[The values of different study designs on the levels of evidence: a descriptive analysis of the researches published in four general medical journals in 2009].

Objective: To discuss the levels of evidence provided by different study designs.

Methods: Websites of N Engl J Med, JAMA, Lancet, and BMJ were accessed to identify research articles (systematic review and meta-analysis included) published in 2009. A standardized data collection form was established using Epidata 3.1 software to extract the "title", "country of lead author", "clinical problem" (such as treatment, diagnosis, etc.) and "study design" of eligible studies. Descriptive statistics was conducted with SPSS 13.0.

Results: Over all, 844 studies were included, among which 35.7% were RCT, 9.4% systematic review and Meta-analysis, and 54.9% other types of studies. Regarding clinical problems, 34.2%, 19.7%, 13.7%, 6.0% and 5.1% of the included researches addressed the issues of treatment, etiology/risk factors, prevention, disease frequency and prognosis, respectively. The study designs that were most frequently adopted to explore these problems were RCT (70.6%), cohort study (44.6%), RCT (68.1%), cross-sectional study (56.9%), and cohort study (93.0%), respectively.

Conclusions: High-level evidence does not come exclusively from RCT and systematic review, as each type of study may have its unique value in health related research. The clinical problem of interest, the previous work that has been done to approach the same issue, as well as other factors should be taken into account when deciding whether the selected study design is appropriate.