National Analysis of 2454 Pediatric Moyamoya Admissions and the Effect of Hospital Volume on Outcomes
Background and Purpose—Comprehensive multicenter data on treatment of pediatric moyamoya in the United States is lacking. We sought to identify national trends in the diagnosis and treatment of this disease.
Methods—A total of 2454 moyamoya admissions from 1997 to 2012 were identified from the Kids Inpatient Database. Demographics, inpatient costs, interventions, and discharge status were analyzed. Admissions with and without surgical revascularization were reviewed separately. The effect of hospital moyamoya volume on outcomes was analyzed by multivariate regression analysis.
Results—Care of moyamoya patients has been concentrating at high-volume centers during the past 12 years. Among moyamoya admission without surgical revascularization, high-volume hospitals show no difference in length of stay, cost, or complications compared with low-volume centers. However, low-volume hospitals have more nonroutine discharges (odds ratio, 2.32; P=0.0005) and inpatient deaths (odds ratio, 12.7; P=0.02) when no revascularization was performed. In contrast, among admissions with surgical revascularization, high-volume centers had decreased length of stay (4.7 versus 6.2 days; P=0.004), reduced cost ($88 000 versus $138 000; P<0.0001), and no increase in complications (P=0.29) compared with low-volume centers. Admissions with revascularization to low-volume hospitals also had increased likelihood of nonroutine discharge (odds ratio, 8.23; P=0.02) compared with high-volume centers.
Conclusions—This is the largest study of US pediatric moyamoya admissions to date. These data demonstrate that volume correlates with outcome, indicating high-volume centers provide significantly improved care and reduced mortality in pediatric moyamoya patients, with the most marked benefit observed in admissions for surgical revascularization.
Moyamoya disease is an arteriopathy of the intracranial internal carotid arteries and their proximal branches that predisposes affected patients to symptoms of cerebral ischemia and stroke.1 Surgical revascularization, including both direct and indirect techniques, is utilized as the primary treatment to reduce stroke risk in children.2 However, because of the rarity of this disorder (prevalence ≈0.08–3/100 000 children3,4), knowledge of moyamoya demographics and outcomes are predominantly derived from small single-institution case series5–8 or meta-analysis.2 Currently, only 1 US population-based study of healthcare utilization among moyamoya has been performed, and this included both adult and pediatric admissions.9 Furthermore, no data are available to objectively compare treatment of pediatric moyamoya and outcomes across institutions.
To address these knowledge gaps, we analyzed current moyamoya admissions both with and without revascularization surgery, using the largest US pediatric inpatient database. Previous literature suggests that hospital and provider case volume is a significant determinant of morbidity and mortality when treating specialized conditions. We hypothesized that institutional moyamoya volume would be a significant predictor of inpatient outcomes. If true, this would favor a clinical strategy of directing children to high-volume centers for treatment safety and saving medical resources. To our knowledge, this is the first national analysis of pediatric US admissions for both surgical and nonsurgical moyamoya, and our hope is that this approach will provide novel insights into the current demographics and treatment of pediatric moyamoya in the United States.
For this analysis, the authors utilized the Kids’ Inpatient Database (KID), a pediatric inpatient sample generated by The Healthcare Cost and Utilization Project (HCUP) triennially from 1997 to 2012. It is the largest US pediatric inpatient database available, containing >18 million records (http://www.hcup-us.ahrq.gov/kidoverview.jsp). Inclusion criteria were age <22 years old and either a diagnosis of moyamoya disease (437.5) or extracranial to intracranial (ECIC, 39.28) bypass. Direct versus indirect ECIC bypass are not separately coded in this database, so any type of revascularization surgery is generically called ECIC bypass. No patient undergoing revascularization lacked a concurrent diagnosis of moyamoya. Patient variables included age, race, payer status, sex, admission source, and admission type. Numbers of diagnoses and procedures were chosen as indicators of case complexity. Institutional variables included hospital region, location (rural or urban), teaching status (teaching or nonteaching), and number of beds. For each year, a hospital’s annual moyamoya admissions were calculated. Then, all hospitals were categorized as high-, moderate-, or low-volume institutions by determining cutpoints that divided each year’s total moyamoya admissions into approximate thirds. This resulted in roughly equal number of admissions in each moyamoya volume category. The 15-year span of the study and the increasing number moyamoya admissions made the use of fixed cutpoints inappropriate.
The primary outcome of interest was discharge disposition, which was dichotomized into routine and nonroutine. Routine discharge was defined as discharge to home or self-care, whereas nonroutine discharges included transfer to short-term hospital, skilled nursing facility, other transfer, home healthcare, or leaving against medical advice. Secondary outcomes included length of stay (LOS), total charges, and inpatient death. Complications were determined from International Classification of Diseases Ninth Revision (ICD-9) CM by previously published methods.10–12 These included wound complications, infectious, urinary, pulmonary, gastrointestinal, cardiovascular, systemic, intraoperative, neurological, anesthetic, and complication requiring additional procedures. The total number of complications was treated as a continuous variable. Methods section in the online-only Data Supplement for ICD-9 definitions of complications.
Hospital moyamoya volume served as the primary exposure of interest. Comparison of categorical variables was by χ2 and continuous was by 2-tailed Student t test. LOS and total complications were analyzed by negative binomial regression, death and nonroutine discharge were by logistic regression, and total hospital charge was by linear regression. Multivariate regression modeling used survey-adjusted generalized estimating equations that were fit to model outcomes and adjusted for patient factors, hospital factors, and admission severity. We adjusted for clustering by hospital. Unfortunately, the HCUP databases lack a unique patient identifier; and therefore, we could not control for multiple measurements in a single patient. Extrapolations to the entire US population were adjusted for survey methodology by weights. Statistical significance was P<0.05. All analyses were conducted in SAS 9.4 (Cary, NC).
The KID database contains 2454 admissions for moyamoya. Male/female ratio was not significantly different from all US hospital admissions. Four hundred fifty-four moyamoya admissions (18.5%) had revascularization surgery. Moyamoya admissions with revascularization surgery had a 4-fold higher percentage of Asian and Pacific Islanders (11.5% versus 2.7%) and a lower percentage of Hispanics (7.2% versus 17.1%) than the US population. Moyamoya admissions with and without revascularization surgery show distinct differences (Figure 1A; Table 1). Blacks composed the largest group of moyamoya admissions without revascularization surgery (34.8%, Figure 1A). Furthermore, the percentage of blacks is 2.3-fold higher among moyamoya admissions without revascularization surgery (34.8 versus 15.2%) and 2.8-fold higher than blacks admission in all US hospital admissions (12.7%).
Most operative moyamoya admissions with revascularization surgery were routine (53.7%), elective (45.6%) admissions of patients with only 2 diagnoses (Table 1). In contrast, admissions without revascularization surgery were most commonly emergent admissions (28.8%) and were more commonly through an emergency room (15.3 versus 2.2%) or hospital transfer (3.85 versus 0.4%). Nonoperative admissions also had a higher number of diagnoses (5 versus 2) indicating higher acuity.
Almost all surgical admissions had a primary diagnosis of moyamoya (n=437, 96.3%), with the few remaining composed of cerebrovascular accidents, strokes, transient ischemic attack, and nonruptured cerebral aneurysms (Table I in the online-only Data Supplement). Admissions with revascularization surgery were notable for 8.6% (n=39) with Down syndrome, 6.4% (n=29) with neurofibromatosis type I, and 1.8% (<10) with dwarfism.
Admissions without revascularization surgery were more diverse, with moyamoya comprising only 24.1% (n=483) of the primary diagnoses (otherwise named as a secondary diagnosis), with a high number of sickle-cell admissions (Table I in the online-only Data Supplement). Disease of the blood comprised 17.2% of principal diagnoses, including 311 patients (15.6%) with sickle-cell disease (SCD). Furthermore, 11.2% of admissions undergoing surgery and 36.7% of admissions without surgery had at least 1 diagnostic code for sickle-cell trait or disease. This increased to 60.9% and 81.3% among black moyamoya admissions. The primary diagnosis of stroke, ischemia, transient ischemic attack, or sequelae of cerebrovascular accident was assigned to 12.5% of nonsurgical patients (n=250). The ICD category of ill-defined symptoms (7.5%, n=150) comprised predominantly central nervous system-related conditions, such as convulsions (n=48), syncope (n=11), and altered mental status (<10). Finally, principal diagnoses related to the central nervous system (4.5%, n=89) included 3.4% (n=68) with epilepsy or seizures and the remainder including intracerebral hemorrhage or subdural hematoma.
Revascularization surgery was performed in 18.5% of all moyamoya admissions (n=454). When revascularization surgery was performed, it was the primary procedure 91% of the time. Among the 2000 moyamoya admissions without revascularization, 1420 procedures were performed (Table II in the online-only Data Supplement). These included 722 miscellaneous diagnostic procedures that incorporates imaging (54.4%), procedures related to blood dyscrasias (16.3%, n=232, including transfusions and therapeutic erythrocytopheresis), arteriography of cerebral vessels (11.1%, n=157), primary procedures occurring in the nervous system (n=299, 21%, including 163 craniotomies for evacuation of hygroma, hematoma, or abscess), CSF diversion procedures (3.6%, n=51), dural repairs (2.7%, n=38), cranioplasties or wound revisions (1.2%, n=17), and primary procedures occurring in the cardiovascular system (n=131, 9.2%).
Low- Versus High-Volume Hospitals
Hospitals varied widely in the annual number of moyamoya admissions. Admissions for moyamoya seemed to concentrate at a small number of high-volume centers over the study with the annual admissions to high-volume centers increasing from 7 to 19 cases per year during the study period (Figure 1B). In 2012, a third of moyamoya admissions were spread over 118 low-volume hospitals (median=2 admissions/y), whereas a similar number of moyamoya admissions were managed by just 13 high-volume centers (median=19 admissions/y; Table 2).
Moderate- and low-volume moyamoya centers tended to be tertiary care centers, with more annual discharges, and mixed age populations (36% and 51% pediatric discharges versus 77%). High-volume moyamoya centers were exclusively teaching institutions and more likely to be free-standing children’s hospitals and located in the south or northeastern United States (Table 2). Admissions to high-volume hospitals were 3× more likely to undergo revascularization surgery than admissions to low-volume institutions (30.9 versus 9.5%; P<0.0001; Table 3).
Hospital Volume Effect on Patient Outcomes
Hospital volume had a direct correlation with patient outcomes, including LOS, charges, complications, and nonroutine discharge/death (Table 4). For admissions with surgical revascularization (ECIC), high- and moderate-volume hospitals had significantly shorter lengths of stay than low-volume institutions (4.7 versus 6.2 days; P=0.004; Table 4; Figure 2A). For nonrevascularization admissions, moderate-volume hospitals had the lowest LOS (4.9 versus 5.9 days; P=0.005). Significant predictors of increased LOS included increasing number of diagnoses (P<0.0001), female sex (P=0.04), payer status (P=0.03), hospital teaching status (P=0.006), and admission in the later years of the study (P<0.0001).
Higher hospital volume correlated with lower charges after correcting for patient factors and revascularization surgery (P<0.001; Figure 2B). Among admissions with surgical revascularization, high-volume and moderate-volume centers had 32% to 48% lower charges than low-volume hospitals (P<0.0001). Again, admissions without a revascularization surgery showed moderate-volume centers had the lowest total charges (P<0.0001). Other patient factors associated with charges included revascularization surgery (+$52 195; P<0.0001), later year of admission (+$1504 per year; P<0.0001), increasing age (−$603 per year; P=0.003), female sex (+$6866; P=0.002), increasing income quartile (+3633; P=0.002), number of diagnosis (+$1076; P<0.0001), race (P<0.0001), and payer status (P=0.0003). Admissions with private insurance were charged nearly $12 000 more than those utilizing Medicaid ($68 100±1700 versus $56 000±1800) and these 2 payers covered >90% of admissions. Hospital factors predictive of total charges included hospital region (P<0.0001) and teaching status (+$28 222; P<0.0001).
Admission to a moderate-volume moyamoya hospital was associated with the lowest risk of complications (Table 4; P=0.004; Figure 2C), but only for nonsurgical admissions, (P=0.008). The number of diagnoses (P<0.0001) and hospital region (P=0.002) were predictive of inpatient complications. Factors associated with a reduced complications include admission during a later year (P=0.03) and older patient age (P<0.0001)
High moyamoya volume is associated with a much lower risk of nonroutine discharge and death (Table 4; Figure 2D). Low-volume hospitals had more nonroutine discharges for both admissions with revascularization surgery (odds ratio [OR], 10.12; P=0.02) and without (OR, 2.11; P=0.0005). Significant associations with nonroutine discharge included year of admission (OR, 0.93; P=0.0009), number of diagnoses (OR, 1.20; P<0.0001), hospital region (OR, 0.76; P=0.0025), and number of hospital beds (OR, 1.46; P=0.027). Admissions to low-volume hospitals also had 2.8-fold higher odds of death than high-volume hospital, even after correcting for acuity (P=0.01). The risk of death cannot be calculated for admissions with revascularization surgery given the low prevalence (n<10). Factors associated with risk of death include year (OR, 0.84; P=0.004) and number of diagnoses (OR, 1.33; P<0.0001).
The KID is the largest pediatric inpatient US data set and functions as the pediatric counterpart to the National Inpatient Sample. The database contains nearly 40% of all national inpatient records for the years acquired. The unique design of the KID enables national and regional estimates of pediatric disorders and allows tracking and analysis of national trends in utilization, charges, quality, and outcomes. This is particularly important for studying rare conditions such as moyamoya, where the relevant literature is typically limited to single-institution reports, prone to sampling error, and bias. We utilized the KID to address knowledge gaps in pediatric moyamoya care, identifying national trends in moyamoya admissions, and determine the role that hospital volume plays in outcome. To our knowledge, this is the first use of a national pediatric database to study US moyamoya admissions.
Admission data from KID reveal that the demographics of US pediatric moyamoya differ from other countries and single-center reports. Previous epidemiological studies in Japan report a female prevalence of moyamoya twice that of males.3,7 Single institution reports from the United States also demonstrate a 71% female prevalence.13,14 However, although our study found a preponderance female admissions, it was slightly less (57%) than the other reports, suggesting possible subtle differences in sex-related admissions that may merit further study.
Historic data suggest a predominance of moyamoya patients with Asian ancestry (nearly all Japanese literature, and some high-volume US centers report ≤32% of Asian patients). However, the KID data confirm a high percentage of white moyamoya admissions (31.6%) and only 4.8% US admissions with Asian ancestry. Moreover, a seemingly unique finding is the high number of black moyamoya admission in this US study (31.2%). Here blacks comprised 15% of surgical (ECIC) admissions and 35% of nonrevascularization surgery pediatric moyamoya admissions. This contrasts against smaller or single-center series that suggest only 5% to 10% of pediatric moyamoya patients are of black background.7,13,15,16 The high number of nonsurgical admissions for black moyamoya patients might be explained by sickle-cell issues unrelated to problems specific to moyamoya. In fact, 75.9% of black admissions in this study carried a codiagnosis of sickle-cell trait or disease. Taken together these data suggest a relatively high number of US moyamoya patients are not of Asian ancestry. Similarly, a study of the National Inpatient Sample incorporating both adults and children with moyamoya found nearly identical racial demographics supporting these findings.9
This important finding highlights the differences in presentations of syndromic moyamoya in the US pediatric population. Case reports have previously associated pediatric moyamoya syndrome with SCD,17,18 neurofibromatosis type I,19 Down syndrome,20 and primordial dwarfism among other conditions.21 One single-institutional pediatric moyamoya series reported a prevalence of 11% neurofibromatosis type I, 7% Down syndrome, and 2% SCD.10 In contrast, the KID data reveal a distribution of 6.4% neurofibromatosis type I, 8.6% Down syndrome, and 16% SCD. Furthermore, SCD or trait was present in 60.9% of surgical and 81.3% of nonsurgical black admissions. These differences highlight the ability of a larger database to more accurately discern national trends from institutional referral patterns and facilitate identification of important subgroups within a disease population.
There is precedent in the literature for linking volume and outcome in studies of adult neurosurgical care. Utilizing the Nationwide Inpatient Sample, high-volume centers are associated with decreased morbidity, mortality, shorter LOS, and lower total hospital charges in treatment of aneurysms,22,23 meningiomas,24 intracranial metastasis,25 and transsphenoidal surgery for pituitary tumors.26 Pediatric neurosurgical literature about volume–outcome relationship, whereas more limited in scope, mirrors its adult counterpart. Analysis of unruptured cerebral aneurysm treatment in children using the KID showed decreased mortality for teaching hospitals (OR, 0.13; P=0.001) and hospitals with more beds (OR, 0.35; P=0.01),27 whereas a study of 4712 admissions for pediatric brain tumor craniotomies demonstrated significantly lower mortality in high-volume hospitals (2.3% versus 1.4%).28 However, there are no similar data for pediatric moyamoya.
Using KID, we analyzed outcomes (as determined by LOS, cost, complications, nonroutine discharge, and death) as a function of a hospital’s pediatric moyamoya volume. Admissions to a high-volume center had significantly improved outcomes. Admission to a high-volume center with moyamoya but without a revascularization surgery had increased cost, slightly increased complications, but much lower chances of nonroutine disposition or death when compared with a low-volume center. Children undergoing surgical revascularization for moyamoya during their admission experienced the greatest benefits when treated at a high-volume center, including decreased LOS (23%), reduced cost (32%), and an 89% decreased likelihood of nonroutine discharge. These data suggest potential benefits to concentrating pediatric moyamoya care at high-volume centers.
Limitations and Future Directions
Several limitations of this study should be addressed. First, the retrospective nature of the KID and the use of national administrative data make it subject to variability in data collection, recording, and individual-level hospital reporting. However, previous studies estimate 80% accuracy of ICD-9 CM codes in national administrative data sources29 and moyamoya is coded as a single diagnosis in ICD-9 CM, maximizing accuracy. Second, KID data classification of ECIC bypass does not distinguish between indirect or direct revascularization. Third, the KID consists of individual patient admissions and lacks any unique patient identifier, making longitudinal tracking of a single patient impossible outside of a given admission. With these points acknowledged, KID is the only data set currently available for studying US pediatric moyamoya admissions from a national vantage point. Moreover, the KID allows for temporal trends analysis during its 15-year span, which we are currently pursuing. Finally, it is relevant to comment that the authors of this study are surgeons at high-volume centers, which may introduce the potential for bias. Consequently, this work will benefit from peer-review and validation from other groups.
To our knowledge, this is the first study to use a robust national database to generate an analysis of American pediatric moyamoya demographics and to evaluate the effect of hospital volume on outcome. These data reveal differences in the US pediatric moyamoya population that contrast to populations in other countries and to previous single-center reports. With lower percentages of female admissions, more syndromic cases and a particularly high number of black admissions with moyamoya, these data support the hypothesis that unique subpopulations of moyamoya pathophysiology may exist in the United States. We also present data that demonstrates—for the first time in this population—that treatment of pediatric moyamoya patients at high-volume centers, especially for surgical revascularization, is associated with significantly lower costs, shorter LOS, and better clinical outcomes. Clinical management of moyamoya includes complex surgical, anesthetic and nursing care, all of which may improve with coordinated teams benefiting from high-volume practice.30,31 These data may support the consideration of policies directing these patients to high-volume centers. In combination, these results offer novel insights into the composition of the United States pediatric moyamoya population, support the efficacy of high-volume centers in delivering better financial and clinical care for this disorder and suggest the need for further research to better understand why these differences exist.
Sources of Funding
This study was supported by American Association of Neurological Surgeons/Congress of Neurological Surgeons Joint Pediatric Section Resident Traveling Fellowship.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.115.012168/-/DC1.
- Received November 22, 2015.
- Revision received February 10, 2016.
- Accepted February 29, 2016.
- © 2016 American Heart Association, Inc.
- Arias EJ,
- Dunn GP,
- Washington CW,
- Derdeyn CP,
- Chicoine MR,
- Grubb RL >Jr,
- et al
- Gonzalez NR,
- Dusick JR,
- Connolly M,
- Bounni F,
- Martin NA,
- Van de Wiele B,
- et al
- Kennedy BC,
- McDowell MM,
- Yang PH,
- Wilson CM,
- Li S,
- Hankinson TC,
- et al
- Bober MB,
- Khan N,
- Kaplan J,
- Lewis K,
- Feinstein JA,
- Scott CI Jr,
- et al
- Hoh BL,
- Rabinov JD,
- Pryor JC,
- Carter BS,
- Barker FG II
- Barker FG II,
- Amin-Hanjani S,
- Butler WE,
- Ogilvy CS,
- Carter BS
- Burns EM,
- Rigby E,
- Mamidanna R,
- Bottle A,
- Aylin P,
- Ziprin P,
- et al