It is easy to design a search strategy that identifies all evidence about an intervention by creating search strings for the population and disease characteristics, the intervention, and possibly the comparator. When a review aims to include randomized trials only, various approaches are available to restrict the search strategy to randomized trials (see Chapter 6).
a) Search for previous reviews of the review question.
b) Use resources, such as CENTRAL or CRG-specific registers, that are ‘rich’ in randomized trials.
c) Use methodological filters and indexing fields, such as publication type in MEDLINE, to limit searches to studies that are likely to be randomized trials.
d) Search trial registers.
To restrict the search to particular non-randomized study designs is more difficult. Of the above approaches, only (a) and (b) are likely to be at all helpful. Review authors should certainly search CRG-specific registers for potentially relevant NRS. Some CRGs (e.g. the EPOC Group) include particular types of NRS in CRG-specific registers (authors should check with their CRG). The process of identifying studies for inclusion in CENTRAL means that some, but not all, NRS are included, so searches of this database will not be comprehensive, even for studies that use a particular design. There are no databases of NRS similar to CENTRAL.
As discussed in Section 13.2.1.3, study design labels are not used consistently by authors and are not indexed reliably by bibliographic databases. Strategy (c) is unlikely to be helpful because study design labels other than randomized trial are not reliably indexed by bibliographic databases and are often used inconsistently by authors of primary studies. Some review authors have tried to develop and ‘validate’ search strategies for NRS (Wieland 2005, Fraser 2006, Furlan 2006). Authors have also sought to optimize search strategies for adverse effects (see Chapter 14, Section 14.5) (Golder 2006b, Golder 2006c). Because of the time-consuming nature of systematic reviews that include NRS, attempts to develop search strategies for NRS have not investigated large numbers of review questions. Therefore, review authors should be cautious about assuming that previous strategies can necessarily be applied to new topics.