Impact of the COVID-19 Pandemic on Inpatient Utilization for Acute Neurologic Disease

Abstract

Background and Objective: The initial months of the Corona Virus 2019 (COVID-19) pandemic resulted in decreased hospitalizations. We aimed to describe differences in hospitalizations and related procedures across neurologic disease. Methods: In our retrospective observational study using the California State Inpatient Database and state-wide population-level estimates, we calculated neurologic hospitalization rates for a control period from January 2019 to February 2020 and a COVID-19 pandemic period from March to December 2020. We calculated incident rate ratios (IRR) for neurologic hospitalizations using negative binomial regression and compared relevant procedure rates over time. Results: Population-based neurologic hospitalization rates were 29.1 per 100,000 (95% CI 26.9–31.3) in April 2020 compared to 43.6 per 100,000 (95% CI 40.4–46.7) in January 2020. Overall, the pandemic period had 13% lower incidence of neurologic hospitalizations per month (IRR 0.87, 95% CI 0.86-0.89). The smallest decreases were in neurotrauma (IRR 0.92, 95% CI 0.89–0.95) and neuro-oncologic cases (IRR 0.93, 95% CI 0.87–0.99). Headache admissions experienced the greatest decline (IRR 0.62, 95% CI 0.58–0.66). For ischemic stroke, greater rates of endovascular thrombectomy (5.6% vs 5.0%; P < .001) were observed in the pandemic. Among all neurologic disease, greater rates of gastrostomy (4.0% vs 3.5%; P < .001), intubation/mechanical ventilation (14.3% vs 12.9%, P < .001), and tracheostomy (1.4 vs 1.2%; P < .001) were observed during the pandemic. Conclusions: During the first months of the COVID-19 pandemic there were fewer hospitalizations to varying degrees for all neurologic diagnoses. Rates of procedures indicating severe disease increased. Further study is needed to determine the impact on triage, patient outcomes, and cost consequences.

Introduction

The Coronavirus Disease 2019 (COVID-19) pandemic resulted in substantial changes to the utilization and allocation of health care resources globally. The US Centers for Disease Control and Prevention initially recommended curtailment of non-emergent in-person ambulatory visits and inpatient admissions, as well the delay of some surgeries. ¹ As a result, the overall number of hospitalizations fell substantially across all services during the first 6 months of the pandemic. ² Reports from individual US centers suggested an initial decline of acute hospitalizations for myocardial infarctions^3,4 and other conditions. ⁵ Trends observed among these individual facilities have been corroborated by analyses of larger health systems, indicating that some hospitals experienced a decline in non-COVID-19 hospitalizations as large as 40% during April 2020. ⁶

During the COVID-19 pandemic, while attention and resources focused on managing the increasing number of cases, ⁷ there was insufficient preparation for patient displacement caused by the overwhelming surge. This discrepancy in care can be attributed, in part, to the uncertainty surrounding the impact of COVID on inpatient utilization for non-COVID cases. Understanding the consequences of such catastrophic events like COVID-19 is crucial, as they provide valuable insights for future disruptions in health care capacities. While it may be unlikely to experience an event of similar magnitude to the COVID-19 pandemic in the near future, historical events such as the 2009 H1N1 influenza pandemic, ⁸ HIV epidemic,^9,10 September 11th terrorist attacks, ¹¹ Hurricane Katrina, ¹² and Sandy^13,14 demonstrate that events which may substantially strain health care services are not uncommon. These examples underscore the importance of anticipating and planning for unforeseen surges in patient numbers or disruptions to health care infrastructure. By studying and learning from these events, we can enhance preparedness and develop effective strategies to not only manage the crisis at hand but also establish safety nets to support otherwise displaced patients.

We therefore examined acute care hospitalizations for neurologic illness before and during the first 10 months of the COVID-19 pandemic in California to describe differences in inpatient admissions and related procedures across multiple categories of neurologic disease.

Methods

Setting and Data Source

We used the California State Inpatient Database (SID) from the Health care Cost and Utilization Project (HCUP) to perform a retrospective analysis of the COVID-19 pandemic’s impact on neurologic hospitalizations and related procedures. The California SID includes a complete enumeration of all-payor claims data on hospital discharges from all non-federal acute care hospitals within California. The SID is available for purchase from HCUP (HCUP-RequestData@ahrq.gov). The University of Rochester Research Subjects Review Board approved the study.

Identification of Neurologic Diagnoses and Procedures

Neurologic hospitalizations were identified using principal International Classification of Disease 10th revision (ICD-10) diagnosis codes which matched our categorization for neurologic disease (eTable 1). Procedures were identified using up to 25 ICD-10 procedure codes for observations within the database which matched our categorization for procedures (eTable 2). Greater detail regarding the methods used to select these codes can be found in the online Supplement (eMethods).

Patient and Hospital Level Characteristics

Patient-level data contained in the SID includes sociodemographic information and up to 36 secondary ICD-10 diagnoses which allowed for calculation of the Elixhauser comorbidity index. ¹⁵ Hospital characteristics were obtained from linkage to the American Hospital Association database and publicly available data on Comprehensive Stroke Center status.^16,17 Further detail regarding the California SID and variables used in the study can be found in the online Supplement (eMethods).

Timeframe

Admission and discharge month and year were known for all admissions, but specific dates of admission and discharge for each patient were not available. Our dataset included all hospitalizations for which the patient was discharged between January 2019 to December 2020. Hospitalizations with admission day during the pandemic period but discharge day after December 2020 were not included, resulting in incomplete ascertainment of admissions that occurred during the final month of the analysis. Therefore, to account for missing data, we performed imputation to estimate December 2020 admission counts (detailed below).

We divided the dataset into a control and pandemic periods based on admission month. The beginning of the pandemic period was set to March 2020 due to admission month being the smallest unit of time in the dataset; roughly in accordance with the World Health Organization’s declaration of the pandemic’s start on March 11, 2020, and extended to December 2020. ¹⁸ The control period was defined as January 2019 to February 2020. We identified troughs and peaks in the graphical trend for neurologic admissions and defined these as nadir and rebound months, respectively. We performed a post-hoc examination of these specific months by patient and hospital characteristics compared to January 2020, similar to prior studies. ⁶

Inclusion and Exclusion Criteria

Hospitalizations for neurologic diagnoses among patients 18 to 100 years of age were included in the study. Data were cleaned for missingness across key variables, restricted to acute inpatient admissions within the pre-defined time period, and limited to hospitals with available data in both time periods (eFigure 1).

Outcome Measures

The primary outcome measures for this study were counts of acute care hospitalizations for neurologic disease and population-based hospitalization rates per 100,000 by sex, age, race and ethnicity, and homelessness based on US Census Bureau and US Interagency Council on Homelessness data for California. ¹⁹

Data Imputation

Due to the missingness of December 2020 hospitalizations that were discharged in the year 2021, we imputed December 2020 hospitalization counts using existing data and a multiplier based on the proportion of December 2019 hospitalizations that were discharged in the year 2020. Multipliers were derived by dividing the number of December 2019 hospitalizations that were discharged in 2020 by the number of December 2019 hospitalizations that were discharged in 2019 for each subgroup.

Statistical Analysis

We descriptively assessed trends in neurologic hospitalizations and population-based hospitalization rates per 100,000 population by month. We report incident rate ratios (IRR) for monthly neurologic hospitalizations comparing the pandemic to control period for each category of neurologic disease. In sensitivity analysis, we calculated IRRs excluding December 2020 data. We calculated percent of hospitalizations receiving procedures for all neurologic disease or the corresponding inpatient diagnoses of interest (ie, percent of ischemic stroke hospitalizations receiving tPA and EVT).

We used Chi-squared and Wilcoxon Rank Sum to determine differences in categorical and continuous variables, respectively. Standard errors for population-based hospitalization rates were calculated using variance-covariance estimators accounting for the hospital clustered sampling design of the SID. To calculate IRRs, we used negative binomial mixed-effects modeling with hospital-level random effects. A priori, we set P < .05. Statistical analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, NC) and Stata v17.0 (StataCorp, College Station, TX).

Results

Across 320 California hospitals, there were 177,191 neurologic hospitalizations during the 14-month control period and an estimated 109,695 neurologic hospitalizations during the 10-month pandemic period. The largest categories of neurologic disease were neurovascular (n = 143,481, 50% of total), neurotrauma (n = 45,991, 16% of total), and epilepsy (n = 28,859, 10% of total).

Changes in Neurologic Hospitalizations

For all months during the pandemic period, monthly neurologic hospitalizations were lower than January 2020, with the lowest incidence in April 2020 (Nadir #1) and highest incidence in October 2020 (Rebound). Following the rebound in hospitalizations, a second trough was observed in December 2020 (Nadir #2) (Figure 1).

Figure 1.

Trends in Neurologic Hospitalizations by Disease Category, January 2019 to December 2020. Trends in neurologic hospitalizations for each subcategory of neurologic disease are represented in terms of change compared to January for the respective year, 2019 and 2020. Total count for December 2020 was imputed based on December 2019 values for all admissions.

Population-based neurologic hospitalization rates were 29.1 per 100,000 (95% CI 26.9-31.3) in April 2020 compared to 39.6 per 100,000 (95% CI 36.8-42.9) in April 2019 and 43.6 per 100,000 (95% CI 40.4-46.7) in January 2020 (Figure 2). Each subcategory of neurologic disease experienced a decline in population-based hospitalization rates in April 2020 compared to April 2019 and January 2020 except for neurooncologic hospitalizations.

Figure 2.

Trends in Neurologic Hospitalization Rates per 100,000 Population by Disease Category, January 2019 to December 2020. Hospitalization rates were calculated using SID hospitalization counts by disease subcategory compared to the California population in the respective year (2019 or 2020) of the US Census Bureau. 95% confidence intervals were calculated using variance-covariance estimators accounting for the hospital clustered sampling design of the State Inpatient Dataset. The vertical line marks the beginning of the pandemic period.

The pandemic period had 13% lower incidence of neurologic hospitalizations per month compared to the control period (IRR 0.87, 95% CI 0.86 to 0.89) (Figure 3). Subcategories of disease with the greatest decline in incidence of neurologic hospitalizations during the pandemic period compared to the control period included headache (IRR 0.62, 95% CI 0.58 to 0.66), neuromuscular (IRR 0.74, 95% CI 0.70 to 0.78) and neuroinfectious/neuroinflammatory disorders (IRR 0.79, 95% CI 0.74 to 0.83). Subcategories with the smallest change in hospitalizations were neurotrauma (0.92 95% CI 0.89 to 0.95) and neurooncologic (0.93, 95% CI 0.87 to 0.99).

Figure 3.

Incident Rate Ratios for Subcategories of Neurologic Hospitalizations in Control vs Pandemic Period. Forest plot conveying incident rate ratio (IRR) estimates and 95% confidence intervals for subcategories of neurologic disease, where a ratio <1 indicates a decrease in neurologic hospitalizations in the control period (January 2019-February 2020) compared to the pandemic (March 2020-December 2020). Estimates were calculated using negative binomial regression modeling with repeated measures by hospital. Total count for December 2020 was imputed based on December 2019 values for all hospitalizations.

In sensitivity analyses excluding December 2020, there were no substantiative changes to the magnitude or significance of IRRs comparing the control and pandemic periods (eTable 3).

Patient Characteristics

Compared with the control period, there were a smaller proportion of neurologic admissions identified as women (52.2% vs 46.6%, P < .001) in the pandemic period. The mean Elixhauser index value within the control period was 10.8 (Standard Deviation [SD] = 16.2) vs 11.5 (SD = 16.6) in the study period (P < .001). The remainder of patient characteristics were similar between the two groups though the difference was statistically significant (P < .001 for all) (Table 1). Throughout the pandemic period, there was an increase in the proportion of principal neurologic hospitalizations with secondary diagnosis of COVID-19 from 0.62% in April 2020 to 6.5% in December 2020 (P < .001) (eFigure 2).

Table 1.

Patient Characteristics of Hospitalizations for Neurologic Disease before and During COVID-19 Pandemic.

Patient characteristic	Control Period, a n (%) (n = 177,191)	Pandemic Period,a,b n (%) (n = 109,695) a	P value
Age category
18-54 years	44,543 (25.1)	27,322 (24.9)	<.001
55-64 years	32,686 (18.4)	20,889 (19.0)
65-74 years	36,914 (20.8)	23,160 (21.1)
75-84 years	35,794 (20.2)	22,203 (20.2)
≥85 years	27,254 (15.4)	16,122 (14.7)
Female	92,449 (52.2)	51,089 (46.6)	<.001
Race and ethnicity c
Asian or Pacific Islander	18,881 (10.7)	11,355 (10.4)	<.001
Black	17,987 (10.2)	10,970 (10.0)
Hispanic	47,153 (26.6)	29,458 (26.9)
Native American or Alaskan native	663 (.4)	292 (.3)
White	83,454 (47.1)	51,123 (46.6)
Other
Homeless	4,658 (2.6)	3,328 (3.0)	<.001
Insurance Payor
Medicare	98,460 (55.6)	60,601 (55.3)	<.001
Private insurance	30,506 (17.2)	18,823 (17.2)
Medicaid	39,703 (22.4)	25,217 (23.0)
Self-pay, No-charge or other	8,486 (4.8)	5,027 (4.6)
Elixhauser comorbidity Index d
First quartile	61,325 (34.6)	36,057 (32.9)	<.001
Second quartile	30,949 (17.5)	18,416 (16.8)
Third quartile	47,404 (26.8)	29,989 (27.3)
Fourth quartile	37,513 (21.2)	25,324 (23.1)

^aControl period was defined as January 2019 to February 2020. Pandemic period was defined as March 2020 to December 2020.

^bTotal count for December 2020 was imputed based on December 2019 values for all hospitalizations.

^cRace and ethnicity as reported by HCUP. Hispanic ethnicity is assigned to any patient identified as Hispanic, including those identified with additional race categories. Other may include multiple race or uncategorized.

^dElixhauser Comorbidity Index is calculated based on the presence of 38 comorbidity measures present in up to 36 secondary diagnosis codes. Higher scores indicate more severe or more numerous comorbidity. ¹⁵ Quartiles cutoffs: first quartile (≤0), second quartile (1-6), third quartile (7-23), fourth quartile (>23).

The largest decrease in neurologic hospitalizations in the nadir months compared to January 2020 occurred among those age ≥85 years, women, and those with low comorbidity while the smallest decrease in neurologic hospitalizations in the nadir months compared to January 2020 occurred among homeless patients (Table 2). Race and ethnicity subgroups experienced similar changes in hospitalizations during nadir and rebound periods.

Table 2.

Difference in Neurologic Hospitalizations During Pandemic Nadir and Rebound Periods Compared to January 2020 Baseline by Patient Characteristic.

Characteristic	January, n (n = 13,325)	April, Nadir #1, % Change a (n = 8,906)	P value	October, rebound, % Change a (n = 12,001)	P value	December, Nadir #2, % Change a (n = 10,585) b	P value
Age category
18-54 years	3,220	−33.2	.03	−8.1	.02	−25.6	.01
55-64 years	2,429	−29.6		−9.0		−18.6
65-74 years	2,769	−31.5		−6.3		−19.9
75-84 years	2,742	−33.1		−10.2		−15.0
≥85 years	2,165	−39.3		−18.1		−23.1
Sex
Male	6,968	−30.9	.01	−9.3	.56	−18.6	.05
Female	6,357	−35.6		−10.6		−22.7
Race and ethnicity c
Asian or Pacific Islander	1,484	−40.0	.052	−15.6	.13	−19.2	.051
Black	1,309	−29.9		−9.0		−21.9
Hispanic	3,520	−33.7		−6.9		−21.3
Native American or Alaskan native	38	−5.3		15.8		−55.3
White	6,225	−31.9		−11.4		−21.3
Other	749	−34.6		−3.9		−9.3
Homelessness d
Housed	12,999	−33.8	<.001	−10.3	.16	−21.0	.06
Homeless	311	−9.0		0.6		−7.6
Insurance Payor d
Medicare	7,519	−33.7	.04	−10.7	.02	−20.2	.87
Private	2,233	−35.2		−8.8		−21.1
Medicaid	2,913	−28.8		−5.8		−20.2
Self-pay, No-charge or other	657	−39.6		−22.5		−24.2
Elixhauser comorbidity Index e
First quartile	4,352	−37.7	<.001	−8.4	.59	−24.1	<.001
Second quartile	2,241	−35.2		−9.2		−17.0
Third quartile	3,706	−32.9		−12.3		−24.2
Fourth quartile	3,026	−25.4		−9.7		−12.4

^aPercent change from January baseline.

^bTotal count for December 2020 was imputed based on December 2019 values for all hospitalizations.

^cRace and ethnicity as reported by HCUP. Hispanic ethnicity is assigned to any patient identified as Hispanic, including those identified with additional race categories. Other may include multiple race or uncategorized.

^dMissingness: Homelessness 0.2%, Insurance Payor 0.02%.

^eElixhauser Comorbidity Index is calculated based on the presence of 38 comorbidity measures present in up to 36 secondary diagnosis codes. Higher scores indicate more severe or more numerous comorbidity. ¹⁵

Quartiles cutoffs: first quartile (≤0), second quartile (1-6), third quartile (7-23), fourth quartile (>23).

Examination of sex, age, race and homeless-specific population-based hospitalization rates over the time period demonstrated similar trends (eFigure 3).

Hospital Characteristics

There were differences for most hospital characteristics across the control and pandemic periods (P < .001 for all except hospital ownership), though the magnitude of these differences were small (eTable 4). During the second nadir of December 2020, large decreases in neurologic hospitalizations (28% decline from January 2020 for the highest quintile) were seen among hospitals with high COVID-19 admission rates compared to hospitals with low COVID-19 admission rates (P < .001) (eTable 5).

Changes in Neurologic Procedures

Within the control period, the proportion of hospitalizations receiving procedures were stable, apart from EVT which was increasing as a proportion of ischemic stroke cases at the end of 2019 (eTable 6, eFigure 4). Compared to the control period, during the pandemic period tPA administration among ischemic stroke (n = 55,398; 15.1% vs 14.2%, P < .001) and LP among all neurologic admissions decreased (4.8% vs 4.6%, P = .03). EVT among ischemic stroke increased (n = 55,398; 5.0% control vs 5.6% pandemic, P < .001), and among all neurologic admissions, intubation/mechanical ventilation (12.9% control vs 14.3% pandemic, P < .001), gastrostomy (3.5% control vs 4.0% pandemic; P < .001), and tracheostomy (1.2% control vs 1.4% pandemic; P < .001) increased. EEG among all neurologic hospitalizations did not change (3.6% control vs 3.5% pandemic, P = .45).

When compared to a January 2020 baseline, in April 2020 (Nadir #1) there were lower proportion of tPA administrations among ischemic stroke patients (15.4% vs 13.7%; P = .04), but no differences in EVT (eTable 6). Among all neurologic hospitalizations, there were higher proportions of intubation/mechanical ventilation (13.5% Jan 2020 vs 15.6% Apr 2020; P < .001), but no change in the proportion receiving EEG, LP, or gastrostomy. In December 2020 (Nadir #2) compared to January 2020, EVT (5.4% vs 6.5%; P = .05) increased as a proportion of ischemic stroke hospitalizations. For all neurologic disease in December 2020 (Nadir #2) compared to January 2020, gastrostomy (3.9% vs 4.7%; P = .001), intubation/mechanical ventilation (13.5% vs 16.3%; P < .001), and tracheostomy (1.2% vs 1.9%, P < .001) increased. Among those receiving intubation/mechanical ventilation, 1.3% had comorbid diagnosis of COVID-19 in April 2020 compared to 12.0% in December 2020 (P < .001).

Discussion

In this study, we found the COVID-19 pandemic was associated with substantial decreases in acute care hospitalizations for neurologic disease in California. The largest change occurred during April and December 2020 when neurologic hospitalizations decreased 33% and 21%, respectively. The magnitude of reduction in hospitalizations was not uniform across disease category. Neurooncologic conditions and neurotrauma experienced small decreases whereas headache and neuromuscular cases fell precipitously.

Our findings are consistent with prior reports that demonstrated decreases in admissions in April 2020 and subsequent months.^6,20,21 A large study of all inpatient admissions of 201 hospitals across 36 states found a 42.8% decrease in non-COVID-19 admissions during the April nadir. ⁶ A decrease in stroke-related admissions was observed in a large global study, ²² which evaluated trends in 457 stroke centers across 70 countries and found a 19% drop in North American monthly hospitalizations during the first 4 months of the pandemic. We generally saw a greater magnitude of decline averaged over a four-month study period (range: −18 to −31%), but this may reflect changes specific to California and our inclusion of non-stroke center hospitals. We add to current knowledge of the pandemic’s impact on neurovascular admissions by demonstrating that they failed to return to normal volumes throughout the entirety of 2020 and began to decline again in December 2020.

The literature regarding admission for other neurologic conditions is nascent; studies have largely been conducted in small US samples,^23,24 international cohorts,^25-29 or focused on survey-based methods.^30,31 Similar to our findings, these studies found decrements in admissions for neurotrauma,^23,24 neuro-oncologic surgical cases, ³¹ epilepsy, ²⁵ headache, ²⁶ neurodegenerative, ²⁸ and neuro-inflammatory disorders ²⁹ during the initial phase of the pandemic. Among the various categories of neurologic disease, we found headache hospitalizations to be most impacted, falling 66.7% in April and still being down by 20.5% by the October rebound period. This reflects a similar trend seen in Italy, where emergency department visits for headache was reduced by 49% during the months of March to May 2020, compared with January to February of the same year. ²⁶ These trends may reflect an ameliorative effect of the work-from-home accommodations imposed by the pandemic on primary headache disorders. ³² However, there may have also been greater reluctance to present for acute treatment for headache, a condition which patients may perceive as a lower priority compared to other health problems. ³³ In contrast to what was seen with headache, we found neurotrauma hospitalizations to be least affected with an incident rate ratio of 0.93. While there was a substantial drop during the first nadir (25.3%) compared to January 2020, neurotrauma hospitalizations surpassed January baseline by 3.8% during the October 2020 rebound. This “resistance” to decline is somewhat unexpected, as reports have suggested a decrease in US motor vehicle related injuries during the pandemic. ³⁴ However, other studies indicated increase rates of gun-related injuries ³⁵ and domestic violence, ³⁶ which may partially account for the relatively minor changes in trauma-related neurologic hospitalizations. While the variability in pandemic-related declines is likely due to multiple factors, there may be a gap in public awareness about the seriousness of certain neurologic signs/symptoms and manner of presentation that warrant further study.

We also identified changes in procedure rates, with tPA administration and LP rates decreasing during the pandemic, though only modestly. Notably, rates of thrombectomy (among ischemic stroke hospitalizations), gastrostomy, intubation/mechanical ventilation, and tracheostomy increased during this time. There was no difference in the proportion of hospitalizations receiving EEG during the pandemic. Consistent with prior reports, ²² we observed a decline in the rate of tPA administration for acute ischemic stroke, though contrary to prior reports ³⁷ rates of EVT during the pandemic period were higher in California and pronounced during the two nadir periods. More severe presentations of disease comprising a greater proportion of total ischemic stroke hospitalizations may be the driving force behind this trend.

Concordant with this observation regarding increased procedure rates indicating severe disease, patients with greater comorbidity (ie, higher Elixhauser comorbidity index scores) made up a higher proportion of neurologic hospitalizations during the pandemic compared with the pre-pandemic, control period of our study. Fear of contagion may have increased the threshold in which patients sought care,^38,39 resulting in fewer “mild” and proportionally greater “severe” presentations. In alignment with this possibility, we found hospitals with high COVID-19 admission rates were associated with greater declines in neurologic hospitalizations in December 2020.

Further study is needed to describe how neurologic outcomes changed during the pandemic. Identification of the most impacted patient populations may help health systems better prepare for future waves of COVID-19 or other pandemics. This should include approaches to coordinate inpatient, outpatient, and emergency teams to best respond to shifts in sites and modes of care delivery that occur during a pandemic. Alternatively, disruption in normal patterns of health care utilization by the pandemic may have forced systems to address previously tolerated inefficiencies, over-triage, and costly care.^40,41 Further studies may be helpful in identifying changes in care during the pandemic which resulted in unintended improvements in efficiencies and cost-savings which did not sacrifice quality of care.

Limitations

There were several limitations to this study. First, our data were restricted to admissions within California. Therefore, the generalizability of the results may be limited as case rates for COVID-19 varied by state. ⁴² The prevalence of some neurologic conditions, such as ischemic stroke, have been found to vary geographically which could also influence inpatient resource utilization. ⁴³

Second, we relied on a combination of CCSR and ICD-10 codes to capture a wide array of neurologic diagnoses and procedures; however, these codes may not have the same level of specificity as ICD-10 codes previously validated for various categories of neurologic disease.^44,45

Third, we lacked hospitalization data for patients admitted in December 2020 but discharged in the subsequent year. Our imputation of admission counts based on 2019 data could represent an under or overestimation of neurologic hospitalizations and procedures which may influence some findings with borderline significance (eg, decline in monthly admissions for neurotrauma or neuro-oncologic cases). However, our analysis of the study period compared with control period was largely unchanged by the omission of December data (eTable 3) and any underestimation would be unlikely to account for a 21% relative difference in admissions compared to January of the same year.

Fourth, our study is limited to a cross-section of individual hospitalizations. We are unable to understand occurrences in health or instances of care prior to and following hospitalizations, thus, limiting us in understanding the effects of delays in care from the COVID-19 pandemic.

Our study had several strengths. Our dataset had a large number of observations in a diverse patient population that included patients with a broad mix of insurance payors, who sought care from a wide spectrum of hospital settings (eg, large academic centers, urban community, or rural community hospitals). This allowed us to assess for potential differences in admissions for various patient and hospital characteristics across the control and pandemic periods. Second, we were able to examine longer study periods compared to earlier studies, allowing us to ascertain overall differences between much of the pandemic and the preceding year.

In conclusion, we found acute neurologic hospitalizations declined during the pandemic with considerable variability. While rates of tPA administration and LP fell, rates of thrombectomy, gastrostomy, intubation/mechanical ventilation, and tracheostomy rates increased, suggesting a shift toward a greater proportion of hospitalizations with more severe disease. Further studies are needed to determine how these decreases in neurologic hospitalizations affected outcomes, and use this opportunity to investigate changes in care during the pandemic which may have resulted in improved efficiency, triage, and cost-savings.

ORCID iD

Benjamin P. George https://orcid.org/0000-0001-9966-5632