Transient Ischemic Attack Requiring Hospitalization of Children in the United States: Kids’ Inpatient Database 2003–2009

Abstract

Background and Purpose

Transient ischemic attacks (TIA) are not well described in children. We assessed the prevalence of and risk factors for TIA requiring hospitalization in children in a large national database.

Methods

Using the Healthcare Cost and Utilization Project Kids’ Inpatient Database, children 1–18 years admitted for TIA in 2003, 2006 and 2009 were identified by ICD-9 code (435). Descriptive analyses identified patient characteristics. Trend analysis determined the change in annual average hospitalization days from 2003 to 2009.

Results

TIA was the primary diagnosis for 531 children. Important secondary diagnoses included sickle cell disease (20%), congenital heart disease (11%), migraine (12%), moyamoya disease (10%) and stroke (4%). Mean length of stay decreased from 3.0 days [95% confidence interval (CI) 2.4–3.6] in 2003 to 2.3 days (95% CI 2.0–2.7) in 2009 (p-value = 0.04). During the same period, 2590 children were admitted with ischemic stroke; 4.8 children with stroke were admitted for every child with TIA.

Conclusion

Recognized risk factors for TIA including sickle cell disease, congenital heart disease, moyamoya, recent stroke and migraine were present in <60% of children. Pediatric admissions for ischemic stroke were about 5-fold more common than for TIA. Further study is required to understand the risk of stroke after TIA in children to guide appropriate evaluation and treatment.

Introduction

Transient ischemic attack (TIA) has rarely been described in children. The Brain Attack Surveillance in Corpus Christi (BASIC) project found an approximate annual incidence of hemorrhagic or ischemic stroke in children of 4.3 per 100,000 and of 0.54 per 100,000 for TIA in a population-based sample of eight children with stroke or TIA.¹ In adults, depending on risk factors, 1 to 10% with TIA will have an ischemic stroke within 2 days.² However, the growing literature on TIA in adults does not apply to children who typically have different risk factors. This is a first step to understand the epidemiology and medical conditions associated with TIA in children.

Methods

The patient sample was taken from the Kids’ inpatient database (KID), part of the Healthcare Cost and Utilization Project (HCUP). A comprehensive synopsis of KID is available at http://www.hcup-us.ahrq.gov/kidoverview.jsp.

National estimates were obtained by use of discharge weights developed using the American Hospital Association as the standard. Children 1–18 years of age from KID years 2003, 2006, and 2009 were included in this study. The International Classification of Disease, 9th Revision, Clinical Modification (ICD-9-CM) primary diagnosis code 435 identified children admitted with TIA. Variables included age; sex; race/ethnicity; comorbidities were obtained from Agency for Healthcare Research and Quality’s comorbidity data files. ICD-9-CM secondary diagnosis codes identified secondary diagnoses in children with TIA including but not limited to hypertension (401, 405), elevated blood pressure (796.2), diabetes (249, 250), congenital heart disease (CHD) (745–747), migraine (346), moyamoya disease (MMD) (437.5), stroke (430–434, 436–437.1) and sickle cell disease (SCD) (282.6). ICD-9-CM procedure codes were used to estimate the percentage of TIA patients who underwent TIA-related procedures like cerebral angiography (88.41) and transfusion (99.04). Length of stay (LOS) and hospital charges were recorded. If there were <11 discharges for a given variable, that variable was excluded according to HCUP policy.

Statistical analysis

Linear regression determined the relationship between hospitalization charges and LOS after adjusting years to control for inflation of hospital charges. Trend analysis was performed to determine the change in annual average hospitalization days from 2003 through 2009 using the Cochran-Armitage trend test. SAS 9.3 software (SAS Institute, Cary, NC) was used to perform analyses.

Results

TIA was the primary hospitalization diagnosis in 531 children (Table I online supplement). Median age was 13 (interquartile range 8–16) years, and 48% were male. TIAs were more common in adolescents; 67% occurred in those age 11–18 years. Important secondary diagnoses included SCD (20%), CHD (11%), migraine (12%), MMD (10%) and stroke during the same hospital admission (4%). Anemia, coagulopathy, diabetes, hypertension, and obesity were rare comorbid conditions, each noted in ≤6% of children. Two hundred seventeen (40.9%) had no risk factor identified. Additional analyses are available in Tables II and III (online supplements). The rates of cerebral angiography and transfusion were 18% and 7%, respectively. No child with TIA died; 97% were discharged home.

Mean LOS decreased from 3.0 days (95% confidence interval (CI) 2.4–3.6) in 2003 to 2.3 days (95% CI 2.0–2.7) in 2009, p<0.04. Hospitalization charges were directly associated with LOS (p<0.0001) even after adjusting for year. During the same period, 2590 children were admitted with arterial ischemic stroke (AIS); therefore, 4.8 children with AIS were admitted for every child with TIA.

Discussion

In recent years, while research has focused on childhood AIS,³ few studies have described TIA in a pediatric population. In adults with TIA, 1 to 10% will have a stroke within the next 48 hours, ² and up to 15% will have a stroke within 3 months.⁴ In this 3-year sample of more than 500 children hospitalized with a primary diagnosis of TIA, 22 (4.2%) had a secondary diagnosis of stroke within the same hospitalization. Of these 22 children, 17 had identifiable risk factors: 10 had SCD, 5 had MMD, and 2 had migraine.

TIAs were more common in adolescents with 67% occurring in those age 11–18 years. Of note, stroke was nearly 5 times more common than TIA in the Kids’ database. While the frequency of stroke following a TIA in children has not been reported, about a third of children who had a confirmed AIS had a history of recent TIAs,⁵ and prior TIA is a strong risk factor for recurrent stroke in children.⁶

Similar to our study, the sparse pediatric TIA literature has primarily included children with MMD and children with SCD. In the Cooperative Study of Sickle Cell Disease, a history of TIA was a strong risk factor for AIS.⁷ In young children with MMD, the most common presenting symptom is TIA.⁸

In our study, of 531 children with TIA, 217 (40.9%) had no risk factor identified. This is consistent with pediatric AIS literature where up to 30% of strokes remain cryptogenic even after a complete vascular, cardiac, and prothrombotic evaluation.³

This study has limitations inherent to the Kids’ data set, namely the accuracy of the diagnosis and procedures codes listed in the discharge summaries. In a previous study, the sensitivity and positive predictive values in adults for a primary diagnosis of ICD-9-CM code 435, TIA, were 75% and 80%, respectively.⁹ It is also likely that TIA is underdiagnosed in children. Without detailed chart review, it is difficult to determine whether migraine is a risk factor for TIA and stroke in children as it is in adults or whether complicated migraine is simply a common differential diagnosis for TIA in children.¹⁰ Definitively distinguishing whether children with a primary diagnosis of TIA had stroke or migraine prior to TIA or after TIA is not possible in this administrative dataset. Finally, application of the diagnosis TIA might not be uniform across all hospitals.

Conclusion

Recognized risk factors for TIA including SCD, CHD, MMD, recent stroke and migraine were present in <60% of children. Admissions for AIS were about 5-fold more common than for TIA in children, and 4.2% with a primary diagnosis of TIA also had stroke during the same hospitalization. Children in this high risk category often had SCD or MMD. Children with TIA may also be quite healthy: LOS was short, no child died and the most were discharged home and not to a rehabilitation facility. Additional study is required to understand the risk of stroke after TIA in children to guide appropriate evaluation and treatment.