Suitability of three different tools for the assessment of methodological quality in ex post facto studies

There is no clear candidate tool for assessing the methodological quality of ex post facto studies in systematic reviews and meta-analyses yet. Our purpose is to thoroughly analyze the psychometric properties of the three most comprehensive assessment tools of this kind published up to 2010. We selected these tools from a previous systematic review, and we applied each one to assess the quality of 10 prospective studies, 10 retrospective studies with quasi-control group, and 10 cross-sectional studies. Inter-rater reliability for the first two aforementioned research designs is moderate only for one of the selected tools, and moderate to high for all of them for cross-sectional studies. Agreement between tools is low in general, although the inferred aspects show that the tools have a relative good conceptual overlapping in most of the domains. According to these results we recommend two tools for assessing cross-sectional studies, but we consider that the tools applicable to prospective studies or retrospective studies with quasi-control group require further testing. The 30 concrete aspects that we have inferred from the items of the three analyzed tools can be used as starting point to develop a new tool of this kin