Prevalence of psychiatric disorder in Europe: the potential and reality of meta-analysis

Abstract

Background: Within the European Mental Health Status Project, over 200 psychiatric surveys concerning members of the European Union (plus Norway) were examined for their potential for meta-analysis with regard to prevalence of psychiatric disorders and basic demographic and social variables. The diversity of samples, methods, analysis and presentation was such that only data derived from GHQ-12 and CIDI studies could be used, and those relating to sex differentials only.

Methods: The statistical program "Stata" was used to compute odds ratios (with confidence intervals) for individual studies, and to produce fixed and random effects estimates of the pooled odds ratio for all studies together, and a measure of heterogeneity. Forrest Plots were also produced.

Results: Analysis of GHQ-12 data with a cut-off point of 4, indicating a current or recent "probable mental health problem", showed, as expected, that women had higher prevalence rates than men. However, there was a relatively high heterogeneity score, suggesting that these studies may not be measuring the same thing. Analysis of CIDI results showed homogeneity for major depressive disorder within the last 12 months, with the risk for men about half of that for women.

Conclusions: In terms of advancing epidemiological knowledge, the results are trivial, at most confirming what is already well known. However, the study shows the potential for pooled analysis, with much greater power in epidemiological investigation if consistency could be achieved in research. Various ways in which this might be done are discussed. It also shows the value of personal knowledge and personal networks in fields which are not well handled by electronic literature databases.