Cognitive sequelae are reported in 20-25% of patients following SARS-CoV-2 infection. It remains unclear whether post-infection sequelae cluster into a uniform cognitive syndrome. In this cohort study, we characterized post-COVID neuropsychological outcome clusters, identified factors associated with cluster membership, and examined 6-month recovery trajectories by cluster.

The Mayo Clinic Institutional Review Board approved study protocols. Informed consent was obtained from all participants. Participants (> 18 years old) were recruited from a hospital-wide registry of Mayo Clinic Florida patients who tested positive for SARS-CoV-2 infection from July 2020 to Feb 2022. We abstracted participant health history and COVID-19 disease severity (NIAID score) from the electronic health record and retrieved Area Deprivation Index (ADI) scores as a measure of neighborhood socioeconomic disadvantage. We assessed objective cognitive performance with the CNS Vital-Signs (CNSVS) and subjective neuropsychological symptoms with the Neuropsych Questionnaire-45 (NPQ-45). Results were used as input features in a K-means clustering analysis to derive neurophenotypes. Chi-square and analysis of variance (AnOvA) tests were used to identify clinical and sociodemographic factors associated with cluster membership. Participants repeated the CNS Vital Signs, NPQ-45, as well as the Medical Outcomes Survey (MOS SF-36) and a posttraumatic stress disorder (PTSD) checklist (PCL-C 17) 6 months following initial testing. Repeated-measures ANOVA was used to assess change in neurocognitive performance over time by cluster. Significance was set at P < 0.05.

Our cohort consisted of 205 participants (171 ambulatory, 34 hospitalized) who completed post-acute outcome assessment a mean of 5.7 (± 3.8) weeks following testing positive for SARS-CoV-2. K-means clustering with elbow method fitting identified three subgroups (see figure). The first cluster (N = 31) is characterized by executive dysfunction, greater socioeconomic disadvantage, and higher rates of obesity. The second cluster (N = 32) is characterized by memory and speed impairment, higher COVID severity, prevalent anosmia (70%), and greater severity of memory complaints, depression, anxiety, and fatigue. The third and largest cluster (N = 142) is absent cognitive impairment. Approximately 39% of participants completed the 6-month outcome assessment (N=79). Regardless of cluster membership, verbal memory, psychomotor speed, and reaction time scores improved over time. Regardless of timepoint, cluster 1 (dysexecutive) showed lower scores on cognitive flexibility and complex attention and cluster 2 (memory-speed impaired) showed lower scores on verbal memory, psychomotor speed, and reaction time. Modeling of cluster by timepoint interactions showed a steeper slope of improvement in complex attention and cognitive flexibility in cluster 1 (dysexecutive). Cluster 3 (normal) showed significant improvement in fatigue while cluster 2 (memory-speed impaired) continued to report moderate-severe fatigue, worse medical outcomes, and higher PTSD symptom severity scores at six months.

Most participants were cognitively normal or experienced cognitive recovery following SARS-CoV-2 infection. The 25-30% of participants who showed cognitive impairment cluster into two different neurophenotypes. The dysexecutive phenotype was associated with socioeconomic factors and medical comorbidities that are non-specific to COVID-19, while the amnestic phenotype was associated with COVID-19 severity and anosmia. These results suggest that cognitive sequelae following SARS-CoV-2 infection are not uniform. Deficits may be influenced by distinct patient- and disease-specific factors, necessitating differentiated treatment approaches.