In-hospital outcomes in patients with and without epilepsy diagnosed with COVID-19—A cohort study

Abstract

Objectives:

Coronavirus disease 2019 (COVID-19) is associated with mortality in persons with comorbidities. The aim of this study was to evaluate in-hospital outcomes in patients with COVID-19 with and without epilepsy.

Methods:

We conducted a retrospective study of patients with COVID-19 admitted to a multicenter health system between March 15, 2020, and May 17, 2021. Patients with epilepsy were identified using a validated International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM)/ICD-10-CM case definition. Logistic regression models and Kaplan–Meier analyses were conducted for mortality and non-routine discharges (i.e., not discharged home). An ordinary least-squares regression model was fitted for length of stay (LOS).

Results:

We identified 9833 people with COVID-19 including 334 with epilepsy. On univariate analysis, people with epilepsy had significantly higher ventilator use (37.70% vs 14.30%, p < .001), intensive care unit (ICU) admissions (39.20% vs 17.70%, p < .001) mortality rate (29.60% vs 19.90%, p < .001), and longer LOS (12 days vs 7 days, p < .001). and fewer were discharged home (29.64% vs 57.37%, p < .001). On multivariate analysis, only non-routine discharge (adjusted odds ratio [aOR] 2.70, 95% confidence interval [CI] 2.00–3.70; p < .001) and LOS (32.50% longer, 95% CI 22.20%–43.60%; p < .001) were significantly different. Factors associated with higher odds of mortality in epilepsy were older age (aOR 1.05, 95% CI 1.03–1.08; p < .001), ventilator support (aOR 7.18, 95% CI 3.12–16.48; p < .001), and higher Charlson comorbidity index (CCI) (aOR 1.18, 95% CI 1.04–1.34; p = .010). In epilepsy, admissions between August and December 2020 or January and May 2021 were associated with a lower odds of non-routine discharge and decreased LOS compared to admissions between March and July 2020, but this difference was not statistically significant.

Significance:

People with COVID-19 who had epilepsy had a higher odds of non-routine discharge and longer LOS but not higher mortality. Older age (≥65), ventilator use, and higher CCI were associated with COVID-19 mortality in epilepsy. This suggests that older adults with epilepsy and multimorbidity are more vulnerable than those without and should be monitored closely in the setting of COVID-19.

1 |. INTRODUCTION

Coronavirus disease 2019 (COVID-19) has been associated with increased morbidity and mortality in patients with pre-existing conditions, particularly cardiovascular and immunocompromised conditions, and in patients of older age, male sex, and non-White race.^1–3 Prior studies have shown that influenza hospitalizations are more common in people with epilepsy compared to those without.^4,5 The literature regarding how COVID-19 affects people with epilepsy, however, is inconsistent. Several studies report that people with epilepsy experience higher rates of morbidity and mortality from COVID-19,^6,7 whereas other studies do not.^8–10 Specifically, one study found that patients with epilepsy are not more likely to contract or have severe COVID-19; however, the authors suggested that COVID-19 may present differently in those with epilepsy (e.g., less cough, more gastrointestinal symptoms).¹¹ Alternatively, a recent meta-analysis of 13 studies reported increased morbidity and mortality after COVID-19 infection in people with epilepsy.¹² However, many of the studies in this systematic review were done early during the pandemic when effective treatment options and preventative measures were limited (e.g., lack of vaccines and anti-viral agents) and/or did not adjust for important confounders associated with mortality, including comorbidities.¹²

Given the conflicting literature on whether COVID-19 increases the risk of morbidity and mortality in people with epilepsy, the primary aim of our study was to compare in-hospital mortality in people with and without epilepsy. For secondary aims, we compared other measures of in-hospital morbidity including rates of intensive care unit (ICU) admission, need for oxygenation support, length of stay (LOS), and discharge disposition. Because people with epilepsy have a higher risk of premature mortality compared to the general population,¹³ our hypothesis was that individuals diagnosed with COVID-19 and epilepsy would have higher rates of in-hospital mortality.

2 |. MATERIALS AND METHODS

2.1 |. Data source

We conducted a retrospective cohort study in a large multicenter New York State health system, the Mount Sinai Health System (MSHS), comparing outcomes in patients of all ages with and without epilepsy admitted with COVID-19. Patient selection consisted of all MSHS admissions between March 15, 2020, and May 17, 2021. MSHS serves a socioeconomically and racially diverse population in New York City’s five boroughs and the greater New York City area. Data from all five boroughs were included for this study. The MSHS electronic medical record (EMR) data (Epic) were obtained from the Mount Sinai Clinical Intelligence Center (MSCIC).

2.2 |. Study population

People of all ages (to avoid bias) were included and defined as having COVID-19 if they had a confirmed positive or presumptive positive severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) polymerase chain reaction (PCR) test either at presentation or during hospitalization at MSHS. If more than one hospitalization occurred, the first COVID-19-positive hospitalization for each individual was included. Initially all patients admitted into a New York City hospital were tested for COVID-19 if they were symptomatic or if they were exposed to someone with COVID-19 regardless of whether they were otherwise healthy or not. However, within a few weeks of the pandemic, all patients admitted into hospital had to be tested for COVID-19. People with epilepsy were identified at or up to 1 year prior to admission, with International Classification of Diseases Ninth or Tenth revision, Clinical Modification (ICD-9-CM or ICD-10-CM) codes paired with antiseizure medications (ASMs). People were defined as having epilepsy if they had either (1) the following ICD-9-CM (345.x [except 345.2 and 345.3]) or ICD-10-CM (G40.x [except G40.xx1]) epilepsy-related codes in any diagnostic position with or without an ASM or (2) one of the following ICD-9-CM (345.2, 345.3, or 780.3x) or ICD-10-CM (G40.xx1, G41.x, or R56.x) epilepsy or seizure-related codes in any diagnostic position with at least one ASM. This case definition was validated in the MSHS EMR and found to have a sensitivity of 95%, specificity of 72%, positive predictive value of 80%, and negative predictive value of 92%.¹⁴ Of the 17 368 inpatient admissions captured by the MSCIC between March 15, 2020 and May 17, 2021, inpatient admissions were removed from the cohort for the following reasons: second or subsequent inpatient admissions (N = 4773), admissions not associated with a positive or presumptive positive PCR test (N = 2734), missing ICD-9-CM and ICD-10-CM codes (N = 14), missing medication data (N = 11), and missing data associated with study outcomes (N = 3). These exclusions yielded a study cohort of 9833 COVID-19-positive inpatient admissions.

2.3 |. Study outcomes

The primary outcome was in-hospital mortality (died in-hospital or discharged) and the secondary outcomes were the following: (a) discharge disposition (routine discharge to home, non-routine discharges to rehab or subacute care/other acute care facility, hospice, chronic care facility, other facility, or discharge against medical advice); (b) in-hospital LOS; (c) oxygenation support (none, nasal cannula/high flow nasal cannula, bilevel positive airway pressure [BIPAP], continuous positive airway pressure [CPAP], ventilator support); and (d) ICU admission (admitted, not admitted).

2.4 |. Study covariates

Potential confounders included age, sex, race (non-Hispanic White, non-Hispanic Black, other/non-Hispanic Asian/Hispanic, unknown), language (English, Spanish, other, unknown), insurance type, location of hospital site, smoking status, alcohol and substance use, Charlson comorbidity index (CCI), hypertension, and admission period (Table 1).¹⁵ The admission period was stratified into three intervals to highlight the trend in COVID-19 cases in New York: March 2020 to July 2020 and August 2020 to December 2020 (alpha variant period), and January 2021 to May 2021 (delta variant period).¹⁶ The CCI was defined using ICD-9-CM and ICD-10-CM codes.¹⁵ Hypertension was defined using the ICD-9-CM and ICD-10-CM Elixhauser comorbidity index codes for presence of either complicated or uncomplicated hypertension. ICD-9-CM and ICD-10-CM codes recorded up to 1 year prior and up to discharge date were used to determine the presence of comorbidities. Due to the variation in recorded measurements during a given hospital stay, the minimum and maximum of select variables of interest were determined. The following lab tests previously found in the literature to be associated with poor COVID-19 outcomes were included in our study: lowest hemoglobin (Hb), platelets and albumin; and highest white blood cell (WBC) count, creatinine (Cr), procalcitonin, d-dimer, ferritin, C-reactive protein (CRP), alanine transaminase (ALT), aspartate transaminase (AST), alkaline phosphatase (ALP), and prothrombin time (PT).^17,18 COVID-19-related medications received during an individual’s hospital stay were summarized according to their general mechanism of action and/or indication (Supplement S1).

TABLE 1.

Baseline characteristics in COVID-19-positive patients stratified by epilepsy status.

	Overall		Epilepsy		No epilepsy
	N = 9833	%	N = 334	%	N = 9499	%	p-value
Age, mean, SD	64.69	18.02	62.47	17.08	64.77	18.05	.022
Sex							.674
Female	4527	46.04	150	44.91	4377	46.08
Male	5306	53.96	184	55.09	5122	53.92
Race							.204
Hispanic	95	0.97	1	0.30	94	0.99
Non-Hispanic Asian	554	5.63	12	3.59	542	5.71
Non-Hispanic Black	2416	24.57	93	27.84	2323	24.46
Non-Hispanic White	2762	28.09	101	30.24	2661	28.01
Other	3629	36.91	113	33.83	3516	37.01
Unknown	377	3.83	14	4.19	363	3.82
Primary Language							.010
English	7675	78.05	278	83.23	7397	77.87
Spanish	1365	13.88	31	9.28	1334	14.04
Other	621	6.32	15	4.49	606	6.38
Unknown	172	1.75	10	2.99	162	1.71
Insurance Type							<.001
Private	2477	25.19	54	16.17	2423	25.51
Medicaid	3872	39.38	170	50.90	3702	38.97
Medicare	2641	26.86	89	26.65	2552	26.87
Other	755	7.68	16	4.79	739	7.78
Unknown	88	0.89	5	1.50	83	0.87
Smoking Status							<.001
Current use	563	5.73	30	8.98	533	5.61
Previous use	2369	24.09	87	26.05	2282	24.02
Never use	5675	57.71	153	45.81	5522	58.13
Never asked	1226	12.47	64	19.16	1162	12.23
Alcohol Use							.966
Yes	2088	21.24	70	21.02	2019	21.25
No/Not Currently	5986	60.88	203	60.96	5783	60.87
Never	1758	17.88	60	18.02	1698	17.87
Illicit Substance Use							.004
Yes	490	4.98	39	11.64	452	4.76
No/Not Currently	5730	58.27	195	58.21	5535	58.26
Never	3614	36.75	101	30.15	3513	36.98
Date of Admission							.401
March 2020 to July 2020	5012	50.97	171	51.20	4841	50.96
August 2020 to December 2020	1566	15.93	45	13.47	1521	16.01
January 2021 to May 2021	3255	33.10	118	35.33	3137	33.02
Charlson comorbidity index (CCI)
CCI score, median [range]	1	[0, 17]	1	[0. 17]	1	[0, 15]	<.001
Myocardial infarction	521	5.30	22	6.59	499	5.25	.285
Congestive heart failure	984	10.01	42	12.57	942	9.92	.112
Peripheral vascular disease	439	4.46	19	5.69	420	4.42	.270
Cerebrovascular disease	563	5.73	84	25.15	479	5.04	<.001
Dementia	561	5.71	47	14.07	514	5.41	<.001
Chronic pulmonary disease	881	8.96	25	7.49	856	9.01	.337
Rheumatic disease	101	1.03	1	0.30	100	1.05	.266
Peptic ulcer disease	41	0.42	2	0.60	39	0.41	.650
Mild liver disease	268	2.73	11	3.29	257	2.71	.517
Diabetes without complications	2162	21.99	77	23.05	2085	21.95	.632
Diabetes with complications	405	4.12	17	5.09	388	4.08	.364
Paraplegia or hemiplegia	60	0.61	14	4.19	46	0.48	<.001
Renal disease	1128	11.47	58	17.37	1070	11.26	<.001
Cancer	484	4.92	21	6.29	463	4.87	.241
Moderate to severe liver disease	62	0.63	5	1.50	57	0.60	.042
Metastatic carcinoma	118	1.20	7	2.10	111	1.17	.126
Human immunodeficiency virus	135	1.37	8	2.40	127	1.34	.102
Hypertension*	3270	33.26	119	35.63	3151	33.17	.349

Note: Hypertension includes complicated and uncomplicated status.

2.5 |. Statistical analysis

Categorical variables were summarized with frequencies and percentages. Continuous variables were presented with mean and standard deviation (SD) if normally distributed, and the median and interquartile range (IQR) if not normally distributed. Due to low sample size in certain racial categories among admitted people with epilepsy, Hispanic race, non-Hispanic Asian race, and other race were combined into the racial category “other.” Similarly, insurance types “other” and “unknown” were combined. Smoking status, alcohol use, and illicit substance use had greater than 10% missingness. The pattern of missingness for each variable was found to be arbitrary (missing completely at random), likely due to individual clinician EMR documentation methods, and were thus multiply imputed using SAS software’s PROC MI specifying the fully conditional specification (FCS) logistic regression method with “glogit” link function. All covariates intended for regression modeling were used to impute missing variables and 10 imputations were performed in order to achieve >97% relative efficiency in each variable.¹⁹ Univariate comparisons between categorical covariates and epilepsy status were calculated using Fisher’s exact or chi-square tests. The non-parametric Mann–Whitney U test was used for continuous variables against epilepsy status in univariate comparisons. The proportion and cumulative incidence of in-hospital mortality in patients with and without epilepsy were examined across time periods. To test the difference in in-hospital mortality across time in people with and without epilepsy, the data were assessed for trend using the autocorrelation and partial autocorrelation functions. The time-series data had no significant autocorrelation, with no seasonality, and thus two nested analysis of variance models examining equal and varying slopes across time period and epilepsy diagnosis were fitted. The F test was used to compare the two models, and to further determine the nature of the interaction between time period and epilepsy status a non-parametric analysis of covariance (ANCOVA) was performed using “sm.ancova()” from R’s “sm” package. In all statistical analysis regarding the study outcomes apart from ventilation status, unadjusted and adjusted models were generated. The following baseline covariates were included in each multivariable (adjusted) model: age, sex, language, race, smoking status, alcohol use, substance use, insurance type, admission period, ICU admission indicator, ventilation status, CCI score, and hypertension indicator. Logistic regressions were conducted to examine any association of epilepsy with mortality and discharge disposition and adjusted odds ratios (aORs) were calculated. Ordinary least-squares (OLS) regression was performed to determine the association between LOS in COVID-19 and epilepsy. All models were adjusted for age, sex, race, language, alcohol and substance use, insurance type, admission period, ICU admissions status, worst ventilation status, presence of hypertension, and CCI score.

Multivariable logistic regressions were performed to examine factors associated with in-hospital mortality and non-routine discharge disposition in people with epilepsy. A sensitivity analysis was done excluding the following COVID severity proxy ventilation status from the in-hospital mortality model to assess the effect of its correlation with in-hospital mortality. OLS regression was performed to determine the association between LOS in COVID-19 and epilepsy after adjusting for the same variables in the previous models. LOS was log-transformed to correct for skewness. The OLS regression model covariates were interpreted as the percentage change in the average LOS for one unit of change in the covariate (i.e., percentage change = e^β−1).²⁰ Kaplan–Meier survival curves for people with and without epilepsy were calculated to illustrate the association with LOS and presence of epilepsy after censoring for individuals who were discharged against medical advice, died in hospital, or surpassed 100 days (N = 2). To test for equality over the stratification variable, epilepsy diagnosis, the log-rank test was conducted. For all analyses, the level of significance was 0.05 and Statistical Analysis Software (SAS) version 9.4 was used (SAS Institute Inc).

2.6 |. Standard protocol approval

The study was approved by the Icahn School of Medicine Institutional Review Board (IRB) under protocol number IRB-20–03271.

3 |. RESULTS

3.1 |. Cohort baseline characteristics

Between March 2020 and May 2021, we identified 334 people (44.91% female) with and 9499 (46.08% female) without epilepsy with confirmed positive (n = 9481) or presumptive positive (n = 352) SARS-CoV-2 PCR (Table 1) who were admitted to MSHS. Mean age was lower in people with epilepsy (62.47, SD: 17.08 vs 64.77, SD: 18.05, p = .022). The majority were English speaking, but more people without epilepsy were Spanish speaking than people with epilepsy (14.04% vs 9.28%, p = .010). More people with epilepsy were on Medicaid (50.90% vs 38.97%, p < .001), whereas more people without epilepsy had private insurance (25.51% vs 16.17%, p < .001). After multiple imputations of missing data for smoking status (N missing = 1634, 16.62%), alcohol use (N missing = 3008, 30.59%), and illicit drug use (N missing = 3431, 34.89%) it was found that more people with epilepsy currently used tobacco products (8.98% vs 5.61%, p < .001) and reported higher illicit drug use than people without epilepsy (11.64% vs 4.76%, p = .004). We identified 334 with epilepsy, of whom 93.7% were taking at least one ASM.

People with epilepsy diagnosed with COVID-19 had significantly worse clinical characteristics than those without in unadjusted analyses. People with epilepsy had higher median maximum heart rates (p < .001), ferritin (p = .003), CRP (p < .001), d-dimer (p < .001), creatinine (p = .002), and WBC levels (p < .001), and significantly lower median minimum Hb level (p < .001) and platelet counts (p < .001) (S1). In addition, people with epilepsy were more likely to be treated with antibiotics (p < .001) and analgesics (p = .013) than people without epilepsy (see Table S1).

3.2 |. Hospitalization outcomes

In univariate analysis, people with epilepsy were more likely to have an ICU admission (39.22% vs 17.69, p < .001) and have a longer median LOS compared to people without epilepsy (12 days vs 7 days, p < .001) (Table 2). People with epilepsy were also more likely to receive ventilator support (37.72% vs 14.34%, p < .001) and have in-hospital mortality (29.64% vs 19.88%, p < .001) but were less likely to be discharged home than those without epilepsy (29.64% vs 57.37%, p < .001).

TABLE 2.

Outcomes in COVID-19-positive patients stratified by epilepsy status.

	Overall		Epilepsy		No epilepsy
	N	%	N	%	N	%	p-value
Worst Ventilation Status							<.001
None	2057	20.92	52	15.57	2005	21.11
Nasal flow cannula/mask	3399	34.57	67	20.06	3332	35.08
Non rebreather/high flow nasal cannula	2047	20.82	67	20.06	1980	20.84
BIPAP/CPAP	842	8.56	22	6.59	820	8.63
Ventilator	1488	15.13	126	37.72	1362	14.34
Discharge disposition							<.001
Home	5549	56.43	99	29.64	5450	57.37
AMA	155	1.58	9	2.69	146	1.54
Rehab/subacute care	463	4.71	25	7.49	438	4.61
Other acute care facility	654	6.65	46	13.77	608	6.40
Other	82	0.83	1	0.30	81	0.85
Chronic care facility	813	8.27	50	14.97	763	8.03
Hospice	130	1.32	5	1.50	125	1.32
Died	1987	20.21	99	29.64	1888	19.88
Death Status							<.001
Alive	7846	79.79	235	70.36	7611	80.12
Died in hospital	1987	20.21	99	29.64	1888	19.88
Overall length of stay, median [Q1, Q3]	7	[3, 13]	12	6, 27	7	[3, 12]	<.001
ICU admission	1811	18.42	131	39.22	1680	17.69	<.001
ICU length of stay, median [Q1, Q3]	6	[2, 13]	11	3, 23	6	[2, 13]	<.001

Abbreviations: AMA, against medical advice; COVID-19, coronavirus disease 2019; ICU, intensive care unit.

3.3 |. In-Hospital mortality

Figure 1A shows the proportion of in-hospital mortality over time in both groups with higher mortality in those with epilepsy during New York City’s second wave from November 2020 to March 2021. Although the contrasting spike in hospital mortality around January 2021 between people with and without epilepsy is notable, the non-parametric ANCOVA test for equality was insignificant (p = .680). Figure 1B shows the cumulative proportion of in-hospital mortality over time. In the multivariable logistic regression analysis, there was no statistically significant difference between those with or without epilepsy (Table 3). A separate model that did not include COVID-19 severity proxy variables did not change the outcomes (see Table S2).

FIGURE 1.

(A) Percentage of in-hospital mortality stratified by epilepsy diagnosis in coronavirus disease 2019 (COVID-19)-positive people. (B) Cumulative percentage of in-hospital mortality stratified by epilepsy diagnosis in coronavirus disease 2019 (COVID-19)-positive people.

TABLE 3.

Logistic regression of (A) in-hospital mortality; (B) discharge disposition in COVID-19-positive patients.

			Unadjusted model			Adjusted model
	Epilepsy (%)	No epilepsy (%)	Odds ratio	95% CI	p-value	Odds ratio	95% CI	p-value
(a) In-Hospital Mortality
Alive	70.36	80.12	Reference	Reference		Reference	Reference
Died in hospital	29.64	19.88	1.70	1.34, 2.16	<.0001	0.84	0.60, 1.17	.301
(b) Discharge Disposition
Routine	29.64	57.37	Reference	Reference		Reference	Reference
Non-routine	40.72	22.75	3.47	2.66, 4.51	<.0001	2.73	2.00, 3.72	<.0001

Note: Unadjusted model tests association between (a) in-hospital mortality or (b) discharge disposition, and epilepsy without controlling for any other factors. CI = confidence interval; COVID-19, coronavirus disease 2019.

Adjusted model after adjusting for age, sex, language, race, alcohol and substance use, insurance type, admission date, intensive care unit admission status, worst ventilation status, Charlson comorbidity score, and hypertension.

Factors associated with higher odds of mortality in people with epilepsy were increasing age (aOR 1.05, 95% confidence interval [CI] 1.03–1.08; p < .001), ventilator support (aOR 7.18, 95% CI 3.12–16.48; p < .001), and higher CCI score (aOR 1.18, 95% CI 1.04–1.34; p = .010) (Table S3).

3.4 |. Discharge disposition

In our multivariable logistic regression analysis, people with epilepsy had a higher odds of non-routine discharge (aOR 2.73, 95% CI 2.00–3.72; p < .001) as compared to those without epilepsy (Table 3). Factors associated with higher odds of non-routine discharge in epilepsy were age (aOR 1.03, 95% CI 1.01–1.05; p = .008) and ventilator support (aOR 4.04, 95% CI 1.45–11.29; p = .008). In addition, people with epilepsy had a lower odds of non-routine discharge if they were admitted between January 2021 and May 2021 (aOR 0.27, 95% CI 0.09–0.81; p = .020) (Table S4).

3.5 |. Hospital length of stay

People with epilepsy had greater cumulative probability of having a LOS longer than people without epilepsy (p < .001; Figure 2). For example, people with epilepsy had a 72% chance of having a LOS longer than 7 days as compared to those without epilepsy who had a 51% chance of stay longer than 7 days (7 days being the median LOS for the entire cohort). In our multivariable regression analysis, on average, people with epilepsy had longer LOS than those without epilepsy (32.50%, 95% CI 22.22%–43.64%’ p < .001; Table 4).

FIGURE 2.

Kaplan–Meier survival curve of length of stay stratified by epilepsy diagnosis in coronavirus disease 2019 (COVID-19)-positive people.

TABLE 4.

Ordinary least-squares regression results after log transformation of length of stay in COVID-19-positive patients.

	Unadjusted model			Adjusted model
	% change	95% CI	p-value	% change	95% CI	p-value
Length of stay (days)
Epilepsy (median 12 days)	66.77%	57.47%, 76.07%	<.0001	32.50%	22.22%, 43.64%	<.0001
No epilepsy (median 7 days)	Reference	Reference		Reference	Reference

Note: Unadjusted model tests association between length of stay and epilepsy without controlling for any other factors. CI = confidence interval; COVID-19, coronavirus disease 2019.

Adjusted model after adjusting for age, sex, language, race, alcohol and substance use, insurance type, admission date, intensive care unit admission status, worst ventilation status, Charlson comorbidity score, and hypertension.

Within the epilepsy cohort, factors associated with a percentage increase in average LOS were admission to the ICU (39.11%, 95% CI 7.17%–80.57%; p = .013), BIPAP/CPAP use (48.86%, 95% CI 1.38%–118.58%; p = .042), ventilator support (95.67%, 95% CI 50.86%–153.79%; p < .001) and increasing CCI score (9.71%, 95% CI 5.47%–14.12%; p < .001). Epilepsy people admitted after July 2020 had a decreased average LOS compared to those admitted prior (Table S5).

4 |. DISCUSSION

Determining which factors are associated with poor outcomes after COVID-19 infection for inpatients with pre-existing conditions can help guide therapeutic decision-making and risk stratification. This is especially important because the literature surrounding COVID-19 outcomes for people with epilepsy is limited and conflicted. The goal of our study was to investigate mortality and relevant in-hospital outcomes in people with epilepsy from a diverse, multicenter health system in New York City. We found that people with epilepsy and COVID-19 were slightly younger and were more likely to have Medicaid insurance compared to those without epilepsy, but we did not observe significant racial or sex differences. People with epilepsy and COVID-19 had a higher frequency of ICU admission, longer mean hospital LOS, and were less likely to be discharged home than those without epilepsy, irrespective of many important confounders (e.g., smoking status, alcohol use). Of interest, mortality was higher among people with epilepsy than people without epilepsy, but this difference was not significant after adjusting for demographic and clinical characteristics. Of note, risk factors for severe COVID-19 in our population seem to mirror the general population, likely explaining why there is no difference in mortality after adjusting for confounders. Although other studies have similarly found that in the general population, older adults with multimorbidity are more vulnerable to COVID-19 infection, our study provides a unique perspective in studying a diverse epilepsy cohort and adjusting for multiple comorbidities not previously adjusted for.

Despite no difference in mortality, our study suggests that people with epilepsy had a higher incidence of poor in-hospital outcomes. These results align with findings from previous studies^8,9 but are discordant with a recent meta-analysis of 13 studies that found that COVID-19 was associated with higher rates of mortality in those with epilepsy.¹² Although these findings appear contradictory, it should be noted that our study also analyzed the incidence of mortality throughout the different phases of the pandemic through May 2021. The initial wave, characterized by a health care system struggling to cope with an unprecedented volume of patients, was marked by significant in-hospital mortality (Figure 1). The second wave in the winter of 2021 was marked by a decline, followed by a plateau, in mortality (Figure 1), likely in part due to reduced overall patient volume, decreased strain on the health care system, and vaccine efficacy starting in late 2020/early 2021.²¹ A significant proportion of the studies cited in the meta-analysis analyzed data from the beginning of the pandemic, during the spring and fall of 2020, which may explain differences in mortality. In addition, several studies in the recent meta-analysis either did not adjust for confounders or adjusted only for age and sex (omitting pre-existing comorbidities). Our purpose in adjusting for individual demographics and clinical features was to make sure we accounted for important factors that are known to be associated with mortality in patients with COVID-19. People with epilepsy in our study who died during their hospitalization often had well-known risk factors associated with COVID-19 severity and mortality (e.g., older age and need for intubation).^17,18 Of note, we found that insurance status and male sex were not associated with mortality in people with COVID-19, which was not consistent with earlier work.^1–3

We found that people with epilepsy admitted with COVID-19 had significantly longer LOS and were less likely to be discharged home, which in prior studies have been found to be associated with increased COVID-19 severity.²² Our findings that people with epilepsy had worse in-hospital outcomes (e.g., ICU admission, intubation) and longer LOS suggest that they experienced more severe COVID-19. Patients with severe COVID-19 are often more likely to have abnormal inflammatory or coagulation markers.¹⁸ Indeed markers such as d-dimer, ferritin, CRP, and platelets were all more likely to be abnormal (e.g., higher d-dimer, ferritin, and CRP, and lower platelets) in people with COVID-19 and epilepsy compared to those without epilepsy in our study.

Our study has several strengths, including the stratification by pandemic time-period. Stratification by time allowed us to contextualize our results in the setting of real-world changes such as the introduction of vaccines, the evolution of SARS-CoV2 variants, and improvements in treatments for COVID-19 (e.g., steroids, antiviral agents, monoclonal antibody, and anticoagulation). Similarly, this study includes a large, racially and ethnically diverse population, which increases the generalizability of our findings. Moreover, all COVID-19 cases were confirmed by PCR, or if PCR was not conclusive, by an infectious diseases specialist using symptomatology and chest X-ray deemed to be consistent with COVID-19. This limits the possibility of error associated with self-report or misdiagnoses, especially early in the pandemic when testing was limited.

Although our study was focused primarily on mortality, we also sought to investigate predictors of other poor in-hospital outcomes. As people with a history of COVID-19 undergo longitudinal follow-up, details from the hospital course are recognized increasingly as important predictors of the long-term effects of this illness.²³ As such, it is valuable to understand what factors may be associated with prolonged hospitalizations, ICU admission, and non-routine discharge. Finally, this study adjusted for both traditional confounders (e.g., age, sex, race) and clinical confounders (such as smoking history and other medical comorbidities). Uniquely, this allowed us to consider both some of the pre-existing condition of epilepsy and some health-related behaviors that may also impact COVID-19 severity (e.g., smoking).²⁴

Our study is not without limitations. Although all those included in the study had a confirmed PCR diagnosis of COVID-19, we were not able to determine if the death was indeed due to COVID-19. ICD-9-CM and ICD-10-CM diagnosis codes were used to identify patients with epilepsy, which could misclassify epilepsy status, including those with acute seizures during hospitalization. Although our case definition has the potential to misattribute acute seizures as diagnosed epilepsy, our case definition has been validated within our health care system and has relatively good accuracy mitigating this risk.¹⁴ Moreover, a study in our health care system found that acute seizures due to COVID-19 are possible but rare.²⁵ We also use validated case definitions for all comorbidities analyzed.¹⁵ Unfortunately we were unable to adjust for some laboratory findings due to missing lab data for some people. We also did not have information about epilepsy type (e.g., convulsive seizures), epilepsy severity, seizure frequency, or ASM type or dose. However, we have found that acute seizures due to COVID-19, although possible, are rare.²⁵ Future work, however, would benefit from considering markers of epilepsy severity. Similarly, although we have a large data set, data were available only from the year prior to admission up until the date of discharge. As such, despite the use of a validated case definition, there may be well-controlled cases of epilepsy that did not have ICD codes for seizures or epilepsy and could therefore have been included in the controls. This type of misclassification (i.e. if some patients with mild epilepsy were classified as not having epilepsy) could lead to falsely elevated ORs since the group with epilepsy would include those with more severe epilepsy. It would have been interesting to be able to have 1 year follow-up on all patients to examine longer-term survival with a survival analysis and also to have at least 1 year of data on comorbidities for all participants. Fortunately the varied length of time for the comorbidity ascertainment would be non-differential. In addition, we captured data from all care settings preadmission and thus included patients with and without well-controlled epilepsy. In addition, although we know where patients were eventually discharged, we do not have information on where they originated prior to hospitalization, which may have influenced discharge disposition. Finally, although this study was performed within a large health care system in New York City, the findings may not be generalizable to other geographic regions or to persons who had COVID-19 in the community and were not hospitalized.

In conclusion, we found that people with epilepsy had worse in-hospital outcomes, albeit without increased mortality, compared to those without epilepsy, suggestive of increased COVID-19 severity among those with epilepsy. Although the observed similarity in mortality is reassuring, these other poor outcomes in people with epilepsy may predispose this population to post-acute sequelae of COVID-19 (also referred to as “long COVID”), which has become an important medical and public health concern. Understanding predictors of these outcomes in epilepsy will continue to be important to create targeted care-pathways to try to improve the management of COVID-19 in this at-risk population.

Supplementary Material

SUPPORTING INFORMATION

Additional supporting information can be found online in the Supporting Information section at the end of this article.

Key points.

• People with coronavirus disease 2019 (COVID-19) and epilepsy had a higher odds of non-routine discharge and longer length of-stay but not higher mortality.

• People with COVID-19 and epilepsy had higher inflammatory markers than those without.

• Older age (≥65), ventilator use, and higher comorbidity index were associated with COVID-19 mortality in epilepsy.

• Older adults with epilepsy are more vulnerable than those without and should be monitored closely in the setting of COVID-19.