Abstract
The GHQ-28 was validated against the short PSE in a New Zealand community study of female psychiatric morbidity. The GHQ-28 total scores were significantly correlated with the PSE scores. Higher coefficients were obtained using the scoring method of Good child and Duncan-Jones than with the standard scoring method. In this data set, the 3/4 cutoff had the best sensitivity and specificity. The correlations of the GHQ-28 subscales with ICD diagnostic classes and ad hoc PSE sub-scores were also statistically significant. Because the distribution of the GHQ-28 scores is positively skewed, non-parametric statistics may be preferable to the traditional Pearson's correlation coefficient. Overall, the results from this study confirm the GHQ-28 to be a valid and practical screen for presence or absence of psychiatric disorder in New Zealand women.
