Reliability, validity and factor structure of the GHQ-28 used among elderly Iranians

The object of this study was to develop an Iranian version of the General Health Questionnaire-28 (GHQ-28) for use with elderly subjects.

The GHQ-28 Farsi version was evaluated for face validity among 204 elderly subjects aged 59 years or older, chosen randomly from residents of Tehran. The Composite International Diagnostic Interview (CIDI) was used to establish a gold standard diagnosis of mental disorders.

The GHQ-28 was an internally consistent measure. Cronbach's alpha, split-half coefficients and test-retest reliability were 0.9, 0.89 and 0.58 respectively. Four factors were extracted using factor analysis: "depression," "psychosocial activity," "anxiety," and "somatic." Using receiver operating curve (ROC) analysis, the optimum cutoff score for the GHQ-28 in this group was 19/20 (sensitivity 0.83, specificity 0.76). Using a loading of 0.6 or greater, a short form of the instrument (GHQ-15) (alpha=0.9) was derived and correlated well with the longer form of the scale (r=0.97). Using ROC analysis, the optimum cutoff score was 10/11 (sensitivity 0.83, specificity 0.69).

The short and long forms of the GHQ-28 are suitable screening instruments for elderly Iranian residents, particularly those living in urban areas.