Thread by @Joni_Coleman

I'm back (after a slightly longer lunch) - @juandelahozco is presenting on longitudinal trajectories in the EHR from the Clínica San Juan de Dios, Manizales, Colombia. #WCPG2022

Diagnoses of severe mental illness from ICD10, validated against structured interview and chart review. Some BIP-MDD mismatches (see tweets on @loldeloo talk earlier), but agreement is generally very good, especially for SCZ #WCPG2022

Want to extract presentation information from notes - developed NLP algorithm, required named entity recognition and negation detection. Extracted features align with ICD10 codes #WCPG2022

Diagnostic trajectories. One primary diagnosis recorded per visit. Can have multiple diagnoses across time - replace diagnosis ("switch") if incompatible, add as comorbidity otherwise #WCPG2022

Only 35% have single diagnostic trajectories. 30% have comorbidity, 20% switches, 15% both. Most switches happen from BIP to SCZ or MDD. Typical comorbidities - MDD +ANX, SCZ + substance abuse #WCPG2022

MDD has low prospective stability compared to BIP and SCZ. All disorders have low retrospective stability - most people start with a different diagnosis than they ended with #WCPG2022

One switch predicts another, especially within psychotic disorders #WCPG2022

Read the paper: medrxiv.org/content/10.110… #WCPG2022

Now Vilte Baltramonaityte will talk on the causal role of stress in multimorbidity, particularly using univariable and multivariable MR in MDD, CAD, and T2D #WCPG2022

Numerous stress phenotypes, phenotype is psychocardiodiabetic multimorbidity (the shared variance from the separate phenotypes) #WCPG2022

Stress sensitivity did not strongly correlated with other stress measures. Maltreatment, neuroticism, SESA (sensitivity to environmental stress and adversity) and worry show some degree of MR evidence. Neuroticism excluded because worry and SESA are components of it.

In multivariable analyses, only maltreatment was significant, although removing SESA resulted in worry being significant too #WCPG2022

Is this just effects on MDD? Yes, although still see worry-multimorbidity effect if restrict to same SNPs. Not so for maltreatment #WCPG2022