[Practice of systematic reviews. VI. Searching, selection and methodological evaluation of etiological research]

Optimal search strategies for the identification of publications on aetiological research are currently not available. The definition of adequate selection criteria is necessary to limit the harvest to a relevant and homogeneous set of publications. Yet, the harvest should be high in order to prevent many relevant papers being missed. The methodological assessment of aetiological research should be aimed at the estimation of the presence of selection bias, information bias, and confounding. Important items of the methodological assessment are independent and standardised assessment of exposure and outcome, high response rate, no selective dropout, and adjustment for confounding in the analysis.