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 cutoff
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
