MSc

Background: Research has identified psychological factors as important predictors of highly prevelant persistent somatic symptoms (PSS). However, this research is generally not based on primary care data and general practitioners (GPs) experience difficulties identifying PSS in their patients. Awareness of the psychological risk factors noted in readily accessible electronic medical records (EMRs) might aid GPs with earlier discovery and treatment. Aim: The objective was to examine the predictive value of psychological registrations in primary care for PSS onset. Method: We employed a retrospective longitudinal cohort design. EMR data of Dutch primary care patients were allocated into different subsamples according to registrations of irritable bowel syndrome (IBS), chronic fatigue syndrome (CFS), fibromyalgia (FM), and a combination of these registrations (COMBI) with age and sex-matched non-PSS cohorts. Candidate predictors were registered psychological symptoms, mental health...Show moreBackground: Research has identified psychological factors as important predictors of highly prevelant persistent somatic symptoms (PSS). However, this research is generally not based on primary care data and general practitioners (GPs) experience difficulties identifying PSS in their patients. Awareness of the psychological risk factors noted in readily accessible electronic medical records (EMRs) might aid GPs with earlier discovery and treatment. Aim: The objective was to examine the predictive value of psychological registrations in primary care for PSS onset. Method: We employed a retrospective longitudinal cohort design. EMR data of Dutch primary care patients were allocated into different subsamples according to registrations of irritable bowel syndrome (IBS), chronic fatigue syndrome (CFS), fibromyalgia (FM), and a combination of these registrations (COMBI) with age and sex-matched non-PSS cohorts. Candidate predictors were registered psychological symptoms, mental health referrals, and psycho-pharmaceuticals registered prior to PSS onset. The relevance of each candidate predictor was determined via L1 regularization in a logistic Lasso regression. The resulting prediction models’ performance was assessed via area under the curve (AUC). Results: The AUC indicated a fair classification performance for IBS (AUC IBS = .77), and good classification performances for CFS, FM, and COMBI (AUC CFS = .82, AUC FM = .88, and AUC COMBI= .87). The IBS-, CFS-, FM-, and COMBI-models, retained a total of 27, 12, 22, and 15 predictors, respectively. The strongest predictor per model was registration of sexual dysfunction for IBS (OR = 4.0), concentration disorder for CFS (OR = 2.4), neurasthenia for FM (OR = 3.0), and concentration disorder for COMBI (OR = 3.8). Neither of these predictors was unique to one specific model. Each final model retained mental health referrals, psycho-pharmaceuticals, and certain psychological symptom registrations as valuable predictors. Based on shared predictors IBS and FM had the closest models while IBS and CFS models were the most dissimilar. Discussion: These findings indicate that several psychological registrations are valuable predictors for onset of IBS, CFS, and/or FM. Therefore, GPs should consider the according predictors when screening for PSS. Future research could examine the comparatively low classification performce for IBS and psychological factors outside of primary care across PSS subtypes.Show less

Leiden University Student Repository

Documents

In Collections

MSc