Older Adults with Epilepsy and COVID-19: Outcomes in a Multi-Hospital Health System

Abstract

Background:

Coronavirus disease 2019 (COVID-19) is associated with high rates of mortality and morbidity in older adults, especially those with pre-existing conditions. There is little work investigating how neurological conditions affect older adults with COVID-19. We aimed to compare in-hospital outcomes, including mortality, in older adults with and without epilepsy.

Methods:

This retrospective study in a large multicenter New York health system included consecutive older patients (age ≥65 years) either with or without epilepsy who were admitted with COVID-19 between 3/2020–5/2021. Epilepsy was identified using a validated International Classification of Disease (ICD) and antiseizure medication-based case definition. Univariate comparisons were calculated using Chi-square, Fisher‟s exact, Mann-Whitney U, or Student‟s t-tests. Multivariable logistic regression models were generated to examine factors associated with mortality, discharge disposition and length of stay (LOS).

Results:

We identified 5384 older adults admitted with COVID-19 of whom 173 (3.21%) had epilepsy. Mean age was significantly lower in those with (75.44, standard deviation (SD): 7.23) compared to those without epilepsy (77.98, SD: 8.68, p=0.007). Older adults with epilepsy were more likely to be ventilated (35.84% vs. 16.18%, p<0.001), less likely to be discharged home (21.39% vs. 43.12%, p<0.001), had longer median LOS (13 days vs. 8 days, p<0.001), and had higher in-hospital death (35.84% vs. 28.29%, p=0.030) compared to those without epilepsy. Epilepsy in older adults was associated with increased odds of in-hospital death (adjusted odds ratio (aOR), 1.55; 95% CI 1.12–2.14, p=0.032), non-routine discharge disposition (aOR, 3.34; 95% CI 2.21–5.03, p<0.001), and longer LOS (46.46% 95% CI 34%−59%, p<0.001).

Conclusions:

In models that adjusted for multiple confounders including comorbidity and age, our study found that epilepsy was still associated with higher in-hospital mortality, longer LOS and worse discharge dispositions in older adults with COVID-19 higher in-hospital mortality, longer LOS and worse discharge dispositions in older adults with COVID-19. This work reinforces that epilepsy is a risk factor for worse outcomes in older adults admitted with COVID-19. Timely identification and treatment of COVID-19 in epilepsy may improve outcomes in older people with epilepsy.

Introduction:

The coronavirus disease 2019 (COVID-19) pandemic has caused disproportionate morbidity and mortality in older adults, especially those with pre-existing conditions.^{1, 2 3} While certain pre-existing conditions such as diabetes and cardiovascular disease have been associated with COVID-19 severity in older adults,² the contribution of other chronic diseases to mortality and morbidity have been less well studied. There is little work investigating how neurological conditions, specifically epilepsy, affect older adults with COVID-19.

Epilepsy is among the most common chronic neurologic disorders and is most prevalent in older adult who comprise up to 25% of new epilepsy diagnoses.⁴ As such, epilepsy affects a significant portion of our aging population, who are themselves already at increased risk from COVID-19.¹ The literature on the association between epilepsy and hospital outcomes with COVID-19 in those of all ages is variable: some studies suggest that those with epilepsy have higher rates of morbidity and mortality,^{5,6, 7} while others report no added risk in individuals with epilepsy and COVID-19.^{8,9, 10} How these findings may differ in older adults who are already at higher risk of death from COVID-19 is unknown.

We sought to understand whether epilepsy was associated with in-hospital mortality in older adults hospitalized with COVID-19. In addition, we sought to understand the relationship between age and epilepsy in those with COVID-19 and other hospitalization outcomes including length of stay (LOS) and discharge disposition. Because epilepsy is associated with premature mortality, we hypothesized that older adults with epilepsy would experience worse in-hospital outcomes compared to older adults without epilepsy.¹¹

Methods

Data Source

Electronic medical record (EMR) data from the Mount Sinai Health System‟s (MSHS) Clinical Intelligence Center (MSCIC) were used in a retrospective cohort study to compare outcomes in people with COVID-19 with and without epilepsy between March 15, 2020 and May 17, 2021. The MSCIC gathered data across five participating MSHS hospitals, including admissions data, sociodemographic information, diagnostic codes, as well as clinical information (e.g., medications, laboratories). The catchment area for the MSHS includes New York City‟s (NYC) five boroughs and the greater New York City area. As a result, the MSHS serves a primarily urban but socioeconomically and racially diverse population.

Study Cohort

The study cohort consisted of 9,833 hospitalized individuals with COVID-19. Eligible patients had to have a positive or presumptive positive SARS-CoV-2 polymerase chain reaction (PCR) test (96.42% had a positive PCR and 3.58% were presumed positive) in the emergency department (ED) or during their inpatient admission to the MSHS, regardless of age. In the beginning of the pandemic, due to shortage of tests, all patients admitted to NYC hospitals were tested for COVID-19 if symptomatic or if they reported a COVID-19 exposure independent of any co-existing risk factors. Within a few weeks, however, all patients entering the hospital were tested regardless of symptoms or exposure. For those with more than one admission during the study period, the first COVID-19 positive inpatient admission was defined as the index admission.

International Classification of Diseases ninth or tenth revision clinical modification classification codes (ICD-9-CM/ICD-10-CM) with antiseizure medications (ASM) one year prior and up to admission were used to identify people with epilepsy. The case definition for epilepsy consisted of a patient having either (1) an epilepsy specific diagnostic code (ICD-9-CM [345.x (except 345.2 and 345.3)] or ICD-10-CM code [G40.x (except G40.xx1)] in any diagnostic position) +/− an ASM or (2) a status epilepticus or seizure code (ICD-9-CM (345.2, 345.3, or 780.3x) or ICD-10-CM codes (G40.xx1, G41.x, or R56.x) in any diagnostic position) with at least one ASM (Supplement 1 (S1)). This epilepsy case definition has been validated in our health system (sensitivity 95%, specificity 72%, positive predictive value 80%, negative predictive value 92%).¹² Individuals were excluded from the study if they had unknown/missing (1) discharge dates; (2) ICD codes; (3) medication information; or (4) sex (see Figure 1).

Figure 1.

Diagram for the COVID-19 positive cohort selection

Study Measures

Outcomes

The primary outcomes of interest were in-hospital mortality, discharge disposition, and length of stay (LOS). For the discharge disposition analysis, we excluded patients who had a discharge status as “expired.” Discharge disposition was then categorized as routine or non-routine, where routine discharges were defined as discharge home and non-routine included discharges anywhere other than home. Examples of non-routine discharge include but are not limited to: discharges to skilled nursing facilities, rehabilitation facilities, chronic care facilities, and discharge against medical advice (AMA).

Secondary outcomes were ventilation status (none/nasal cannula/high flow nasal cannula), bilevel positive airway pressure (BIPAP), continuous positive airway pressure (CPAP), ventilator support; and intensive care unit (ICU) admission. Individuals who did not have ventilation status information and who were not admitted to the ICU were assumed to be on room air for the duration of their admission.

Covariates

Age, sex, race or ethnicity, primary language, insurance type, hospital, smoking status, alcohol and substance use, Charlson comorbidity index (CCI), hypertension, and period of admission (March 2020 – July 2020, August 2020 – December 2020, and January 2021 – May 2021) were obtained. The CCI was defined using standardized validated ICD-9-CM and ICD-10-CM codes.¹³

Blood pressure, oxygen level, weight, and other vital signs through admission were captured from the EMR. Laboratory tests found in previous studies to be associated with poor COVID-19 outcomes were summarized over the hospital stay and by reporting the median of the individual‟s highest (white blood cell count, creatinine, procalcitonin, D-dimer, ferritin, C reactive protein, alanine transaminase, aspartate transaminase, alkaline phosphatase, prothrombin time (PT)), and lowest (hemoglobin, platelet count, and albumin) observed values.^14,15 Medications prescribed during the hospital stay were summarized by pharmacologic class as it relates to COVID-19 treatment (S1).

Statistical Analysis

All outcomes and covariates were described using summary statistics: frequencies and percentages for categorical variables, and mean and standard deviation (SD) or median and interquartile range (IQR) for continuous variables. Due to low sample size among patients with epilepsy, Hispanic, Asian, Alaska Native/American Indian, Native Hawaiian and Pacific Islander, and people of reported “other” or unknown race were included as „other race‟ for analytic purposes. Variables with greater than 10% missingness (alcohol use, illicit substance use, smoking status) were multiply imputed. These variables had no clear pattern of missingness (missingness appeared to be random) likely due to individual clinician EMR documentation methods. These variables were then 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.¹⁶ We compared discharge disposition, in-hospital mortality, and LOS in those with and without epilepsy over 65 years of age and in those with epilepsy under and over 65 years of age.

Multivariable logistic regression was used to investigate the association of epilepsy with mortality and discharge disposition in older adults. Kaplan-Meier curves were plotted to show how the proportions of patients‟ LOS changed over time after censoring patients who died in the hospital, were discharged AMA, or had stays exceeding 100 days (N=2) (Figure 2). The log-rank test was performed and p-values were corrected for multiple comparisons using the Bonferroni correction. Ordinary least squares regression (OLS) was performed to examine the association between LOS in older patients with and without epilepsy who were COVID-19 positive (Table 2). To correct for skewness, LOS was log-transformed (LOS shorter than one day were considered as one day). OLS model covariates represent the percentage change in the LOS for one unit of change in the covariate (i.e., percentage change = e^β−1).

Figure 2.

Kaplan-Meier survival curve of length of stay in COVID-19 positive older adults stratified by epilepsy status

Table 2.

Multivariable logistic regression of in-hospital mortality and discharge disposition in older adults with COVID-19; ordinary least squares regression with log-transformed length of stay in older adults with COVID-19

Variable	Adults 65 or older
	In-Hospital Mortality		Non-Routine Discharge Disposition		Length of Stay
	AOR (95% CI)	P-value	AOR (95% CI)	P-value	% Difference from Mean (95% CI)	P-value
Epilepsy
No epilepsy	Reference		Reference		Reference
Epilepsy	1.55 (1.12, 2.14)	0.032	3.34 (2.21, 5.03)	<0.001	46.46% (34%, 59%)	<0.001
Age	1.04 (1.03, 1.05)	<0.001	1.05 (1.05, 1.06)	<0.001	0.32% (0.08%, 0.57%)	0.044
Sex
Female	Reference		Reference		Reference
Male	1.40 (1.24, 1.58)	<0.001	1.14 (1.00, 1.31)	0.232	−0.84% (−5.16%, 3.48%)	1
Language
English	Reference		Reference		Reference
Spanish	1.08 (0.89, 1.30)	1	0.87 (0.70, 1.07)	0.772	−6.44% (−13.11%, 0.22%)	0.184
Other	1.35 (1.11, 1.63)	0.008	0.82 (0.65, 1.04)	0.392	−2.23% (−9.35%, 4.88%)	1
Race
Non-Hispanic White	Reference		Reference		Reference
Non-Hispanic Black	1.01 (0.86, 1.20)	1	1.20 (1.01, 1.44)	0.176	−2.11% (−7.99%, 3.78%)	1
Other	1.06 (0.91, 1.23)	1	0.82 (0.69, 0.96)	0.064	−2.76% (−8.06%, 2.54%)	1
CCI	1.07 (1.03, 1.10)	<0.001	1.09 (1.05, 1.13)	<0.001	9.71% (8.54%, 10.89%)	<0.001

Note: CCI = Charlson Comorbidity Index, CI = confidence interval, AOR = adjusted odds ratio

All multivariable models were adjusted for age, sex, language, race or ethnicity and CCI. The odds ratios from the multivariable models are reported as adjusted odds ratios (aOR). For all analyses except Kaplan-Meier curves and log-rank tests, Statistical Analysis Software (SAS) version 9.4 was used (SAS Institute Inc., Cary, NC, USA). R 4.1.2 (Vienna, Austria) was used to do Kaplan–Meier curves and log-rank tests. The level of significance in all tests was measured at 0.05.

To test the difference in in-hospital mortality across time in adults 65 and older with and without epilepsy, we assessed for trend using the autocorrelation and partial autocorrelation functions. This analysis showed that the time series data had no significant autocorrelation, with no seasonality. We then fitted two nested analysis of variance models examining equal and varying slopes across the time period and epilepsy diagnosis. The F-test was used to compare the two models. Finally, to further analyze the the interaction between time period and epilepsy status a non-parametric ANCOVA was performed using “sm.ancova()” from R‟s “sm” package.

Standard Protocol Approval

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

Data Availability Statement

Researchers with the appropriate credentials, IRB training, and certification can apply to the senior authors to request access after a data use agreement has been executed with Mount Sinai.

Results

Baseline characteristics

Between March 2020 and May 2021, we identified 5384 individuals ages 65 years or older admitted with COVID-19; 173 (3.21%) of these older adults also had an epilepsy diagnosis (Figure 1). Table 1 shows the patient and hospital characteristics for this cohort. Mean age was statistically significantly lower for those with epilepsy compared to those without epilepsy (75.44, SD: 7.23 vs. 77.98, SD: 8.68, p=0.007). The majority of patients were English speaking in both groups, but a larger proportion of individuals were Spanish speaking in those without epilepsy (15.28% vs. 9.83%, p=0.039). A larger proportion of older adults with epilepsy had cerebrovascular disease (29.48% vs. 6.54, p<0.001), congestive heart failure (18.50% vs. 13.05%, p=0.037), dementia (20.81% vs. 9.54%, p<0.001), and either paraplegia or hemiplegia (1.73% vs. 0.42%, p=0.045) compared to those without epilepsy. Older adults with epilepsy were also more likely to be ventilated (35.84% vs. 16.18%, p<0.001), were less likely to be discharged home (21.39% vs. 43.12%, p<0.001), had a longer median LOS (13 days vs. 8 days, p<0.001) and were more likely to die in the hospital (35.84% vs. 28.29%, p=0.030) compared to those without epilepsy (Table 1). Older adults with epilepsy also more frequently had inflammatory or coagulation biomarkers previously reported to be associated with more severe COVID-19 than those without epilepsy (S2).^{14, 15}

Table 1.

Baseline characteristics and outcomes in COVID-19 positive older adults stratified by epilepsy status

	Overall		Epilepsy		No epilepsy		P-value
	N	%	N	%	N	%
	5384	100.00	173	3.21	5211	96.79
Baseline Characteristics
Age (mean, SD)	77.90	8.65	75.44	7.23	77.98	8.68	0.007
Sex							0.891
Female	2704	50.22	86	49.71	2618	50.24
Male	2680	49.78	87	50.29	2593	49.76
Race							0.220
Hispanic	51	0.95	1	0.58	50	0.96
Non-Hispanic Asian	288	5.35	6	3.47	282	5.41
Non-Hispanic Black	1257	23.35	50	28.9	1207	23.16
Non-Hispanic White	1763	32.75	59	34.10	1704	32.70
Other	1826	33.92	48	27.75	1778	34.12
Unknown	199	3.70	9	5.20	190	3.65
Primary language							0.039
English	3999	74.28	144	83.24	3855	73.98
Spanish	813	15.09	17	9.83	796	15.28
Other	457	8.49	8	4.62	449	8.62
Unknown	115	2.14	4	2.31	111	2.13
Insurance type							0.069
Private	931	17.29	27	15.61	904	17.35
Medicaid	1862	34.58	72	41.62	1790	34.35
Medicare	2236	41.53	70	40.46	2166	41.57
Other	332	6.17	3	1.73	329	6.31
Unknown	23	0.43	1	0.58	22	0.42
Smoking status							0.002
Current use	194	3.60	6	3.47	188	3.61
Previous use	1553	28.84	55	31.79	1498	28.75
Never use	2937	54.55	74	42.77	2863	54.94
Never asked	700	13.00	38	21.97	662	12.70
Alcohol use							0.845
Yes	872	16.20	26	15.03	846	16.23
No/Not currently	3590	66.68	116	67.05	3474	66.67
Never	922	17.12	31	17.92	891	17.10
Illicit Substance use							0.216
Yes	116	2.15	10	5.78	106	2.03
No/Not currently	3438	63.86	111	64.16	3327	63.85
Never	1830	33.99	52	30.06	1778	34.12
Date of admission							0.438
March 2020-July 2020	2747	51.02	86	49.71	2661	51.07
August 2020-December 2020	856	15.90	23	13.29	833	15.99
January 2021-May 2021	1781	33.08	64	36.99	1717	32.95
Charlson comorbidity index
Overall index, median [range]	1	[0, 17]	1	[0, 17]	1	[0, 15]	0.001
Myocardial infarction	387	7.19	16	9.25	371	7.12	0.286
Congestive heart failure	712	13.22	32	18.50	680	13.05	0.037
Peripheral vascular disease	287	5.33	10	5.78	277	5.32	0.789
Cerebrovascular disease	392	7.28	51	29.48	341	6.54	<0.001
Dementia	533	9.90	36	20.81	497	9.54	<0.001
Chronic pulmonary disease	518	9.62	14	8.09	504	9.67	0.488
Rheumatic disease	56	1.04	1	0.58	55	1.06	1
Peptic ulcer disease	29	0.54	2	1.16	27	0.52	0.239
Mild liver disease	109	2.02	3	1.73	106	2.03	1
Diabetes without complications	1263	23.46	49	28.32	1214	23.30	0.125
Diabetes with complications	242	4.49	8	4.62	234	4.49	0.933
Paraplegia or hemiplegia	25	0.46	3	1.73	22	0.42	0.045
Renal disease	718	13.34	27	15.61	691	13.26	0.372
Cancer	335	6.22	14	8.09	321	6.16	0.301
Moderate - severe liver disease	19	0.35	1	0.58	18	0.35	0.463
Metastatic carcinoma	73	1.36	3	1.73	70	1.34	0.509
HIV	47	0.87	3	1.73	44	0.84	0.191
Hypertension *	2210	41.05	76	43.93	2134	40.95	0.433
Outcomes
Worst ventilation status							<0.001
None	807	14.99	21	12.14	786	15.08
Nasal flow cannula/mask	1819	33.79	36	20.81	1783	34.22
Non-rebreather/High flow nasal cannula	1267	23.53	40	23.12	1227	23.55
BIPAP/CPAP	586	10.88	14	8.09	572	10.98
Ventilator	905	16.81	62	35.84	843	16.18
Discharge disposition							<0.001
Home	2284	42.42	37	21.39	2247	43.12
AMA	43	0.80	2	1.16	41	0.79
Rehab/subacute care	241	4.48	12	6.94	229	4.39
Other acute care facility	445	8.27	23	13.29	422	8.10
Other	37	0.69	0	0	37	0.71
Chronic care facility	679	12.61	34	19.65	645	12.38
Hospice	119	2.21	3	1.73	116	2.23
Died	1536	28.53	62	35.84	1474	28.29
Vital status							0.03
Discharged alive	3848	71.47	111	64.16	3737	71.71
Died in hospital	1536	28.53	62	35.84	1474	28.29
Overall length of stay (median [Q1, Q3])	8	[4, 14]	13	[7, 26]	8	[4, 14]	<0.001
ICU admission	969	18.00	60	34.68	909	17.44	<0.001
ICU length of stay (median [Q1, Q3])	6.5	[3, 13]	12	[4, 22.5]	6	[2, 13]	0.008

Note: AMA = against medical advice, ICU = intensive care unit,

^*Hypertension includes complicated and uncomplicated hypertension

In-Hospital Mortality

In multivariable models, epilepsy was associated with increased odds of in-hospital death (aOR, 1.55; 95% CI 1.12–2.14, p=0.032) in older hospitalized adults with COVID-19 (Table 2). Increasing age (aOR, 1.04; 95% CI 1.03–1.05, p<0.001), male sex (aOR, 1.40; 95% CI 1.24–1.58, p<0.001), and higher comorbidity burden (aOR, 1.07; 95% CI 1.03–1.10, p<0.001 per one point increase in CCI) were also associated with mortality. In supplementary analyses looking at patients with epilepsy of all ages, age over 65 years was the strongest predictor of in-hospital mortality (aOR, 2.04; 95% CI 1.22–3.40, p=0.012) (S3).

Discharge Disposition

In our multivariable models, epilepsy (aOR, 3.34; 95% CI 2.21–5.03, p<0.001), increasing age (aOR, 1.05; 95% CI 1.05–1.06, p<0.001 per one-year increase in age), and higher comorbidity burden (aOR, 1.09; 95% CI 1.05–1.13, p<0.001 per one-point increase in CCI) were all significantly associated with non-routine discharge (Table 2).

Length of Stay

The following variables were associated with longer LOS in the multivariable logistic regression: epilepsy (46.46% longer LOS; 95% CI 34%−59%, p<0.001), increasing age (0.32% longer LOS; 95% CI 0.08%−0.57%, p=0.044 per one-year increase in age), and higher comorbidity (9.71% longer LOS; 95% CI 8.54%−10.89%, p<0.001 per one-point increase in CCI) (Table 2). Older adults with epilepsy also had greater cumulative probability of having a LOS longer than older adults without epilepsy (p<0.0001) (Figure 2). For example, older adults with epilepsy were more likely (56.34% vs. 35.72%) to have a LOS longer than 14 days compared to older adults without epilepsy.

Discussion:

Overall, our results confirm that older adults are more at risk for severe disease and complications when infected with SARS-CoV-2. We also found that older hospitalized adults with COVID-19 and epilepsy had higher mortality, longer LOS, and poorer discharge disposition than older adults without epilepsy. Hospitalized older adults with COVID-19 and epilepsy were also more likely than older adults without epilepsy to be mechanically ventilated, less likely to be discharged home, and more frequently had inflammatory and coagulation markers previously reported to be associated with increased COVID-19 disease severity.^{14, 15}

Our findings align with previous research that older adults with pre-existing neurological conditions have a higher rate of mortality with COVID-19.¹⁷ While there are limited studies on the pathogenesis behind increased disease severity in epilepsy and COVID-19 and even fewer on older adults with epilepsy, there have been several studies investigating other neurological conditions.¹⁸ Hypotheses on why neurological conditions lead to increased COVID-19 disease severity appear to be multifactorial. For example, those with dementia may be at higher risk of more rapid cognitive impairment with infection. Similarly, those with Parkinson‟s disease can suffer from dysphagia leading to aspiration pneumonia. Many patients with pre-existing neurological conditions also have comorbid cerebrovascular diagnoses, with epilepsy in older adults often being secondary to stroke.¹⁹ Studies investigating pre-existing cerebrovascular disease in older adults hypothesize that the combination of an older immune system and previous organ damage from comorbid disorders allow the viral-induced cytokine storm caused by COVID-19 to be associated with even more severe multisystemic effects and increased mortality.^17,20

Our study also found that older adults with COVID-19 and epilepsy had significantly longer LOS than those without epilepsy. There are a variety of factors that contribute to a patient requiring prolonged care or delayed discharge.²¹ In the setting of COVID-19, longer LOS have previously been associated with increased disease severity. Need for ventilation is associated with severe disease; in our patient population, older adults with epilepsy were ventilated significantly more than older adults without epilepsy.^3,21 Moreover, disease severity may also correlate with the level of various inflammatory or coagulation biomarkers, such as D-dimer, during acute infection. These are thought to reflect the ongoing inflammatory damage caused by SARS-CoV-2. Again, the group with epilepsy had higher levels of inflammatory markers that have been previously associated with COVID-19 disease and its severity.^{15, 22} Higher need for mechanical ventilation and higher inflammatory markers in the group with epilepsy support the idea that those with epilepsy had more severe illness requiring significantly longer LOS.

Lastly, our work showed that older adults with COVID-19 and epilepsy were more often discharged to a location other than their home. As with LOS and mortality, we suspect this finding is likely multifactorial in etiology. It most likely reflects a combination of increased severity of COVID-19 infection, requiring additional rehabilitation after hospitalization as well as possibly worse functional status at baseline in those with epilepsy. This is because people with epilepsy are more likely to have pre-existing neurologic disease that impairs physical functioning and cognition (e.g. traumatic brain injury, stroke), which could affect their ability to care for themselves.

This study has several strengths. It is a large epidemiological study specifically looking at epilepsy and COVID-19 hospital outcomes specifically in older adults with epilepsy for which the data are scarce. This cohort also allowed us to compare outcomes between people of different races or ethnicity which were previously reported to be associated with different outcomes in the setting of COVID-19.^{23, 24} Finally, our EMR data allowed us to confirm COVID-19 diagnoses via PCR tests.

There are, however, several limitations to this study. Firstly, our analysis includes a large sample and attempts to adjust for multiple known risk factors for poorer COVID-19 outcomes including age and comorbidity (and those associated with COVID-19 such as cardiovascular disease), however it should be acknowledged that residual confounding variables may contribute to bias in the results. Additionally, epilepsy diagnosis was ascertained using ICD-CM codes and ASMs, which could lead to misclassification if the coding was incorrect. However, we used an ICD-CM and ASM-based case definition that our group has validated in our health system.^{12, 13} Similarly, other comorbidities included in this study were identified using ICD-CM based definitions validated in different US samples. Second, the study took place in a large, multi-hospital system geographically located in New York City and our results may therefore not be generalizable to other populations. However, we believe these data are likely broadly applicable as it uses data from a number of different hospitals within a large system (not only a tertiary/quaternary referral center) and it draws on an international, racially, ethnically and linguistically diverse patient population. Third, our study is restricted to hospitalized patients. As such, it is unknown whether the factors found to be associated with our outcomes of interest including mortality would be different in a community-based cohort or with longer follow-up times. An additional limitation is that our dataset does not include full information about epilepsy type (e.g. convulsive seizures) or epilepsy severity. Moreover, we do not have information regarding epilepsy etiology or outpatient anti-seizure medication type or dose. Lastly, we have information on discharge disposition but not on where patients were admitted from (e.g. some were likely admitted from nursing homes) and this may have influenced eventual discharge location although we believe this would likely be a non-differential mis-classification.

Conclusion:

We found that among older adults with COVID-19, epilepsy was associated with in-hospital mortality, LOS and worse discharge disposition. Our work confirmed that older age and comorbid disease are associated with worse outcomes in those with COVID-19. These findings may inform future studies of COVID-19 outcomes with long-term ambulatory data, to help design care pathways for multimorbid older populations. Moreover, these findings suggest that clinicians should consider older adults with epilepsy as being higher risk for poor outcomes which is important for counselling about the risks and benefits of vaccination and other preventative measures. Finally, this work suggests potential health and mortality benefits of earlier detection and more aggressive treatment of COVID-19 in older adults with epilepsy.

Highlights.

• Older adults with epilepsy had significantly higher in-hospital mortality with COVID-19 than those without epilepsy even after adjusting for multiple confounders.

• Older adults with epilepsy and COVID-19 also have longer length of stay and poorer discharge disposition than those without epilepsy.

• Epilepsy is a risk factor for worse hospitalization outcomes in older adults admitted with COVID-19.