A principal component analysis of dietary patterns in bipolar disorder

Considering the critical interaction between diet and mood, this study aimed to identify dietary patterns through a principal component analysis (PCA) in 724 adults diagnosed with bipolar disorder from the Mayo Clinic Bipolar Disorder Biobank. Participants completed the thirteen-item Rapid Eating Assessment for Participants-Shortened Version (REAP-S) to assess diet quality. The number of components to retain was determined based on a scree plot. Regression models were used to assess associations between principal component (PC) scores and clinical characteristics adjusting for age, sex, and site. The mean (SD) age of our sample was 39.6±14.6 and 64.0% were women. Type I BD (59.8%) was the most frequent diagnosis, followed by type II BD (36.9%) and schizoaffective disorder, bipolar subtype (3.3%). The first three PCs were retained, capturing 41.3% of the variation in the data (PC1: 21.3%, PC2: 11.6%, PC3: 8.4%). As anticipated, PC1 was highly correlated with the total REAP-S. PC2 contrasted overall high versus low intake across both healthy and unhealthy foods, while PC3 contrasted primarily lower meat intake alongside more frequent sweets versus higher meat intake with fewer sweets. With the exception of body mass index, no clinical characteristics were significantly associated with these dietary. Two dietary patterns independent of overall diet quality were identified in patients with BD, describing behavioral heterogeneity that could inform future studies of diet and clinical outcomes. • PCA reveals two distinct dietary patterns in bipolar disorder. • The first pattern is defined by an overall healthy vs unhealthy diet. • A second pattern represents a balance between healthy and unhealthy intake. • Findings lay groundwork for personalized diet interventions in BD patients.