Risks of neurological and psychiatric sequelae 2 years after hospitalisation or intensive care admission with COVID-19 compared to admissions for other causes

Abstract

The association between COVID-19 and subsequent neurological and psychiatric disorders is well established. However, two important questions remain unanswered. First, what are the risks in those admitted to intensive care unit (ICU) with COVID-19? Admission to ICU is itself associated with neurological and psychiatric sequelae and it is not clear whether COVID-19 further increases those risks or changes their profile. Second, what are the trajectories of neurological and psychiatric risks in patients admitted to hospital or ICU with COVID-19, and when do the risks subside? We sought to answer these two questions using a retrospective cohort study based on electronic health records (EHR) data from the TriNetX Analytics Network (covering 89 million patients, mostly in the USA). Cohorts of patients admitted to hospital or ICU with COVID-19 were propensity score-matched (for 82 covariates capturing risk factors for COVID-19 and more severe COVID-19 illness) to patients admitted to hospital or ICU (respectively) for any other reason. Matched cohorts were followed for up to two years and the risk of 14 neurological and psychiatric outcomes were compared. A total of 280,173 patients admitted to hospital and 46,573 patients admitted to ICU with COVID-19 were successfully matched to an equal number of patients admitted to hospital or ICU for any other reason. Those hospitalised with COVID-19 were found to be at a greater risk of a range of neurological and psychiatric outcomes including seizure/epilepsy, encephalitis, myoneural junction/muscle disease, Guillain-Barré syndrome (GBS), dementia, cognitive deficits, psychotic disorder, mood and anxiety disorders, but not ischaemic stroke or intracranial haemorrhage. When risks were elevated after COVID-19, most remained so for the whole two years of follow-up (except for mood and anxiety disorders). Risk profiles and trajectories were substantially different among those admitted to ICU: compared to those admitted for any other reasons, those admitted with COVID-19 were at a greater risk of myoneural junction/muscle disease, GBS, cognitive deficits and anxiety disorder, but at a significantly lower risk of ischaemic stroke, intracranial haemorrhage, encephalitis, and mood disorder. When elevated, the risks in those admitted to ICU with COVID-19 were mostly short-lived. In summary, risks of neurological and psychiatric sequelae in patients hospitalised with COVID-19 are wide ranging and long standing whereas those in patients admitted to ICU with COVID-19 are similar to, or lower than, the risks observed post-ICU admission for any other cause. These contrasting risk trajectories are relevant for researchers, clinicians, patients, and policymakers.

Keywords: Electronic health records; ICU; Neuropsychiatric; SARS-CoV-2; long COVID; post-acute COVID-19 syndrome.