroke Care Capacity for Proper dEsignation of Comprehensive Stroke CenTer in Japan (J-ASPECT)The J-ASPECT is a large-scale database study that started in 2010 with the automatic extraction of Diagnosis Procedure Combination (DPC) information and electronic medical records over the past 11 years (940 facilities, approximately 6.63 million DPC data of approximately 1.54 million stroke cases)2). The DPC has been developed as a measurement tool to standardize, evaluate, and improve the quality of healthcare in Japan and to clarify the content of acute-phase hospital care30). The use of DPC data has allowed for the collection of a large amount of patient information using existing extensive data sources. Moreover, the extraction method has enabled continuous collection of information with minimal effort, ensuring complete coverage and continuity. Similar to the Get With The Guidelines® - Stroke program in the United States, J-ASPECT has conducted various studies on the quality of care to close the gap between guidelines and clinical practice.
First, quality indicators (QI) for primary and comprehensive stroke care have been developed and evaluated using these indicators31, 32). Facilities in the top 20% of the Comprehensive Stroke Center score comprising 25 recommended requirements had a 26% lower in-hospital mortality rate compared with that of facilities in the bottom 20%, revealing that structured indicators are associated with clinical outcomes33). Additionally, compliance rates with stroke care QIs were investigated, showing temporal changes around the time of guideline revision for thrombectomy therapy for acute intracranial large artery occlusion. The impact of evidence-based QIs on outcomes was examined for the first time at the national level34). Among the 20 QIs evaluated for their association with clinical outcomes, 14 indicators were significantly associated with a decrease in in-hospital mortality rate (OR [95% CI]), such as the achievement of intravenous recombinant tissue plasminogen activators (rt-PA) therapy within 60 min from hospital arrival (0.80 [0.69–0.93]), initiation of thrombectomy therapy within 90 min from hospital arrival (0.80 [0.67–0.96]), and achievement of successful reperfusion (0.40 [0.34–0.48]), whereas 11 indicators were associated with increased functional independence at discharge. Continuous improvements have been achieved in the time to thrombectomy and the rate of successful reperfusion since 2015, suggesting the achievement of technical proficiency and standardization of thrombectomy therapy. Furthermore, evaluation of the quality of care for patients with AIS treated with reperfusion therapy (intravenous rt-PA and endovascular thrombectomy) revealed an insufficient achievement of the goals in six items, including the evaluation of cerebral vascular imaging within 30 min from hospital arrival, rt-PA administration within 60 min, appropriate statin administration at discharge, and prevention of deep vein thrombosis35).
Thus, this study not only allows the continuous evaluation of the quality of stroke care but also provides evidence necessary for improving the healthcare system in Japan.
Nationwide Health Insurance Claims DataRecently, studies have been conducted worldwide using insurance claims data. In Japan, a national health insurance system has been established, and the National Database of Health Insurance Claims and Specific Health Checkups (NDB), based on insurance claims data for the entire population, is available. It can be used extensively in various types of clinical research, including studies on disease risk factors, outcome research, healthcare utilization, and prescription patterns. The NDB is a large-scale nationwide database on health insurance claims data covering almost the entire Japanese population (125–126 million people). Such nationwide administrative claims databases are generally considered to have large sample sizes and high representativeness. In addition, they contain continuous longitudinal data, thus being suitable for longitudinal outcome studies. However, administrative claims data are collected for insurance reimbursement purposes rather than for research purposes. Therefore, the NBD data may not necessarily include all the information that researchers need. For example, important clinical information such as the onset date and severity of stroke may be lacking. Therefore, when conducting stroke research based on insurance claims, it is necessary to thoroughly consider the validity of the research findings.
We examined nationwide trends in the implementation status of reperfusion therapy and the 30-day mortality rate after reperfusion therapy in 69,920 patients with AIS who received thrombolysis or endovascular therapy using the NDB36) and found that the number of patients with AIS who were eligible for reperfusion therapy increased because of the extension of the time window for intravenous rt-PA therapy and the dissemination of endovascular therapy. These trends did not result in an increased risk of postreperfusion mortality. Furthermore, we investigated regional disparities in the implementation rate of reperfusion therapy and the regional factors associated with its implementation in patients with AIS37). Over time, the regional disparities improved, and the density of stroke centers and interventional neurologists, as well as market concentration, were associated with an increased utilization rate of reperfusion therapy, whereas the proportion of rural residents and delays in ambulance transportation were negatively associated with the utilization rate.
Future PerspectivesAs mentioned previously, registries based on the collection of individual patient information have certain limitations. In this rapidly advancing era of electronic information utilization, studies based on existing data sources such as claims data are expected to expand. Ideally, it is desirable to link registry databases, which include rich and detailed clinical information, with electronic medical records and long-term outcome databases.
The Canadian Stroke Network Registry Study (RCSN) is a useful example38). It included patients who provided informed consent, which led to a selection bias and threatened the validity of the registry-based study39). In addition, the study increased expenses and led to the loss of follow-up.
However, under the new healthcare privacy legislation in Canada, the RCSN was designated as a “prescribed registry.” This permitted it to operate the registry without the consent of patients and accordingly to conduct follow-up surveys using population-based administrative databases through record linkage with unique identifiers40). Similarly, the use of record linkages in Japan’s JSDB and J-ASPECT aims to visualize stroke care.
However, achieving data linkage is difficult as several concerns exist, such as understanding personal-information protection and constructing the data linkage system.
In the era of big data and the Internet of Things, numerous parameters are automatically recorded and incorporated into databases. Medical data often include unstructured records; however, the use of natural language processing tools enables the extraction of valuable information from such unstructured records41). Therefore, collaboration between healthcare professionals and information technology experts is required42).
In the future, establishing a multinational network of stroke registries will be beneficial to enable the generalization of research findings, provide clear insights into differences such as racial disparities, and contribute to a better understanding and development of stroke care4, 43).
This review discusses the status of stroke registry studies in Japan. Data input involves both manual collection through individual forms and automatic collection through electronic medical records. As healthcare digitization progresses, the linking of stroke registries with electronic information is expected in Japan.
None.
