Table 6 The effect of considering study limitations when judging quality of evidence

Study limitations

When conducting our comprehensive systematic review of prognostic factors of headaches in children, we reported for example that the evidence of headache severity as a prognostic factor for persistence of headache had serious limitations. This evidence comes from three studies [22–24] and all of them have moderate risk of bias. The sources of potential bias in these studies were: not appropriately accounting for important potential confounders in the design and in the analysis (in [22–24]), not presenting sufficient data to assess adequacy of analysis (in [23, 24]), selective reporting of results (in [23, 24]), and not using any conceptual framework when building a model (in [22–24]). Moreover, we sometimes observed no description of inclusion/exclusion criteria (in [23]), no description of sample frame and recruitment (in [23, 24]), and inadequate information about participation rate (in [22–24]).