Disparities in the Use of Seizure Medications After Intracerebral Hemorrhage

Abstract

Background and Purpose

We investigated potential disparities in the use of prophylactic seizure medications in patients with intracerebral hemorrhage.

Methods

Review of multicenter electronic health record (EHR) data with simultaneous prospective data recording. EHR data were retrieved from HealthLNK, a multi-center EHR repository in Chicago, IL, from 2006–2012 ("multicenter cohort"). Additional data were prospectively coded (“single center cohort”) from 2007 through 2015.

Results

The multicenter cohort was comprised of 3,422 patients from four HealthLNK centers. Use of levetiracetam varied by race/ethnicity (P=0.0000008), with whites nearly twice as likely as blacks to be administered levetiracetam (OR 1.71, 95% CI 1.43 – 2.05, P<0.0001). In the single center cohort (N=450), hematoma location, older age, depressed consciousness, larger hematoma volume, and no alcohol abuse, and race/ethnicity were associated with levetiracetam administration (P<=0.04). Whites were nearly twice as likely as blacks to receive levetiracetam (OR 1.9, 95% CI 1.25 – 2.89, P=0.002), however, the association was confounded by history of hypertension, higher blood pressure on admission, and deep hematoma location. Only hematoma location was independently associated with levetiracetam administration (P<0.00001), rendering other variables, including race/ethnicity, non-significant.

Conclusions

While multicenter EHR data showed apparent racial/ethnic disparities in the use of prophylactic seizure medications, a more complete single center cohort found the apparent disparity to be confounded by the clinical factors of hypertension and hematoma location. Disparities in care after intracerebral hemorrhage are common, however, administrative data may lead to the discovery of disparities that are confounded by detailed clinical data not readily available in EHRs.

Seizures are a noted complication of intracerebral hemorrhage (ICH). Although prophylactic phenytoin was previously recommended,¹ revised guidelines no longer recommended prophylactic seizure medications^{2, 3} after phenytoin⁴ and seizure medications generally⁵ were associated with worse outcomes. Yet, levetiracetam continues to be administered to nearly 40% of patients with acute ICH nationwide,⁶ suggesting that clinicians often consider their use to be appropriate.

Disparities in care for patients with ischemic stroke are well described,⁷ however, there are few data on disparities in care for patients with ICH.⁸ We hypothesized that there are racial/ethnic disparities in the use of prophylactic seizure medications.

Materials and Methods

Multi-center cohort

We utilized the Chicago HealthLNK Data Repository (”HealthLNK”). The logistics, procedures and patient privacy issues have been previously described.⁹ Data from 2007 – 2012 were available from Loyola, Rush, and Northwestern Universities, and the University of Chicago. Patients with diagnostic code 431 (Intracerebral hemorrhage) from the International Classification of Diseases, 9^th Ed were identified.

Single-center cohort

The Northwestern University Brain Attack Registry (NUBAR) has enrolled patients with ICH since 2007 (Figure).⁴

Figure.

Organizational chart of the prospective cohort.

Standard Protocol Approvals, Registrations, and Patient Consents

The study was approved by the Northwestern University Institutional Review Board. Patients or a legally authorized representative were asked for written consent.; Patients who died or were not consentable were enrolled under a waiver.

Statistical Analysis

Normally distributed data were compared with analysis of variance, categorical data were analyzed with chi-squared and non-parametric data were analyzed with the Mann-Whitney U or Kruskal-Wallis H. Logistic regression set prophylactic levetiracetam as the dependent variable using forward selection. The analysis was repeated in patients who survived at least five days after ICH.¹⁰

Results

Multi-center Cohort

The multicenter cohort was composed of 3,422 patients, of whom 1,777 (51.9%) were male with a mean age of 57.7 +/− 15.9 years. Race/ethnicity and levetiracetam were associated (Table 1). Whites were almost twice as likely as blacks to be administered levetiracetam (OR 1.71, 95% CI 1.43 – 2.05, P<0.0001). Age and sex were not associated with levetiracetam administration.

Table 1.

Race/ethnicity and levetiracetam use were associated (P=0.0000008) in the multi-center cohort.

Race	No Levetiracetam (N=2386)	Levetiracetam (N=748)
1 Native American	12 (1)	4 (1)
2 Asian	81 (3)	23(3)
3 Black	1041 (44)	239 (32)
4 Hispanic or Latino	78 (3)	16 (2)
5 Pacific Islander	5 (0)	6 (0)
6 White	1051 (44)	413 (56)
7 Declined	51 (2)	23 (3)
8 Other	67 (3)	24(3)

Data are N(%). Data do not sum to 3,422 due to missing data in some patients on race/ethnicity.

Single Center Cohort

Among 450 patients with spontaneous ICH (Table 2), compared to blacks, whites had lower systolic (185 +/− 43 versus 176 +/− 38 mm Hg, P=0.03) and diastolic blood pressures (102 +/− 29 versus 90 +/− 25 mm Hg, P=0.00001) on admission. Compared to blacks (N=163, 51%), whites (N=138, 43%) and Hispanics (N=19, 6%) were less likely to have a prior history of hypertension (P=0.00007).

Table 2.

Demographics of the 450 patients in the prospective cohort stratified by prophylactic levetiracetam use. Lobar hematoma, depressed Glasgow Coma Scale, older age, and known alcohol abuse were associated with levetiracetam administration. In multivariate analysis, only lobar hematoma was associated with levetiracetam administration.

Variable	Levetiracetam	No Prophylactic Levetiracetam	P
N	138	312
Age	67.7 +/− 15.23	64.5 +/− 14.11	0.03
Women	77 (57)	159 (51)	0.2
Race/Ethnicity Missing	2 (1)	3 (1)	0.3
American Indian	0	2 (1)
Asian	3 (2)	5 (2)
Black	45 (33)	148 (47)
Hispanic or Latino	11 (8)	24 (8)
Native Pacific Islander	2 (1)	9 (3)
White	75 (55)	121 (39)
Hematoma location, lobar	77 (59)	73 (25)	<0.00001
Deep	46 (46)	156 (54)
Infratentorial	6 (5)	47 (16)
Other	9 (6)	36 (11)
GCS, 13–15	71 (52)	196 (63)	0.007
5–12	54 (39)	77 (25)
3–4	11 (9)	39 (12)
Hematoma >= 30 mL	54 (38)	77 (25)	0.08
Intraventricular hemorrhage	68 (50)	151 (48)	0.9
Supratentorial hematoma	128 (94)	258 (83)	0.02
Age >= 80 years	36 (26)	48 (15)	0.01
ICH Score	2 [1 – 3]	1 [0 – 2.5]	0.01
Previous ICH	8 (6)	8 (3)	0.2
Previous ischemic stroke	17 (13)	35 (11)	0.8
History of coronary disease	23 (17)	48 (15)	0.3
Atrial fibrillation	16 (12)	27 (9)	0.5
Historical hypertension	93 (68)	242 (78)	0.08
Alcohol abuse	4 (3)	31 (10)	0.04
Historical dementia	4 (3)	7 (2)	0.9
Warfarin use	12 (8)	39 (13)	0.5

Data are N(%), median [Q1 – Q3] or mean +/− SD

Factors associated with levetiracetam administration are shown in Table 2. Whites were nearly twice as likely as blacks to be administered levetiracetam (OR 1.9, 95% CI 1.25 – 2.89, P=0.002). In multivariate analysis, hematoma location entered first and rendered other variables non-significant. Similar results were found for patients who survived at least 5 days and patients with seizures.

Discussion

Whites were approximately twice as likely as blacks to receive levetiracetam. This apparent disparity was confounded by hematoma location, with whites more likely to have lobar hematomas, the clinical factor most closely linked to a higher risk of seizures.¹¹

Although prophylactic seizure medications are no longer recommended after ICH,^{2, 3} their use remains common.⁶ The use of levetiracetam in patients with a lobar hematoma is rational, if not strictly compliant with the guidelines, one of many scenarios where clinical practice deviates from evidence-based recommendations.¹² Disparities in prophylactic phenytoin were not examined because we discontinued its use in 2009.

Associations between seizure medications and outcomes were not evaluated. Adverse events associated with phenytoin are relatively easy to detect.⁴ Levetiracetam has fewer side effects than phenytoin, and any adverse effects on cognition or behavior are likely to be detected only in patients already considered to have a “good outcome.”¹³ Other validated scores of health-related quality of life and cognition could be a useful adjunct.^{14, 15}

In sum, we found that an apparent disparity in the use of seizure medications was confounded by hematoma location. There are likely to be other disparities found in the care of patients with ICH, some of which are confounded.