13. Systematic Review Article

A systematic review paper should have a structured Abstract of no more than 250 words using headlines as Objective, Data Sources, Study Selection, Data Extraction, Data Synthesis and Conclusions and with 3-10 key words for indexing.

Objective: Give precise statement of the primary objective for the review. Define if the review emphasizes cause and diagnosis, prognosis, therapy and intervention, or prevention. Define if the review would be highly selective as including only randomized controlled trials (RCT) or have wider inclusion criteria.

Data Sources: Present data sources used, including any time restriction.

Study Selection: Describe criteria to select studies for detailed review. Specify methods used, as blinded review, consensus, multiple reviewers.

Data Extraction: Describe how extraction was made, including assessment of quality and validity.

Data Synthesis: Present the main results of the review and state major identified sources of variation between studies.

Conclusion: Give a clear statement of the conclusions made, its generalizability and limitations. The Introduction of the paper could be similar to an original report, but without any longer literature survey, only reviewing shortly previous structural reviews and stating the reason and aim of the present review.

The Methodology section may have subheadings corresponding to the Abstract (Data Sources, Study Selection, Data Extraction) and should include clearly defined and reported inclusion and exclusion criteria, and specification of databases and other formal register, conference proceedings, reference lists and trial authors, which are used as sources. The full search strategy should be given so that it is easy to reproduce. If it is considered too long to be published in the article, an electronic document as an Appendix may be alternative. The stages of selection usually include several steps, each undertaken by at least two independent researchers (identified in the Methods). There will be an initial selection from titles/ abstracts to select the articles to be examined in full. The full articles should be re-screened against the selection criteria. The articles fulfilling the criteria should be subjected to quality assessment. Summarize in a flow chart with the number of articles selected and reasons for rejection at each stage. The quality of the methodology should be assessed having an appropriate tool and also for outcome measures and blinding of outcome assessors. The tool that is most appropriate will depend on the extent and nature of the anticipated research evidence.

The Result sectioncorresponds to Data synthesis in the Abstract and may present tables with long lists of selected articles. Extracted data from trials should, when available, include report of randomization method, study population, intervention methods and delivery, reasons to losses at follow-up, information related to treatment monitoring, post-intervention assessments and follow-up. Report the major outcomes, which were pooled, and include odds ratios or effects sizes. Use when applicable meta-analysis. Numerical values should, when possible, be accompanied with confidence intervals. State the major identified sources of variation between reported studies, as differences in treatment protocols, co-interventions, confounders, outcome measures, length of follow-up, and dropout rates. Tables and figures must be self-explanatory and have appropriate title or caption. The methods for synthesis of evidence should be pre-determined. Sometimes it may not be possible to pool the data, but a synthesis of best evidence ought to be given.

The Discussion sectionshould be structured similar to an original report. The findings should be discussed with respect to the degree of consistency, variation, and generalizability. New contribution to the literature based on the review conducted and where information is insufficient must be stated. Providing the limitations of the review would be helpful. Suggest the need for new studies and future research agenda.

Length of paper: The total length of the text should usually not be more than 5000 words (corresponding to 8-9 printed pages) and in addition tables and the reference list. The reference list should be comprehensive and will therefore often be rather long. However, in the printed version of a review paper normally not more than 100 references will be accepted. If needed and without an upper limit, additional references may be published only electronically with a link to such an Appendix given in the original version of the paper.