Acute cognitive complications following COVID-19 infection have been appreciated in a subset of patients since the early months of the global pandemic. Emerging data reveal that some patients go on to experience cognitive improvement, whereas others may experience further cognitive decline. We aimed to assess trajectories and predictors of cognitive change in a sample of post-COVID-19 patients.

This prospective cohort study assessed longitudinal cognitive change in adults receiving care for COVID-19 in the Johns Hopkins Post-Acute COVID-19 Team (JH PACT) clinic. Participants self-administered the Digital Automated Neurobehavioral Assessment (DANA) battery of seven cognitive tests and a performance-based measure of cognitive fatigue on up to six occasions over six weeks. Improvement or decline between the first and last assessment was defined as change of >1 standard deviation of the baseline mean of each outcome. Potential predictors of change included demographic features (age, sex, race/ethnicity, education), COVID-19 illness characteristics (hospitalization or ICU stay, months since symptom onset), and comorbid disease burden. Analyses included measures of central tendency, independent samples t-tests, and chi-square tests of independence.

Of the 36 enrolled participants, 29 (81%) completed at least one DANA assessment (M = 4.7 assessments, SD = 1.8). Those completing at least three assessments (n = 24, 66.7%) were included in the present analyses (71% female; 58% white; M age = 54 years, SD = 10.9; M education = 14.6 years, SD = 2.4; M months since COVID-19 symptom onset at recruitment = 9.8, SD = 4.7; M comorbidities = 2.8, SD = 2.0). Fatigue was the most frequently improved outcome measure, with 41.7% of participants scoring >1 standard deviation above the baseline mean fatigue score at their final assessment. Among cognitive outcomes, the greatest frequency of improvement was observed on tests assessing rapid spatial processing (37.5%), processing speed (33.3%), and memory (33.3%). There were no consistent predictors of improvement, but several subtest-specific findings emerged. Specifically, (a) more comorbidities were positively associated with rate of fatigue reduction (p = .04), (b) longer duration since COVID-19 illness was positively associated with rates of memory improvement (p = .02), (c) older age, male sex, and more comorbidities were positively associated with rate of improvement in reaction time (ps < .05), and (d) more assessments completed was positively associated with rates of improvements in working memory (ps < .05). Response inhibition (12.5%), simple reaction times (16.7%), and working memory (16.7%) showed the lowest rates of improvement over time. Declines in cognition were infrequent, with 4.2 - 8.3% (n = 1 to 2) declining on measures of procedural reaction time, spatial processing, inhibitory control, or working memory.

At an average of >9 months following acute COVID-19 illness, we observed longitudinal improvements in cognitive fatigue as well as processing speed, memory, and spatial reasoning. Consistent predictors of recovery were not identified, although age, sex, comorbid conditions, and time since illness predicted rates of improvement in select domains. Further analyses with a larger sample size and more stringent analyses are needed to confirm and extend these findings.