Context: Systematic reviews offer us a diversity of implications for practice and research, but the uncertainty seems to be a hallmark for consumers, decision-makers and health professionals.

Objective:

To evaluate the proportions of implications for practice and research of the Cochrane systematic reviews.

Design and Setting: Cross-sectional study of systematic reviews published in The Cochrane Library, Issue 4, 2004.

Main Outcomes Measures: 1016 systematic reviews published throughout 50 Cochrane Collaborative Review Groups were analysed randomly. Data extraction was based on the completed systematic reviews. Implications for practice were classified in three categories: "A", evidence supporting the intervention as beneficial; "B", evidence supporting the intervention as harmful; and "C", absence of enough evidence to support recommendations. Implications for research were categorized according to recommending or not further research. Number of included studies and meta-analysis into systematic reviews were also quantified.

Results:

Percentage and 95% confidence intervals (95% CI) of different implications for practice and research of the Cochrane Library were: "A", 44.39 (95% CI, 42.16 - 46.62) %; "B", 6.79 (95% CI, 5.66 - 7.92)%; and "C", 48.82 (95% CI, 46.57 - 51.07)%. 95.96% (95% CI, 95.08 - 96.85)% asked for further researches. 13830 (range 0-292, median 8, mode 2) included studies and 6461 (range 0-177, median 2, mode 0) meta-analysis were computed in all analysed systematic reviews.

Conclusion:

This study systematically identified one aspect frequently noticed in systematic reviews, the inconsistence of evidences to recommend purposed interventions for health care. The results suggest a low methodological quality of primary studies, highlighting the need of efforts and resources to conduct better studies. Thus, systematic reviewers offering a summarized project for future randomized controlled trials would be one of the most important steps. Entities involved with medical research regarding effectiveness, efficacy and safety of human health interventions should be aware of these results.