Related Experiment Videos
Why GHQ threshold varies from one place to another
D P Goldberg1, T Oldehinkel, J Ormel
1Institute of Psychiatry, London.
Psychological Medicine
|September 2, 1998
Summary
The best threshold for the General Health Questionnaire (GHQ) varies with diagnosis prevalence. Higher thresholds correlate with better GHQ performance, but lower thresholds protect sensitivity in diverse settings.
Area of Science:
- Mental Health Research
- Psychometric Evaluation
- Epidemiology
Background:
- The optimal threshold for the General Health Questionnaire (GHQ) varies across different settings, lacking a clear explanation.
- Understanding this variation is crucial for accurate mental health screening.
Purpose of the Study:
- To investigate the factors influencing the optimal threshold for the GHQ in diverse populations.
- To identify the relationship between diagnostic prevalence and GHQ performance.
Main Methods:
- Analysis of data from a World Health Organization (WHO) study involving the GHQ and the Composite International Diagnostic Interview (CIDI) in 15 cities.
- Statistical examination of the correlation between diagnostic prevalence, GHQ discriminatory power, and optimal threshold selection.
Main Results:
- The prevalence of single and multiple diagnoses did not vary significantly in relation to the number of CIDI symptoms across cities.
- The optimal GHQ threshold was strongly associated with the prevalence of both single and multiple diagnoses within a center.
- Higher thresholds correlated with improved GHQ discriminatory power (Area Under the ROC Curve), while item performance varied by location.
Conclusions:
- The prevalence of multiple diagnoses is a key determinant of the GHQ threshold, with higher prevalence linked to higher thresholds.
- The mean GHQ score can serve as a general guide for the optimal threshold.
- In centers with lower GHQ discriminatory power, a reduced threshold is necessary to maintain sensitivity, albeit at the cost of lower positive predictive value.