Uncovering psychiatric test information with graphical techniques of Exploratory Data Analysis

Area of Science:

• Psychiatry
• Health Services Research
• Data Science

Background:

• Conventional statistical methods are standard for evaluating psychiatric tests.
• Exploratory Data Analysis (EDA) offers complementary approaches to uncover complex patterns.

Purpose of the Study:

• To evaluate the predictive ability of repeated psychiatric screening tests (General Health Questionnaire [GHQ]) for healthcare service utilization.
• To identify subpopulations based on GHQ scores and their relationship with service use and Health Maintenance Organization (HMO) dropout.
• To demonstrate the utility of EDA in psychiatric research.

Main Methods:

• Utilized two interactive computer programs for EDA, including three-dimensional graphs and scatterplot matrices.
• Analyzed data from two stratified random samples of new HMO enrollees (n=244 and n=213).
• Applied EDA to General Health Questionnaire (GHQ) scores to predict medical/psychiatric service use and HMO dropout.

Main Results:

• Identified two subpopulations: low GHQ scorers (no service use prediction) and high GHQ scorers (service use prediction).
• Observed that improving GHQ scores predicted increased service use, while chronically high scores predicted diminished use.
• Found that high and unchanging GHQ scores predicted HMO dropout, suggesting potential patient immobilization and disengagement.

Conclusions:

• EDA methods can reveal nuanced findings missed by conventional statistics in psychiatric evaluations.
• Chronically high or unchanging GHQ scores may signal patients at risk for disengagement and dropout from healthcare systems.
• EDA's visual and interactive capabilities enhance the understanding of complex patient behaviors and outcomes in mental health research.