COVID-19 and Acute Neurologic Complications in Children

We conducted a cross-sectional, multicenter study of children hospitalized at academic centers that contribute data to the Pediatric Health Information System (PHIS) database. The PHIS administrative database includes billing claims and utilization data from 49 tertiary care children’s hospitals located in the United States that are affiliated with the Children’s Hospital Association (Lenexa, KS).^15,16  Multiple process measures are performed by the Children’s Hospital Association and participating hospitals to ensure data quality and reliability. Discharge data, including demographics, diagnoses, pharmacy information, and procedures using the International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) are included in the database.

We included children from 2 months up to 18 years of age hospitalized at a PHIS hospital with either (1) a primary ICD-10-CM discharge diagnosis of COVID-19 or (2) a secondary discharge diagnosis of COVID-19 with a primary diagnosis of a known COVID-19 complication or associated condition (Supplemental Table 4).^3,17  An ICD-10-CM discharge diagnosis of COVID-19 has a >95% positive predictive compared with reverse-transcription polymerase chain reaction results.¹⁸  To best identify children hospitalized for COVID-19, children cared for in the NICU and those with a surgical primary diagnosis were excluded from the study. Children <2 months of age were excluded because there are special considerations for hospitalization in this age group, similar to previous studies.⁴ 

We identified a priori risk factors for COVID-19 complications on the basis of previous studies in children.^4,19  These included: age, race/ethnicity (as a social construct and included non-Hispanic white, non-Hispanic Black, Hispanic, Asian, other, defined as Pacific Islander, American Indian, multiracial, and other unspecified race/ethnicity), chronic neurologic conditions (neurologic conditions defined using the pediatric complex chronic conditions [CCC] classification system version 2),²⁰  nonneurologic CCCs, predominant SARS-CoV-2 strain at time of hospitalization, and COVID-19 pharmacologic treatment (remdesivir or dexamethasone). Predominant SARS-CoV-2 strain or variant time periods were defined as hospitalization occurring during a week in which >50% of US SARS-CoV-2 infections were attributed to a single circulating strain or variant according to the Centers for Disease Control and Prevention (CDC) COVID-19 tracker and the Oxford Our World in Data COVID-19 surveillance data (Supplemental Table 5).^21,22  We used 2 sources for our data because neither source had data for the entire study period. For periods in which both data sources overlapped in variant data, there was identical concordance in the proportion of variants in the United States. Washout periods were defined as periods in which no SARS-CoV-2 strain or variant accounted for >50% of infections. Race and ethnicity were examined as associated factors because of previous studies in SARS-CoV-2 and other respiratory viruses in children that have found neurologic complications to be independently associated with race or ethnicity.^4,11,23 

The primary outcome was a neurologic complication, defined as the presence of an ICD-10-CM diagnostic code during the hospitalization with COVID-19. Neurologic complications were categorized on the basis of previous studies and included: encephalopathy, encephalitis, aseptic meningitis, febrile seizure, nonfebrile seizure, brain abscess and bacterial meningitis, Reye’s syndrome, or cerebral infarction (Supplemental Table 6).^4,19  We did not include more subjective neurologic complications such as headache, dizziness, or fatigue in our outcome definition. Our secondary outcomes focused on in-hospital outcomes and included length of stay (LOS), admission to the ICU, ICU LOS, 30 day readmissions (defined as any hospitalization for any indication within 30 days of index discharge date), in-hospital death, and costs associated with hospitalizations in which a COVID-19-associated neurologic complication occurred compared with those without neurologic complications.^3,4  We estimated costs using hospital-specific cost-to-charge ratios and costs were adjusted for hospital location by using the Centers for Medicare and Medicaid price wage index.²⁴ 

Demographic characteristics were summarized by using frequencies and percentages for categorical variables and medians with standard deviations for continuous variables. We used χ² tests for bivariate comparisons between children with and without neurologic complications. The proportion of children with neurologic complications was calculated by dividing the number of COVID-19-associated hospitalizations with neurologic complications by the total number of COVID-19-associated hospitalizations. Bivariate comparisons using χ² tests for hospital outcomes were made between children without and without neurologic complications. Factors associated with neurologic complications were evaluated by using multivariable logistic regression incorporating the following covariates: age, sex, race/ethnicity, payer, number of nonneurologic CCC, neurologic CCC, remdesivir or dexamethasone treatment, and predominant circulating SARS-CoV-2 strain, similar to previous studies.⁴  We used generalized linear mixed effects models with a binomial distribution and random intercepts for each hospital to generate adjusted odds ratios (aOR) for each risk factor as performed previously.^4,15,19  Results were stratified by age (2 months to 4 years, 5–11 years, 12–17 years) similar to the CDC age stratification in children with COVID-19.²⁵  All analyses were performed by using SAS version 9.4 (SAS Institute, Inc, Cary, NC). We used a cut-off of P <.05 to determine statistical significance.

A total of 15 137 COVID-19-associated hospitalizations were included in the study. Of these, 12 427 (82.1%) had a primary diagnosis of COVID-19, and 17.9% had a secondary diagnosis of COVID-19 with a concurrent primary diagnosis of a known COVID-19 complication. Those with COVID-19-associated hospitalizations were more likely to be male (7943, 52.5%), older (12 to18 years, 6152, 40.6%), and have government insurance (9868, 65.2%). Most patients received remdesivir (11 473, 75.8%) or dexamethasone (10 974, 72.5%) during their hospitalization. More than one-third (37.1%) of children had a CCC, and 9.8% had at least 1 neurologic CCC. Demographic characteristics are revealed in Table 1.

Of the 15 137 COVID-19-associated hospitalizations, 1060 (7.0%) children had a neurologic complication. The most common neurologic complication was febrile seizures (582, 3.8%) followed by nonfebrile seizures (352, 2.3%) and encephalopathy (332, 2.1%). The least common neurologic complications were brain abscess or bacterial meningitis (19, 0.13%), and cerebral infarction (8, 0.05%); Reye’s syndrome was not reported (Table 2).

When comparing children without neurologic complications, children with neurologic complications had more ICU admissions (29.8 vs 21.8; P <.001), longer ICU LOS (3.2 vs 2.5 days; P <.001), and more in-hospital deaths (1.8% vs 0.6%; P <.001). Children with neurologic complications also had more readmissions and higher hospital costs compared with those without neurologic complications (Table 3).

In multivariable analyses, the presence of a neurologic CCC was strongly associated with increased odds of neurologic complication. Factors associated with a decreased odds of neurologic complications include treatment with remdesivir or dexamethasone, SARS-CoV-2 δ variant, and the presence of 1 nonneurologic CCC (Fig 1).

In stratified analyses by age, the odds of neurologic complications in those with a neurologic CCC were higher as age increased (Fig 2, Supplemental Table 7). The decreased odds of neurologic complications associated with SARS-CoV-2 δ variant predominance and those with remdesivir or dexamethasone treatment were consistent across age strata (Fig 2, Supplemental Table 7).

In this large, multicenter study evaluating neurologic complications among children hospitalized with COVID-19, we determined that neurologic complications are common, occurring in ∼1 out of every 15 COVID-19-associated hospitalizations, and associated with increased risk of ICU admission, readmissions, in-hospital mortality, and increased hospital costs compared with hospitalizations without neurologic complications. Neurologic CCC was associated with greater odds of neurologic complications, whereas increasing age, use of remdesivir or dexamethasone treatment, presence of nonneurologic CCC, and infection during the SARS-CoV-2 delta variant were associated with lower odds of neurologic complications.

Previous reports on the prevalence of neurologic complications in children hospitalized with COVID-19 vary widely, and the estimated 7% prevalence in our study falls on the lower end of published estimates.^11–13  There are a number of factors that may contribute to these differing estimates, including differences in age, population characteristics, study period dates, and neurologic complication outcome definitions. For example, our definition excluded mild complications and those of unclear significance (weakness, dizziness, headache, anosmia, etc.) that were included in outcome definitions in studies with a much higher reported prevalence of neurologic complications.¹¹  Nevertheless, the prevalence of neurologic complications in the current study is similar to that of a recent study in children hospitalized with influenza that used an identical outcome definition.⁴ 

We also found that neurologic complications were associated with more intensive resource use, increased ICU admissions, longer ICU LOS, and higher hospital costs, as compared with those without neurologic complications. Deaths and readmissions were also more common among those with neurologic complications. These findings suggest that children hospitalized with COVID-19 who experience neurologic complications are at risk for worse clinical outcomes.

Previous studies have revealed that neurologic comorbidities are associated with greater COVID-19 disease severity in hospitalized children.^3,26  In this case, it is unclear whether neurologic complications result from more severe COVID-19 disease, or whether undiagnosed or underlying neurologic conditions contribute to the increased complications. It is likely that those with severe COVID-19 disease are at higher risk for neurologic complications. It is also likely that those with underlying neurologic disorders are at higher risk of having neurologic complications with any concurrent illness, including COVID-19. Our finding of a 4-fold increase in the odds of neurologic complications in children with neurologic CCC, and the consistency of this association across age strata, supports the idea that COVID-19 disproportionately affects this vulnerable population. This finding is similar to the well-described relationship between influenza and chronic neurologic conditions.^4,23,27–29  Neurologic complications of viral infection in children have been reported in other viruses as well, such as respiratory syncytial virus, although the importance of underlying neurologic comorbidity is less well described.

Our findings emphasize the strategic importance of immunization and other prevention efforts in children overall, and specifically in children with chronic neurologic conditions. There is strong evidence that immunization prevents severe COVID-19 in children and adolescents and reduces the likelihood of infection.^30,31  Neurologic complications in children may be reduced by immunization through decreased infection and severity of the disease. As seen in the case with influenza, vaccination of those at the highest risk of poor outcomes, such as children with underlying neurologic conditions, is critically important.

A strength of our study was the ability to examine associations between predominant circulating SARS-CoV-2 variants and neurologic complications. We found that hospitalizations occurring during the delta-predominant period were associated with a decreased odds of neurologic complications compared with wildtype SARS-CoV-2. Although we could not confirm that all of the COVID-19-associated hospitalizations during the Δ-predominant period were caused by this variant, we note that delta accounted for >90% of infections reported during the majority of this period. For example, on June 28, 2022, at the start of the delta-predominant period, delta accounted for 51.9% of new infections and by July 12 it accounted for 80.2%. Delta remained responsible for greater than 90% of new infections until December 13, with the delta period ending December 18.^21,22 

Reasons for the association between delta and lower odds of neurologic complications are unclear. We could not account for immunity from previous infections or immunization in our analysis and it is likely that both natural and immunization-related immunities at a community level were greater during the delta period compared with the wildtype period. For example, the Pfizer-BioNTech vaccine received Emergency Use Authorization for those 12 and older on May 10, 2021 and for those >5 years of age on May 17, 2022. It is possible that older children and those at higher risk of neurologic complications were more likely immunized during the delta period, which would suggest a potential protective effect of immunization. According to the CDC, by the end of the delta period, 54.2% of those 12–17 years and 13.8% of those 5–11 years of age had completed the 2 dose COVID-19 immunization series.³²  At the end of the wildtype period, 0.2% of those 12–17 years and 0% of those 5–11 years of age had completed the immunization series.³²  If true, we would expect subsequent variants to also have lower odds of neurologic complications compared with wildtype SARS-CoV-2. We did not find any association between the ο variant and neurologic complications. However, the power to detect differences may have been limited by the relatively short portion of the study period in which omicron was the dominant circulating strain. Additional study is needed to determine the relationship between omicron, subsequent variants, and neurologic complications in children.

Our finding of markedly decreased odds of neurologic complications with COVID-19 pharmacologic treatment (remdesivir and/or dexamethasone) must be carefully considered. This was a cross-sectional study in which the exposure and outcomes were identified during the same period and therefore the temporal association between treatment and neurologic complications could not be determined. For example, it is possible that neurologic complications were present before pharmacologic treatment. In addition, national protocols for pharmacologic treatment of COVID-19 in hospitalized children were mainly driven by respiratory criteria (eg, need for supplemental oxygen).³³  We were not able to examine the temporal sequence of events or management decisions in our study. Nevertheless, our findings of an association between remdesivir or dexamethasone exposure and reduced neurologic complications warrant additional consideration. It is biologically plausible that reduced viral replication and decreased inflammation could reduce the risk of neurologic complications. Rigorous pharmacoepidemiologic investigations are needed to better elucidate whether pharmacologic treatment may prevent neurologic complications.

Limitations of our findings include the retrospective nature of the study and that we only included freestanding children’s hospitals. Therefore, results may not be generalizable to the community hospital setting. Discharge diagnosis codes were used to identify our exposure, and outcomes, which may have led to some degree of nondifferential misclassification. However, the diagnosis of COVID-19 in the hospital setting is highly specific for SARS-CoV-2 infection. Because of limitations of the PHIS database, we were unable to ascertain COVID-19 immunization status, nor were we able to determine the timing between neurologic complication and medication exposure (eg, remdesivir or steroids). Finally, SARS-CoV-2 variant information was not available at the patient level, and it is possible that there was a misclassification of SARS-CoV-2 variant determination. However, the high rates of single predominant strain during the defined variant time periods minimize this concern.

Neurologic complications occurred in 7% of children hospitalized with COVID-19 in our study and were associated with more resource utilization and worse hospital outcomes. There was a strong association between neurologic complications and neurologic comorbidities and younger age. Our findings emphasize the strategic importance of SARS-CoV-2 immunization and continued monitoring of new variants in children to protect this vulnerable population.

aOR

adjusted odds ratio

CCC

complex chronic condition

CDC

Centers for Disease Control and Prevention

CI

confidence interval

COVID-19

coronavirus disease 2019

ICD-10-CM

International Classification of Diseases, Tenth Revision, Clinical Modification

LOS

length of stay

PHIS

Pediatric Health Information System

SARS-CoV-2

severe acute respiratory syndrome coronavirus 2