Epidemiology of Moyamoya Disease in Taiwan
A Nationwide Population-Based Study
Background and Purpose—Previous studies have shown regional and temporal variations in epidemiological features of moyamoya disease, but population-based studies in regions other than Japan are limited. We investigated the incidence and patients characteristics of moyamoya disease during 12 years in Taiwan using claims databases of a universal health insurance system.
Methods—From the inpatient databases of the Taiwan National Health Insurance program, we identified subjects who had an initial hospitalization with moyamoya disease and had been underwent cerebral angiography as incidence cases during 2000 to 2011. The incidence and the patient characteristics were described by age and time periods of the hospitalization.
Results—During the 12-year period, 422 patients were identified, representing an annual incidence of 0.15 per 100 000 person-years. Adults exhibited an upward trend in incidence with an incidence rate ratio of 1.74 (95% confidence interval [1.17–2.58]) in years 2010 to 2011 comparing with years 2000 to 2001. However, children had a decreased incidence except a slightly increase in the last 2 years. Compared with patients hospitalized during 2000 to 2005, patients identified during 2006 to 2011 had greater women-to-men ratio (1.7 versus 1.1, P=0.048). Children were more likely to have comorbid epilepsy than were adult patients (25.0% versus 3.4%, P=0.002). Hemorrhagic stroke was rare among pediatric patients but presented more frequently in adults. However, ischemic stroke was more prevalent in both groups.
Conclusions—The incidence of moyamoya disease has increased in adults but not in children from 2000 to 2011 in Taiwan. Sex ratio and comorbid conditions differed by age and study period.
Moyamoya disease is a rare cerebrovascular disorder of unknown pathogenesis. It is characterized by progressive occlusion of large intracranial vessels around the bifurcation of internal carotid artery, and development of collateral circulation from superficial scalp arteries, meningeal arteries, and deep perforating arteries. Patients with moyamoya disease often have transient ischemic attacks and stroke, and may have debilitating neurological sequelae or cognitive impairment.
Epidemiological features of moyamoya disease have been reported in various regions during the past 3 decades. The observations have shown regional1–3 and ethnic differences4,5 in incidence, prevalence, and patient characteristics. Furthermore, temporal variations in characteristics of moyamoya disease have been observed in nationwide surveys in Japan. Compared with surveys in 1990s, the prevalence and incidence rate in 2000s has doubled, and the onset age was older.6–8
Studies depicting characteristics of a disease and its temporal trends in a given population would aid in generating hypothesis for pathogenetic studies, healthcare resource allocations, and evaluation and improvements in healthcare practices. Several studies have reported moyamoya disease in Asia, particularly in Japan,6–8 South Korea,9,10 and China,11,12 where the incidence and prevalence were recognized high. However, most of these studies, except those from Japan, were performed a decade ago or in a single center or selected hospitals, and may be vulnerable to selection bias.
Taiwan is a Southeast Asian country where people are predominantly ethnic Chinese. The National Health Insurance (NHI) program of Taiwan provides universal health coverage to the population since its inception in 1995.13 The size and representativeness of the research database of NHI allows for study of rare diseases and population-based observations on temporal changes in epidemiological features. Using the database, we documented the incidence and characteristics of patients with moyamoya disease and observed their changes during 12 years in Taiwan.
We used multiple data sets distributed by the National Health Research Institute, which has collaborated with the NHI Administration to construct the NHI research database. The data sets of inpatient expenditures contain inpatient claims of the 23 million NHI beneficiaries that represent >99% of the population. Information included were date of admission, date of discharge, diagnostic codes (up to 5), and procedures (up to 5) for all hospital admissions. The files of inpatient orders contain claims for inpatient services, and registry for beneficiaries contains limited demographics such as sex, date of birth, and date of enrollment in and withdrawal from NHI. All data sets can be linked by using a unique individual identification number, which were scrambled before the release of data sets to protect patient privacy. The Institutional Review Board of the National Taiwan University Hospital approved this study.
Within inpatient claim data sets from 1996 to 2011, we identified 627 patients with an initial hospital admission with moyamoya disease between 2000 and 2011, based on a discharge diagnosis coded with International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 437.5. In an attempt to reduce the possibility of including prevalent cases, we used a wash-out period of ≥4 years. Patients were excluded if they had a hospital admission with moyamoya disease before their hospitalization during 2000 to 2011, identified by retrospectively searching claims back to 1996. Patients also were excluded if they had been diagnosed with conditions associated with moyamoya syndrome (n=13), including neurofibromatosis type 1 (ICD-9-CM 237.71), cranial irradiation (ICD-9-CM v15.3), Down syndrome (ICD-9-CM 758.0), and sickle cell anemia (ICD-9-CM 282.6), or if they did not undergo conventional cerebral angiographic study (n=192).
Demographics and Comorbid Conditions
The inpatient data sets were searched for records with discharge diagnosis of conditions presented during or within 3 years before the hospitalization with moyamoya disease, including ischemic stroke (ICD-9-CM 433.xx-437.xx), hemorrhagic stroke (ICD-9-CM 430.xx-432.xx), and epilepsy (ICD-9-CM 345.xx, 780.3, 779.0). To observe the use of surgical treatments, claims records of extracranial-intracranial revascularization were also searched (ICD-9-CM procedures code 39.28, 39.29, 39.31).
We calculated the annual incidence rate with the first hospital admissions with moyamoya disease divided by the number of subjects insured in the NHI program in each year during 2000 to 2011. Age standardization was performed using the direct method to the population insured in year 2000 in age bands of 5 years. The incidence was also estimated by age of the hospitalization with moyamoya disease (age of diagnosis), to compare whether there are differences in incidence trends between children (<18 years) and adults (≥18 years). The confidence intervals (CI) of incidence were estimated by assuming Poisson distribution, and incidence rate ratios and 95% CIs were estimated using Poisson regression with the incidence of years 2000 to 2001 as the reference. To observe peak of incidence in age distribution, the annual average age-specific incidence rate during the 12-year period was estimated in 5-year age categories.
To assess whether there are temporal variations of characteristics of moyamoya disease, we evaluated patient characteristics at the hospitalization, including sex, age, and medical conditions, survival status at discharge and the surgical treatment, extracranial-intracranial revascularization, by year of the hospitalization in two 6-year categories, 2000 to 2005 and 2006 to 2011. The differences between the 2 periods were tested by using χ2 test or Fisher exact test. We also compared the characteristics of patients across 4 age groups, that is, 0 to 6, 7 to 12, 13 to 18, and ≥18 years, using χ2 tests or Monte Carlo Estimate for the Exact Test. Hemorrhagic stroke and ischemic stroke were identified for children and adults by year of their hospitalization. We performed all analyses using SAS software, version 9.3 (SAS Institute Inc), and P values were calculated with 2-tailed tests.
Incidence, the Trend, and Prevalence
From 2000 to 2011, we identified 422 subjects as incidence cases of moyamoya disease from the initial hospitalization; the annual average incidence rate was 0.15 (95% CI [0.14–0.17]). The age-standardized incidence rate increased from 2000 through 2011 by 44% in all subjects (annual incidence per 100 000, 0.14–0.20), 40% in women (0.18–0.25), and 48% in men (0.10–0.15; Figure 1). The overall incidence rate was higher in pediatric population than in adult population (Table 1). However, adults exhibited an upward trend in incidence for the years 2000 through 2011, but the rates fluctuated among pediatric population. Using the incidence of years 2000 to 2001 as the reference, the incidence rate ratios of years 2010 to 2011 were 1.16 (95% CI [0.66–2.04]) for subjects <18 years and 1.74 (95% CI [1.17–2.58]) for those aged ≥18 years.
We calculated the prevalence of moyamoya disease in 2011. Among 422 patients with moyamoya disease, 384 subjects who were alive and remained enrolled in NHI by December 31, 2011, were the prevalent cases as of the end of 2011. Prevalence was calculated by dividing the number of prevalent cases in 2011 by number of subjects insured by NHI in the same year. The resulting estimate was 1.61 per 100 000.
The incidence rates differed by age, with the highest peak at 5 to 9 years in men and 10 to 14 years in women and the second peak at ages 40 to 44 years in both sexes (Figure 2A). When the study duration was split into 2, the highest peak at 5 to 9 years of age was observed for both men and women who had their first hospitalization with moyamoya disease between 2000 and 2005 (Figure 2B). In the time period of 2006 to 2011, the highest peak was observed at 5 to 9 years in men. In women, the incidence fluctuated with age, with the highest incidence appeared at 10 to 14 years (Figure 2C).
Demographics and Comorbid Conditions Across Age Groups and Time Periods
Of all patients, women accounted for 58.8% (sex ratio, 1.4; Table 2). The mortality rate at discharge was 3.6%. Ischemic stroke and hemorrhagic stroke had been diagnosed in 46.7% and 26.5% patients, respectively. Extracranial-intracranial revascularization was performed in 37% patients. Compared with patients with hospitalization between 2000 and 2005, women accounted for a higher percentage in the period of 2006 to 2011 (sex ratio 1.1 versus 1.7, P=0.048). Patients of the later period were older (P=0.002) and more likely to have ischemic stroke (P=0.007). Mortality at discharge decreased by half in the period of 2006 to 2011 (5.1% versus 2.4%), but the difference was not statistically significant (P=0.14).
Patients aged ≤18 years were 3 times more likely to have epilepsy than were adult patients (P=0.003, Table 3). Patients aged 0 to 6 years had the highest extracranial-intracranial revascularization rate (75.7%), followed by patients in the 7- to 12- (68.9%), 13- to 18-year-old group (51.7%), and the lowest in those aged ≥18 years (24.5%; P<0.001).
The proportion of presence of ischemic stroke was higher than that of hemorrhagic stroke in both children and adults (Table 3 and Figure 3). Of patients aged <18 years, 24.0% had ischemic stroke and 4.0% had hemorrhagic stroke. The corresponding figures for adults were 56.2% and 36.0%. The proportion of ischemic stroke in adults has increased since 2004. Adults were more likely to have both ischemic stroke and hemorrhagic stroke. Among patients aged ≥18 years, 9.4% (28 cases) had both subtypes of stroke; the corresponding figures were 6.5% (2 cases) for patients who were 13 to 18 years old, 0% for those were 7 to 12 years old, and 3% (1 case) for those who were ≤6 years old (P=0.053, data not shown).
Several studies have reported epidemiology of moyamoya disease and shown variations in incidence and characteristics of patients (Table in the online-only Data Supplement).4,6–8,12,14–17 Explanations for the discrepancy may include the differences in survey methods and the studied regions, ethnic groups, and time periods. In general, studies from regions other than Asia yielded lower incidence. However, our observation showed that the incidence of moyamoya disease in Taiwan was far below that of Japan, Korea, and Nanjing, China, although the incidence increased in the last 12 years. The reason is not known, and future studies are needed to address the role of genetic and environmental factors in the pathogenesis of moyamoya disease. The incidence estimated in the present study is 7 to 8 times higher than that (0.02 per 100 000) revealed by a survey across regional medical centers in the late 1990s.14 Less awareness of moyamoya disease in the medical profession as well as the society may partly explain the low incidence in early years.
In line with our observations, previous investigations have shown increased incidence of moyamoya disease over time. In Japan, the incidence (or detection rate) increased from 0.35 per 100 000 in 1994,8 0.54 per 100 000 in 20036 to 1.13 per 100 000 in 2006.18 An upward trend was also observed in a study in Korea, which revealed an increase of 15% of new cases per year during 2004 to 2008.15 We are unaware of any other population-based studies reporting temporal changes in incidence. In our analysis, adult patients exhibited increased incidence over time, whereas children seem to have decreased incidence except a slightly increase in 2010 to 2011. Kuroda et al19 have found that radiological findings detected abnormal conditions in asymptomatic patients with moyamoya disease, of which the prevalence may be higher than previously recognized. The increasing incidence may reflect increased detection of the disease resulting from the widespread use of noninvasive diagnostic techniques in adults.2 Decreased incidence of moyamoya disease in pediatric population has been reported previously, but the reasons remain unclear.2
Our analysis showed differences in patient characteristics between patients who had first hospitalization with moyamoya disease in time periods of 2000 to 2005 and 2006 to 2011. The ratio of women to men has been reported in many regions (Table in the online-only Data Supplement), but none of these studies included data on temporal changes in sex ratio within the same population in the same study. The overall women-to-men ratio in our analysis was similar to the observations of an earlier study in Taiwan (1.3)14 and was much higher in patients hospitalized during 2006 to 2011 (1.7) than during 2000 to 2005 (1.1). Our analysis also revealed increased incidence in ≥10-year-old subjects in the later period, and the highest peak of age distribution increased in women. Baba et al7 has reported a similar pattern, suggesting that the highest peak of incidence shifted from children to adult, and the higher peak in adults than in children was more pronounced in women. The increase in percentage of adults in the latter period may partly explain the increased prevalence of comorbid ischemic stroke in this period. Additional studies on temporal changes of characteristics may help confirm the observations.
Previous studies have reported differences in clinical presentations between children and adults. The majority of pediatric patients present with ischemic attacks, whereas adults present with both hemorrhagic and ischemic events at almost equal proportion.1,2,20 The present study showed that both children and adult patients have more ischemic strokes than hemorrhagic events, although the ratio of hemorrhagic stroke to ischemic stroke was greater in adults. We also found that children were more likely to have comorbid epilepsy. Routine use of magnetic resonance angiography as a noninvasive screening tool could help detect moyamoya disease in children <12 years old who had epilepsy. This strategy may improve the detection rate of moyamoya disease in Taiwan.
Comparison on clinical presentations across studies needs to take into account the differences in survey methods. Using Nationwide Inpatient Sample (NIS) database of the United States, Starke et al21 identified 2280 patients hospitalized for moyamoya disorder between 2002 and 2008. Of these patients, ≈70% did not have a diagnosis of stroke. This is in stark contrast to the result of studies performed using patient registry or case series based on hospital records. Hoshino et al20 studied 941 Japanese patients with moyamoya disease from the patient registry of Research Committee on Moyamoya Disease, whose clinical presentations were transient ischemic attack in 46%, infarction in 20%, hemorrhage in 21%, headache in 6%, epilepsy in 4%, and asymptomatic in 3%. Scott et al22 reported 143 patients with moyamoya treated in the Boston Children’s Hospital and found presentation of stroke in 67.8%, transient ischemic attack in 43.4%, hemorrhage in 2.8%, and asymptomatic in 4.2%. These observations revealed that studies based on insurance claims database may underestimate the incidence of stroke because of coding procedures or other methodological issues. In our analysis of the NHI database of Taiwan, 46.7% of moyamoya cases had ischemic stroke 26.5% had hemorrhagic stroke, and 10.7% had epilepsy. Although the proportion of hemorrhagic stroke is similar to that reported in Japanese series, the proportion of ischemic stroke is lower. The latter could be partly because of patients with transient ischemic attack were listed as moyamoya disease alone.
This study has several limitations, primarily because of the use of administrative databases. First, misclassification of the disease may occur because the accuracy of diagnostic codes for identifying patients with moyamoya disease is not yet validated. Although we included only patients who underwent conventional cerebral angiography, the imaging results were not available in the databases for us to confirm the diagnosis of moyamoya disease. In addition, patients may have not been identified, if diagnostic codes of other events such as ischemia or hemorrhage were coded instead of that of moyamoya disease. These could lead to underestimate of the incidence of the disease. Second, our analyses did not include patients whose diagnosis was made solely with magnetic resonance angiography. This may cause underestimation of the incidence. Third, patients with moyamoya disease treated only in outpatient clinics were not included. Lack of outpatient claims can also result in underestimating both the incidence of moyamoya disease and the presence of the comorbid conditions. Fourth, the claims data include healthcare information for subjects with symptoms who seek care. Thus, the incidence in this study is actually the detection rate. Fifth, we were unable to capture hospitalizations earlier than 1996 because the claims database was not available until the year. A patient who had been admitted to hospital before 1996 may have been classified as a new case in the period of 2000 to 2011 if he or she had a recurrent admission. The incidence in the earlier years of our analysis, and the incidence of adults who may have had incident events in their childhood or an early age, may have been overestimated. Finally, in analyses of administrative claims database, it is difficult to identify clinical presenting symptoms or complications accurately because information on clinical symptoms is limited and might be underreported. The inherent limitation of claims database should be taken into account when interpreting results of our analyses and comparing them with those of others.
There are strengths of this study. Our observations were population based and national representative. The large size of the database allows the relative comparison of epidemiological feature between children and adults and between time periods. This study was consistent with previous studies on several observations on characteristics of moyamoya disease such has dual age peaks and women predominance.1 The epidemiological patterns observed in this study are believed to be valid.
In conclusion, this study documented the incidence and patient characteristics of moyamoya disease during a 12-year period using a national inpatient database in Taiwan. The incidence rate has increased in adults from 2000 to 2011, but no upward trend was observed in patients <18 years old. Patient characteristics including sex ratio and comorbid conditions differed by age and study period.
Sources of Funding
This study was supported by grants from the National Science Council of the Republic of China, Taiwan (Grant No. NSC 101-2321-B-002-085) to Dr Kuo and National Taiwan University and Hospital, Taiwan (UN-102-063) to Drs Kuo and Chen.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.113.004160/-/DC1.
- Received November 14, 2013.
- Accepted February 21, 2014.
- © 2014 American Heart Association, Inc.
- Kleinloog R,
- Regli L,
- Rinkel GJ,
- Klijn CJ
- Chiu D,
- Shedden P,
- Bratina P,
- Grotta JC
- Kuriyama S,
- Kusaka Y,
- Fujimura M,
- Wakai K,
- Tamakoshi A,
- Hashimoto S,
- et al
- Baba T,
- Houkin K,
- Kuroda S
- Han DH,
- Kwon OK,
- Byun BJ,
- Choi BY,
- Choi CW,
- Choi JU,
- et al
- Ikezaki K,
- Han DH,
- Kawano T,
- Kinukawa N,
- Fukui M
- Duan L,
- Bao XY,
- Yang WZ,
- Shi WC,
- Li DS,
- Zhang ZS,
- et al
- 13.↵Universal health coverage in Taiwan. National Health Insurance Administration, Ministry of Health and Welfare, Taiwan, 2012. http://www.Nhi.Gov.Tw/english/webdata/webdata.Aspx?Menu=11&menu_id=290&webdata_id=2974&wd_id=290. Accessed 30 September 2013.
- Kuroda S,
- Hashimoto N,
- Yoshimoto T,
- Iwasaki Y