Distinct and shared contributions of diagnosis and symptom domains to cognitive performance in severe mental illness

Objective: Neurocognitive impairment is prominent in severe mental illness (SMI). We assessed relationships among diagnosis, symptom domains and neurocognitive performance in EMR-ascertained, diagnosis-naive recruited SMI cases, investigated with healthy controls, from the genetically and culturally homogenous "Paisa" population of Colombia. Methods: 1,772 participants (1,220 cases, 552 controls; mean age 42 years, 59% female) were assessed for speed and accuracy across nine tests of the Penn Computerized Neurocognitive Battery (CNB). Cases carried diagnoses of schizophrenia (SCZ, n=141), bipolar-I (BP-I, n=453), bipolar-II (BP-II, n=137) and major depressive disorder (MDD, n=489) based on structured diagnostic interviews. Exploratory factor analysis applied to symptom data produced factor scores. Residualized z-scores were modeled as a function of diagnosis, sex, and all interactions using linear mixed models, with CNB tests as repeated measures. Analyses were repeated in cases, additionally modeling symptom factor scores as main effects. Results: BP-I and SCZ displayed nearly identical impairment patterns on CNB tasks for both accuracy and speed. BP-II or MDD, however, performed similarly to controls with more subtle deficits in specific domains. A four-factor model (psychosis, mania, depression, and suicidality) represented symptom data. Controlling for diagnosis, high psychosis factor scores associated with reduced accuracy and slower speeds. Conclusions: Severity of neurocognitive impairment separates SMI into two profiles (BP-I/SCZ and BP-II/MDD), and associates with lifetime psychotic symptoms across diagnoses. These findings suggest partially independent contributions of psychiatric diagnosis and lifetime symptomatology to cognitive functioning. Genetic investigations may help identify common risk factors predisposing to both psychosis and impaired cognition across SMI categories.