Factors associated with being a false positive on the General Health Questionnaire

T Bell, M Watson, D Sharp, I Lyons, G Lewis

Research output: Contribution to journalArticle

27 Citations (Scopus)

Abstract

Background The General Health Questionnaire ( GHQ) has been used extensively in community and primary care research as an alternative to longer, time-consuming and more expensive assessments of the common mental disorders of depression and anxiety. The sensitivity and specificity of the GHQ compared with longer more detailed assessments is between 70 and 80%. Though satisfactory, this raises the concern about the possibility of bias in relation to longer assessments. We studied socio-demographic factors that were associated with being a false positive on the GHQ in order to investigate any ascertainment bias in relation to more detailed assessments. Method A total of 7,357 consecutive patients aged 16 and over, in five general practices in Cardiff, Bristol and Pontypridd, were invited to complete the 12-item GHQ. Of these, 1,154 patients scored 3 or more, our case definition on the GHQ, and completed a computerised version of the Revised Clinical Interview Schedule (CIS-R) together with a short sociodemographic questionnaire. Results Of the 1,154 subjects who were cases on the GHQ, 344 (30%) (95% CI 27%-32%) were false positive and were not cases on the CIS-R. After adjustment for the other variables, including GHQ score, false positive subjects were more likely to be employed [odds ratio ( OR) 2.7, 95% CI 1.4-5.3], owner-occupiers (OR 1.6, 95% CI 1.0-2.4) and to have a close friend or relative to talk to about personal problems (OR 2.2, 95% CI 1.4-3.5). Conclusion Our results suggest that in this study there was an ascertainment bias on the GHQ in relation to the CIS-R. Studies that use the GHQ to study the relationship between socio-economic status and common mental disorder could lead to biased estimates of association compared to studies that use the CIS-R. It is likely that the GHQ will lead to a higher estimate of prevalence than the CIS-R in subjects who are better off financially and who have better social support.

Original languageEnglish
Pages (from-to)402-407
Number of pages6
JournalSocial Psychiatry and Psychiatric Epidemiology
Volume40
Issue number5
DOIs
Publication statusPublished - 2005

Keywords

  • measurement
  • common mental disorder
  • primary care
  • ascertainment bias
  • validity
  • psychiatric-disorder
  • mental-disorders
  • prevalence
  • community
  • care

Cite this