The Prediction of Diagnostic Change From Bipolar Disorder to Schizophrenia and Schizophrenia to Bipolar Disorder in a Population-Based, Longitudinal, National Swedish Sample

Abstract

Abstract Background and Hypothesis To clarify, in a large, representative, longitudinal sample, the rate and predictors of diagnostic conversion from Bipolar Disorder (BD) to Schizophrenia (SZ) and from SZ→BD. Design From individuals born in Sweden 1950–1995 and living there in 1970 or later, we identified at least one initial diagnoses of SZ (n = 8449) and BD (n = 8438) followed for a minimum of 10 and a mean of 24 years. Diagnostic conversion required, respectively, at least two final diagnoses of BD and SZ 30 days apart with no intervening diagnosis of SZ or BD. Results At follow-up, rates of BD→SZ and SZ→BD conversion were 10.1 and 4.5%, respectively. Conversions occurred slowly, with around 50% completed in the first decade. Using a diverse range of variables available at first onset including family genetic risk scores, BD→SZ conversion was predicted with greater accuracy (AUC = 0.78) than SZ→BD conversion (AUC = 0.65). The strongest predictors of BD→SZ conversion were earlier years of birth, younger age at BD onset, low BD genetic risk, and being unmarried at BD onset. SZ→BD conversion was most strongly predicted by high BD genetic risk, being married at SZ onset, female sex, early age at SZ onset, and an MD episode prior to SZ onset. Cases of BD and SZ in the highest decile for conversion risk had HRs for a diagnostic change of, respectively, 12.5 and 3.4. Conclusions Diagnostic conversion of BD→SZ and SZ→BD are not rare, are moderately predictable, and should likely be accounted for in many research designs.