Population-Based Assessment of the Long-term Risk of Seizures in Survivors of Stroke

Alexander E. Merkler, MD; Gino Gialdini, MD; Michael P. Lerario, MD; Neal S. Parikh, MD; Nicholas A. Morris, MD; Benjamin Kummer, MD; Lauren Dunn, MD; Michael E. Reznik, MD; Santosh B. Murthy, MD, MPH; Babak B. Navi, MD, MS; Zachary M. Grinspan, MD, MS; Costantino Iadecola, MD; Hooman Kamel, MD

Disclosures

Stroke. 2018;49(6):1319-1324. 

In This Article

Methods

Design

We performed a population-based, retrospective study of 2 large cohorts of patients from across the United States. In the first analysis, we used administrative claims data from all nonfederal emergency department and acute care hospital discharges in California from 2005 through 2011, New York from 2006 through 2013, and Florida from 2005 through 2013; these dates were chosen as they represent the most recent data available for analysis. These data were collected by the California Office of Statewide Health Planning and Development, the New York State Department of Health, and the Florida Agency for Health Care Administration and were provided to the Agency for Healthcare Research and Quality for its Healthcare Cost and Utilization Project.[19]

In the second analysis, we used both inpatient and outpatient claims data from a 5% random sample of nationwide Medicare beneficiaries from 2008 through 2014. The US federal government's Centers for Medicare and Medicaid Services provides health insurance to a large majority of US residents once they reach 65 years of age. Centers for Medicare and Medicaid Services makes available to researchers deidentified data sets that include data on claims submitted by providers and hospitals in the course of Medicare beneficiaries' clinical care.[20] In keeping with standard practice in analyzing Medicare data,[21] we limited our cohort to beneficiaries with continuous coverage in traditional fee-for-service Medicare (both parts A and B) for at least 1 year.

In both data sets, all patients are assigned a deidentified personal linkage number that enables them to be followed anonymously over multiple years.[22] Each claim includes the dates of service and up to 25 International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis codes justifying the claim. The complementary nature of these 2 data sets facilitated the study of patients regardless of age or insurance status through the Healthcare Cost and Utilization Project data and analysis of outpatient as well as inpatient diagnoses drawn from Medicare data.

Standard Protocol Approval

In performing this study, we adhered to the Report of Studies Conducted Using Observational Routinely-Collected Health Data guidelines for analyses of administrative claims data.[23] The data that support the findings of this study are available on request. The institutional review board at Weill Cornell Medicine approved this study and waived the requirement for informed consent.

Subjects

In the multistate analysis, we identified all patients 18 years or older at the time of a first-ever recorded hospitalization for stroke. Since Medicare eligibility generally begins at 65 years of age, in the analysis of claims data on Medicare beneficiaries, we included only patients ≥66 years of age to allow adequate time for beneficiaries to enter medical care and for their providers to document any preexisting comorbidities. In both analyses, stroke was defined as the composite of ischemic stroke, intracerebral hemorrhage (ICH), and subarachnoid hemorrhage (SAH). In accordance with a validated algorithm, ischemic stroke was defined as ICD-9-CM codes 433.x1, 434.x1, or 436 in any hospital discharge diagnosis code position without a concurrent primary discharge code for rehabilitation (V57) or any codes for trauma (800–804 or 850–854), SAH (430), or ICH (431). ICH was defined as ICD-9-CM discharge code 431 without concomitant codes for rehabilitation, trauma, or SAH. Finally, SAH was defined as ICD-9-CM discharge code 430 without concomitant codes for rehabilitation or trauma. This algorithm has been previously validated to have a sensitivity of ≥82% and a specificity of ≥92% for correctly identifying both ischemic and hemorrhagic stroke subtypes.[24] Patients who died during the index hospitalization were excluded. Patients coded as having multiple encounters for stroke were identified at the time of their first recorded hospital admission for stroke. To focus on new-onset seizures after stroke, patients with diagnoses of seizure before the index stroke hospitalization were excluded.

Measurements

Our primary outcome was a seizure occurring after hospital discharge for stroke. In the multistate analysis, our primary outcome was an emergency department visit or hospitalization for seizure after discharge from the index stroke hospitalization, defined as any ICD-9-CM code for epilepsy (345.x) in any discharge diagnosis position; this approach has been shown to have a positive predictive value ranging from 84% to 98% in adults.[25–27] Our secondary outcome was subsequent hospitalization for status epilepticus (ICD-9-CM code 345.3) documented as being present on admission. As Medicare claims data includes information on outpatient visits, in the second analysis, our outcome of seizure was defined as any inpatient or outpatient claim with an ICD-9-CM code for epilepsy (345.x); to avoid rule out diagnoses used to justify diagnostic tests (eg, neuroimaging scans, electroencephalograms) we included only outpatient diagnoses associated with an evaluation and management claim.

Statistical Analysis

We used standard descriptive statistics with exact confidence intervals (CI) to report crude rates. Baseline characteristics were compared using the χ2 test and Student t test. Survival statistics were used to calculate incidence rates of seizure per 100 000 patients per year.

After the approach of prior studies on the long-term risk of seizures,[28] we compared the risk of seizures in those with stroke versus the general population. For the multistate analysis, we used publicly available demographic data from the 3 states included in this analysis to determine seizure risk in the general population;[29] and for the Medicare analysis, we compared seizure risk in stroke patients versus the remaining beneficiaries. For both analyses, we used Poisson regression to calculate age-, sex-, and race-standardized incidence rate ratios (IRR). Subgroup analyses were performed stratified by patients' age (<65 versus ≥65), gender, and race (white versus nonwhite). All statistical analyses were performed by HK using SAS version 9.3 (Cary, NC) and STATA/MP version 13 (College Station, TX). Statistical significance was set at α=0.05.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....