Cognitive consequences of <scp>COVID</scp> ‐19 infection: current evidence and future directions

It was evident early in the COVID-19 pandemic that some people experienced persistent symptoms beyond acute infection. Cognitive difficulties (including brain fog and memory complaints) were commonly reported features of what came to be termed Long COVID. Cohort studies soon supported these reports, showing elevated rates of persistent cognitive symptoms, even among those with mild or moderate illness1. Objective assessment of cognitive performance was enabled at scale by advances in online testing platforms. An early example was a citizen science study that assessed 81,337 UK participants during the first national lockdown2. Though not originally intended for this purpose, the scale and timing of the study enabled analysis of cognitive performance in relation to COVID-19 features. Small statistically significant cognitive deficits were observed in the group who reported having recovered from COVID-19 infection, after adjustment for demographic and clinical covariates. These deficits correlated weakly with self-reported fatigue, anxiety or depression, indicating that they were unlikely to be due to affective disturbance. They were most pronounced in people who had been hospitalized, particularly those who had received respiratory support. However, small but measurable deficits were also evident in the non-hospitalized group who had recovered from biologically confirmed SARS-CoV-2 infection with milder acute symptoms. Across domains, the profile included attention, memory and executive deficits. While limited by their incidental nature, these results accorded with patient-reported symptoms, providing converging evidence that COVID-19 infection might have a persistent multi-domain impact on cognitive performance. This hypothesis was subsequently tested in studies with rigorous designs, including the Long COVID extension of the Realtime Assessment of Community Transmission programme3. This programme had utilized random population sampling with biological test confirmation to monitor infection rates in England. Among 276,840 participants who responded to a follow-up survey (REACT-LC) in 2022-2023, 141,583 undertook an assessment on the same online platform (Cognitron) as the prior citizen science research. Compared to a group without known infection, those who had recovered from SARS-CoV-2 showed a small average deficit in cognitive performance. More pronounced deficits were evident in those with ongoing symptoms persisting >12 weeks, consistent with Long COVID. The pattern across domains was again most pronounced for memory and executive tasks, with immediate memory deficits pointing to a role of attention. Deficits were greatest and spanned additional cognitive domains in those who had been hospitalized. Importantly, deficits were more prominent in individuals infected during early relative to late pandemic waves. This, viewed positively, indicated a mitigation of cognitive risk over time, likely through vaccination and viral evolution. Lower-than-expected cognitive performance (at least −1 standard deviation, SD) was observed in 22.5% of individuals with ongoing persistent symptoms, 18.9% of those with resolved persistent symptoms, and 15.7% of the asymptomatic group, with rates of moderate impairment (at least −2 SD) being 5.2%, 3.5% and 2.2%, respectively. Conversely, superior performance (at least +1 SD) was seen in 11.8%, 13.5% and 16.0%, respectively. This cross-sectional pattern indicated that cognitive performance might improve when Long COVID symptoms eventually resolve, although longitudinal confirmation of recovery is pending. REACT-LC was a large-scale study which employed rigorous epidemiological and statistical methods. However, absence of pre-infection cognitive data limited inference about cognitive change. A study from Norway of 111,992 participants with confirmed infection status partially addressed this limitation4. Participant-reported memory function, assessed using the Everyday Memory Questionnaire, was worse up to 36 months after a positive vs. a negative SARS-CoV-2 test, while being essentially the same for these groups before the test. Experimental confirmation of longitudinal change in cognitive function was provided by a UK-based human challenge study5. Using the same Cognitron assessment platform as REACT-LC, healthy young adults were longitudinally monitored from pre- to 1-year post-inoculation with wild-type SARS-CoV-2. Relative to participants for whom the inoculation did not lead to infection, those who developed sustained viral load showed small persistent baseline-corrected cognitive deficits spanning from acute illness until the one-year study endpoint. Performance differences were again most pronounced in immediate memory and executive function. Intriguingly, participants only developed mild illness and did not report persistent cognitive symptoms when debriefed. Further insight into long-term cognitive trajectories came from the PHOSP-COVID cohort6. In follow-up assessments 2-3 years post-hospitalization, patients continued to show objective cognitive impairments. In this more severe group, cognitive deficits occurred alongside psychiatric symptoms, including fatigue and depression. Notably, the scale of cognitive deficit was better predicted by symptom burden at six months than by acute illness severity, highlighting the importance of the early post-acute phase. The biological underpinnings of post-COVID cognitive deficits remain less clear. In a UK-based neuroimaging study of previously hospitalized individuals7, persistent cognitive impairment one year post-infection was associated with elevated serum markers of neuronal injury – neurofilament light chain and glial fibrillary acidic protein (GFAP) – and reduced anterior cingulate cortex volume. Tentative evidence of GFAP elevation was also reported in the wild-type SARS-CoV-2 human challenge study, where illness was mild5. These findings indicate a likely role for immune-mediated neuronal damage across varying severities of acute illness. A recent transdiagnostic review characterized brain fog as a multifaceted symptom cluster involving cognitive, affective and fatigue-related elements, with heterogeneous neural correlates and weak associations with objective task performance8. It highlighted evidence of elevated blood-brain barrier permeability and pro-inflammatory cytokine profiles in individuals reporting brain fog, supporting a neurobiological basis for cognitive symptoms. In addition to large-scale cohorts, frequent monitoring designs have potential for studying the mechanisms that underlie cognitive problems after COVID-19 infection9. They have shown that cognitive symptoms in Long COVID fluctuate and may be exacerbated by mental exertion and fatigue. Such designs could be extended to study whether objectively measured cognitive deficits also fluctuate, and how fluctuations covary with biological and psychological factors within-subject, mapping differences in potential causal mechanisms and informing the search for targeted interventions. In summary, evidence from observational, experimental and mechanistic studies converge to support the hypothesis that COVID-19 infection can have a persistent impact on cognition and brain function. Despite considerable heterogeneity, a characteristic cognitive profile has emerged, with objectively measurable deficits in memory, executive function and attention. The scale of these deficits is greater in those with ongoing persistent symptoms, but may attenuate when their symptoms finally resolve. Objective deficits are weakly correlated with subjective symptoms in Long COVID, but appear more strongly related amongst hospitalized patients. Subjective cognitive symptoms fluctuate in Long COVID, but it remains unclear whether cognitive performance deficits also fluctuate. More research is required to develop an integrative model linking heterogeneity in cognitive symptoms and deficits to underlying psychological and biological mechanisms. Large-scale longitudinal studies and detailed individual monitoring designs can offer tractable approaches for addressing this challenge.