Results
Our primary study cohort included 161 239 patients (158 267 adults and 2972 children) with acute COVID-19 infection, of which 35 993 (22.3%) patients were admitted to an ICU and 125 246 (77.7%) were hospitalized in non-ICU beds (Figure 1 and Supplementary Figure 1). Among the ICU cohort, 15 961 (44.3%) were admitted to the ICU on the same day as initial COVID-19 hospitalization. Demographic characteristics and comorbidities of the COVID-19 cohort are summarized in Table 1. Overall, 56.7% were male and median age was 69 years (IQR = 54–81). The median time from symptom onset to hospitalization was 5 days (IQR = 1–8). After hospitalization, 65.6% of patients were discharged alive and 24.1% died; the remaining patients were transferred to other facilities for further treatment (7.1%) or had recovered from COVID-19 but remained hospitalized (3.2%). Among the ICU cohort (n = 35 993), more than half of all patients (52.3%) were admitted to the ICU on the first day of admission (Supplementary Table 4). ICU patients were younger than non-ICU patients (61 versus 73 years) and had a higher frequency of obesity (21.4% versus 11.4%; Table 1). Additional characteristics of the ICU cohort, including the use of invasive mechanical ventilation and ECMO, are presented in Supplementary Table 4.
Neurological Manifestation at Presentation
Adults Versus Children. Fatigue was the most commonly reported neurological manifestation of acute COVID-19 at admission (adults: 37.4%; children: 20.4%). All neurological manifestations were more frequent in adults than in children, except for seizures (adults: 1.1%; children: 5.2%). One in 20 children presented with a seizure, a frequency approximately 5 times greater than that in adults (Table 2). Notably, altered consciousness was substantially more common in adults (20.9%) than in children (6.8%), and prevalence increased with age (Table 2 and Supplementary Figure 2).
ICU Versus Non-ICU. Altered consciousness, fatigue and myalgia were more prevalent in children admitted to the ICU than in children admitted to a non-ICU floor (P < 0.001), whereas anosmia, dysgeusia and seizure were similarly present in both cohorts. Surprisingly, adults with COVID-19 infection requiring ICU admission were less likely to present with altered consciousness than those on non-ICU floors (10.8 versus 24.0%; OR = 0.39; 95% CI = 0.37–0.40, P < 0.001) and less likely to have seizure (0.8 versus 1.2%; OR = 0.67; 95% CI = 0.58–0.76, P < 0.001) as their initial neurological presentation (Table 2).
In-hospital Neurological Complications
Adults Versus Children. In-hospital neurological complications (CNS infection, seizure and stroke) were rare in both adults and children. In the overall cohort, 0.22% (95% CI = 0.20–0.24%) had CNS infection, 1.0% (95% CI = 0.98–1.10%) experienced seizures and 1.5% (95% CI = 1.4–1.5%) suffered acute stroke during the index hospitalization with COVID-19. Again, seizure was more frequent in children (3.0%) than in adults (1.0%; Table 2); reported in-hospital seizures decreased with increasing age (Figure 2). The frequency of stroke increased with increasing age. In contrast, CNS infection and seizure proportions steadily decreased with increasing age (Figure 2).
Figure 2.
Results of multivariable analysis of neurological complications. (A) Age trends. (B) Trends over time. (C) Forest plot for remaining fixed effects, including confounders. Raw values (C) are presented in Supplementary Table 5. UK-CCP = United Kingdom Clinical Characterization Protocol; COVID-19 CCC = COVID-19 Critical Care Consortium.
ICU Versus Non-ICU. In children, ICU patients (n = 443) were more likely than non-ICU patients (n = 2529) to have in-hospital neurological complications, whereas the frequency of neurological complications was not as distinct in ICU and non-ICU adult cohorts (Table 2). Notably, ICU patients who received ECMO had a higher prevalence of stroke (ECMO: 7.2%; non-ECMO: 1.6%; OR = 4.68; 95% CI = 3.48–6.28, P < 0.001) and seizure (ECMO: 2.8%; non-ECMO: 1.4%; OR = 2.02; 95% CI = 1.30–3.14, P < 0.001; Supplementary Table 5) than those who did not receive ECMO.
Risk Factors for In-hospital Neurological Complications
Chronic neurological disorder was associated with all neurological complications (CNS infection, seizures and stroke; Figure 2). Specifically, underlying hypertension (aOR = 1.38; 95% CI = 1.25–1.52) and chronic neurological disease (aOR = 1.34; 95% CI = 1.21–1.48) increased the odds of acute stroke (Supplementary Table 6). Among initial neurological manifestations, only altered consciousness and seizure at presentation were consistently associated with in-hospital neurological complications (Figure 2). In other words, patients with acute COVID-19 infection who developed neurological complications more frequently presented with altered consciousness and seizure at admission. As expected, seizure at initial presentation had a strong effect on recurrent seizures (aOR = 69.42; 95% CI = 60.67–79.43; Supplementary Table 6). Altered consciousness at hospital admission was strongly associated with CNS infection (aOR = 5.31; 95% CI = 4.01–7.04) and moderately associated with seizures (aOR = 1.77; 95% CI = 1.55–2.03) and stroke (aOR = 1.95; 95% CI = 1.77–2.15; Supplementary Table 6).
Neurological complications were reported more often among patients who received invasive mechanical ventilation during hospitalization versus patients who did not. The adjusted odds of stroke (aOR = 3.77; 95% CI = 2.74–5.19) indicated higher incidence of stroke reported among ECMO patients, as reflected in unadjusted estimates (Supplementary Table 5). The reported incidence of all complications decreased over time, most notably for stroke, which decreased from 3.5% at the start of the initial COVID-19 outbreak (95% CI = 2.63–4.55) to 0.25% by the end of the study time frame (95% CI = 0.13–0.46; Figure 2). Steady declines in seizure and CNS infection were also observed; however, absolute changes were small in line with low baseline incidence (seizure: 0.64 to 0.44%; CNS infection: 0.63 to 0.004%).
Mortality
Overall, mortality was significantly higher in adults than in children (24.5% versus 2.2%, OR = 14.3, 95% CI = 11.3–18.4, P < 0.001). This contrast held true in both ICU (adults versus children: 32.5% versus 7.4%, OR = 5.99, 95% CI = 4.27–8.71, P < 0.001) and non-ICU settings (adults versus children: 22.2% versus 1.3%, OR = 21.6, 95% CI = 15.6–31.0, P < 0.001). Death was more frequent for patients admitted to the ICU than for those not admitted to the ICU (32.2% versus 21.8%, OR = 1.71, 95% CI = 1.67–1.76, P < 0.001; Table 1). The likelihood of death rose steadily with increasing age, especially after 25 years of age, in both ICU and non-ICU patients, although mortality at any age was lower in non-ICU patients (Supplementary Figure 3). As the COVID-19 pandemic progressed from 2020 to 2021, mortality in the non-ICU cohort decreased significantly but changed little for ICU patients (Supplementary Figure 2).
Among ICU patients with neurological complications, the cumulative probability of death increased over the first 30 days of ICU admission (Figure 3, Supplementary Table 7 and Supplementary Figure 4). In non-ICU patients with stroke, the cumulative probabilities of death and discharge were similar regardless of admission duration (Figure 3).
Figure 3.
Cumulative probability (unadjusted, days) for in-hospital mortality (death) and discharge alive from hospital (discharge) for patients who developed neurological complications. Results are stratified by ICU and non-ICU cohorts.
Brain. 2023;146(4):1648-1661. © 2023 Oxford University Press
