Although primarily a respiratory virus, COVID-19 can adversely affect multiple organ systems, including the nervous system. Subjective cognitive complaints post-COVID infection are common and the rate of impairment on objective cognitive tests vary widely in the reported literature, ranging from 0 to 100%. To better understand the effect of cognitive functioning following COVID-19, it is important to aggregate empirical studies and carefully consider factors recognized to affect outcomes (e.g., disease severity). The following study is a meta-analytic review to elucidate post-COVID cognitive sequalae and factors that may moderate cognitive outcomes.

PubMed and Grey Literature were searched and screened for relevance. Studies were included if participants with confirmed SARS-CoV-2 infection were administered well-established neuropsychological assessments and cognitive performance was compared to either a non-COVID control group or a demographically-corrected normative sample. A random effects meta-analysis was conducted to aggregate effect sizes and meta-regression analyses were used to explore possible moderators of between-study heterogeneity. Primary outcome was cognitive functioning (Cohen’s d) after COVID-19 infection. Moderator variables of interest included demographic variables (age, education, sex of study sample), mean time since infection, COVID severity (mild/moderate, moderate/severe, and mixed), use of comparison group (non-COVID controls vs. normative sample), and if participants were specifically recruited based on subjective cognitive concerns (e.g., Long COVID clinic).

K = 14 studies were included in the analysis. Two studies included two independent sub-groups, resulting in K = 16 independent samples for analysis (N = 945 COVID, N = 354 non-COVID controls). Publication bias was evident, even after trimming outlying effect sizes (Egger’s Regression Intercept = 5.59, t(16) = 4.48, p < .001). Aggregating across all cognitive domains, the observed effect size was small-medium (Cohen’s d = .46, p < .001) with significant between-study heterogeneity (Cochran’s Q(17)  = 92.81, p < .001, I² = 81.68%). Unexpectedly, when entered into a meta-regression, demographic variables (age, education, sex), mean time since infection, and COVID severity were not significant moderators (all p’s > 0.05). Surprisingly, only a model accounting for the control group of the study (normative sample vs. non-COVID control; β = -.504, p = .03) and whether participant recruitment was done based on subjective cognitive difficulties (β = .404, p ­= .07) reached significance (F(2,15) = 4.30, p = .03), though significant between-study heterogeneity remained (Cochran’s Q(15) = 50.81, p < .001, I² = 70.48%) .

Although there has been an onslaught of research examining post-COVID cognitive outcomes, there is significant heterogeneity across studies, limiting firm conclusions regarding post-COVID cognitive sequalae at this time. In addition to careful consideration of methodological differences (e.g., recruitment, comparison group) it remains to be fully ecludated how other factors, such as comorbid mental health conditions, may account for outcomes following COVID infection.