The SLR specification manual

The Systematic Literature Review Specification Manual is the result of detailed expert consultation with the Methodology Task Force for the conduct of systematic literature reviews (SLRs) relevant to food, nutrition, physical activity and the causes of cancer worldwide.

A novel methodology

The methodology in this manual drew on the accumulated knowledge and success from other centres of excellence but had many important differences from previous methodologies. In particular, it did not employ a strict hierarchy of evidence approach to reviewing the literature.

An important aspect of a systematic literature review is that all stages of searching, selection, assessment and analysis are pre-specified, objective and reproducible, openly documented and subject to peer review at critical stages.

The SLR manual provides full details of the approach taken and is designed to ensure a comprehensive and consistent approach to the analysis and also a common format for displaying the evidence.

Testing the methodology

The SLR manual was tested by a process set up to evaluate whether two independent centres, in two continents, drew similar conclusions regarding the association of food, nutrition and physical activity and endometrial cancer, when provided with the same general instructions and with availability to similar resources.

The findings of this process published as an journal article suggested that while the two SLRs showed some differences in terms of numbers of citations retrieved or decisions on relevance, the overall conclusions, particularly regarding which studies were most important and pooled risk estimates, were comparable.

This invaluable testing process enabled the SLR Specification Manual to be amended and improved, based on practical experience, prior to the other SLRs being undertaken.

Using the SLR Specification Manual

First stage

The first stage of an SLR was a comprehensive search of the scientific literature and other sources catalogued on electronic databases, using all relevant keywords and terms.

The papers identified were assessed for relevance using reproducible criteria. Study characteristics and results were extracted and recorded. Data from different studies were combined and analysed using meta-analysis when appropriate. Existing SLRs in the literature were also identified to ensure, as far as possible, that all relevant papers were included.

The SLRs include evidence published up to the end of 2005, and the Panel's conclusions are based on those SLRs. To ensure that the Panel's recommendations, which are derived from their conclusions and judgements, took into account developing evidence, a further review of studies published during 2006 was conducted. This review was more limited than the full SLRs, being confined to exposures that had been judged "convincing", "probable", "substantial effect on risk unlikely", and "limited-suggestive", based on the SLRs.

Second stage

At this second review stage, no further meta-analyses were performed and a review of study quality was not included. For these reasons, the results of this 2006 review have been noted but have not been used to alter the Panel's judgements based on the full SLRs.

The SLR Specification manual and systematic literature reviews can both be accessed from our resource downloads section.

Modifications to the methods for the Continuous Update Project (CUP)

The following changes to the methods were made to limit the amount of work for the CUP team without reducing the quality of the reports produced.

• Literature search uses Medline only
• Search includes randomised controlled trials, cohort and case-control studies
• Information on case-control studies is not analysed if there are large numbers of cohort studies
• Publications are restricted to those in English language
• Studies reporting relative risks or odds ratios for cancer incidence are included in meta-analyses and forest plots.
• Studies reporting on mortality are not included in the analyses.