Discussion
In this study to characterize neurological manifestations of COVID-19 among hospitalized adults and children in the ISARIC registry, we found that non-specific symptoms of fatigue and altered consciousness were the most common at admission. Altered consciousness was 3.5 times less common in children than in adults, whereas seizure (as an initial manifestation) was 5 times more frequent in children. Altered consciousness and seizure at admission were strong risk factors for in-hospital neurological complications after adjusting for covariates (Figure 2). Although there are limited data in CSF or imaging data to establish the causality or direction association, an important clinical implication of this analysis is that the possibility of CNS infection should be considered for patients presenting with seizures or altered consciousness at the time of hospital admission for COVID-19. Neurological manifestations on presentation, such as anosmia, ageusia, fatigue and myalgia, were more common in adults admitted to the ICU than in those admitted to a non-ICU floor. However, caution is needed when interpreting these results, as these non-specific neurological symptoms are reported in up to 80% of surveyed patients with COVID-19.[7,8]
In-hospital neurological complications were infrequent in our cohort, with 1.5% for strokes, 1.0% for seizures and 0.2% for CNS infections. These rates are in keeping with prior data on adults with COVID-19.[7,8] Authors of the Global Consortium Study of Neurologic Dysfunction in COVID-19, which used detailed definitions of neurological complications for hospitalized patients, reported a 3% incidence of strokes, 1% incidence of seizures and <1% incidence of CNS infection.[7,8] In the International Multicentre Coronavirus Disease 2019 Critical Care Consortium Study, acute stroke was reported in 2.2% of patients, with haemorrhagic stroke being the dominant type in ICU patients. That study also noted that this risk was 10 times higher in the subset of patients receiving ECMO.[7,8]
Overall mortality was lower in our study cohort at 24.1%, likely because the proportion of patients who required ICU care was relatively lower (22.3%) compared to a systematic review and meta-analysis of 24 983 patients demonstrating 32% ICU admission and 39% in-hospital mortality.[15] Although neurological complications were not common in our study, they have been noted to be the most strongly associated with reduced ability for self-care and worse functional outcome on hospital discharge.[16] In our study, such complications were also associated with in-hospital mortality in our multivariable model estimates (Supplementary Table 7). Therefore, given the high prevalence of COVID-19, neurological complications will be a substantial global public health and social care burden in the near future.
Our study showed that the cumulative probability of in-hospital mortality increased most acutely in the first 30 days for ICU patients who had in-hospital neurological complications and was most pronounced for those with stroke. However, it continued increasing up to 100 days after hospital admission, emphasizing the importance of vigilant neurological evaluation for patients with long hospitalizations (Figure 3) as large vessel occlusion in acute ischaemic stroke is common (>20%)[17] and early detection with standardized neuromonitoring may improve the neurological outcome in ICU patients.[18] Also, it is important to note that the rate of change in the cumulative probability curves decreased over time, indicating the risk and hazard of neurological complications are high early in the disease course. In a previous study that used a 31-day follow-up, the increased frequency of ischaemic stroke was 10 times higher than normal in the 14 days after a COVID-19 diagnosis, and the risk remained up to 6 times higher than normal at 31 days.[14,19] The risk of acute myocardial infarction was also assessed to be 5 times higher in the 14 days after a COVID-19 diagnosis. The authors postulated that the underlying mechanisms may include cytokine-mediated plaque destabilization and hypercoagulability.[14,19] This is likely in line with the fact that early variants were associated with more severe illness requiring hospitalization and ICU admission.[20]
Notably, our study showed a dramatic decline in stroke frequency whereas seizure frequency remained steady over time (Figure 2). Several possible explanations might account for the decrease in stroke frequency during the pandemic. Treatment of COVID-19 changed rapidly after the initial clinical experience, such as with widespread use of high-intensity thromboprophylaxis and avoidance of mechanical ventilation (with the concomitant need for more sedation), for example. Global trends in these management approaches may have reduced the impact of COVID-19-related coagulopathy or reduced hypotension and shock associated with aggressive use of mechanical ventilation. Another possibility is that early variants of SARS-CoV-2 had greater inflammatory and coagulopathy effects. Other explanations are that resources for neuroimaging became reduced as the pandemic progressed, with parallel reductions in surveillance for stroke, or that the initial population of patients enrolled in the registry had a greater baseline risk of stroke, before public health messages about high-risk, vulnerable groups taking extra precautions against contracting COVID-19 became widespread. More research is needed to better understand the factors related to this strong trend.
Evidence regarding the neurological effects of COVID-19 in children is more limited than that for adults. Our study included 2365 patients younger than 18 years and noted a different profile and frequency of neurological manifestations in this cohort. Except for seizures, all neurological manifestations and complications were less frequent in children than in adults. Interestingly, we showed a linear decrease in the prevalence of seizures as age increased. This finding is likely consistent with paediatric seizures where febrile seizure or CNS infection related seizures are more common in younger age.[21] A similar pattern was observed for CNS infection, which decreased with age, whereas the prevalence of stroke increased sharply with increasing age (Figure 2). A prevalence study in the UK paediatric and adolescent population (<18 years) identified neurological and psychological complications in 52 cases of 1334 children linked to COVID-19.[6] The authors reported a 0.4% incidence of CNS infection and 0.07% incidence of transient ischaemic attack.
Severe illness requiring ICU admission was closely associated with in-hospital neurological complications. Invasive mechanical ventilation and especially extracorporeal support were associated with elevated risks of neurological complications (Supplementary Table 5). The risk of stroke was 8.3% among those receiving extracorporeal support, substantially higher than the 4.5% frequency reported in the Extracorporeal Life Support Registry among patients receiving veno-venous ECMO for non-COVID-19 acute respiratory distress syndrome (ARDS).[22]
Limitations
The spectrum of neurological manifestations and complications of COVID-19 is broader than the CRF terms included in the patient registry. Patient recruitment strategies varied between sites and were subject to staff and resource limitations, introducing the possibility of recruitment bias in some sites. Challenges exist in defining and capturing neurological manifestations[1] and in establishing causation,[23] especially in complex ICU patients with ARDS and when COVID-19 therapies can have iatrogenic neurological side effects. For analysis of neurological manifestations, we used data available at the time of hospital admission; however, data availability on neurological complications was limited to reports at any time during hospitalization. Having neurological manifestations at admission such as seizure or altered consciousness may have biased clinicians to investigate and find more in-hospital neurological complications. The use of sedative and analgesic drugs within the ICU setting may have influenced the reporting of these variables. Additional information on the timing of neurological complications would have allowed for more detailed analysis on the risk of complications over the course of hospitalization and time-dependent associations with mortality, as well as the timing of invasive ventilation and ECMO relative to the development of complications.[8] Certain variables such as GCS and smoking status where there is a large missingness in data should be interpreted carefully. Also, the absence of control patients without COVID-19 in the ISARIC dataset prevented estimation of specificity or positive and negative predictive values. Future studies need to more rigorously apply standardized definitions, report temporal relationships and exclude alternate aetiologies to better differentiate primary COVID-19 neurological sequelae from associated comorbidities and iatrogenic causes. Additionally, it is essential that ongoing research into neurological complications of COVID-19 record vaccination status, the temporal relationship between presentation and neurological complications and the specific COVID-19 variant affecting the patient. Finally, we did not investigate Guillain-Barré syndrome (GBS), as it was not part of the CRF. However, it is of particular relevance given the COVID-19 vaccine hesitancy worldwide. GBS is reported to occur at 145 excess cases per 10 million people after a positive SARS-CoV-2 test, which is greater than the 38 excess cases of GBS per 10 million people receiving ChAdOx1 nCoV-19 vaccinations.[24]
Brain. 2023;146(4):1648-1661. © 2023 Oxford University Press
