Abstract
BACKGROUND AND HYPOTHESIS: Little is understood about whether the high reported prevalence of dementia in people with psychosis reflects impairment present at psychosis onset. We hypothesized that relying on longitudinal data would help to distinguish stable deficits associated with psychotic disorders from subsequent decline. STUDY DESIGN: We prospectively assessed the Clinical Dementia Rating (CDR) in individuals with first-admission psychosis who have been followed for 25 years (109 with schizophrenia; 135 with other psychoses) alongside 238 demographically matched never-psychotic adults. We scored the CDR sum of boxes using data collected only once at the 25-year follow-up assessment and as a change from baseline, 25 years prior. Impairment was categorized as not observed (NI: CDR < 2.4), mild-moderate (MI: 2.4 ≤ CDR < 4.8), or severe (SI; CDR ≥ 4.8). Prevalence of MI/SI by clinical group was assessed. Multivariable-adjusted risk ratios (aRR) were estimated using multinomial logistic regression. STUDY RESULTS: MI/SI was common when assessed cross-sectionally, but prevalence decreased by 44.4-65.9% when longitudinal scoring was applied. Using longitudinal data, 11.6%/18.2% of participants with schizophrenia and 9.1%/3.8% of those with other psychosis had MI/SI respectively. Participants with schizophrenia were at an increased risk of MI (aRR = 4.98, 95% CI, 1.85-13.46, P = .002) and SI (aRR = 25.98 [3.50-192.75] P = .001) relative to never-psychotic adults. Secondarily, homozygotic apolipoprotein-ε4 carriers (2.6% of genotyped participants) with schizophrenia were at higher risk for SI at midlife (SRR = 2.57 [1.04-6.33] P = .04). CONCLUSIONS: Future work should include observed decline in cognition and functioning to examine cognitive impairment in psychosis. Changes in CDR score may be important when managing treatment.