Sutton 2018

Appraisal of: Sutton, A.J. Galvan De La Cruz, M.C., Leaviss, J. Booth, A. Searching for trial protocols: a comparison of methods. Res Synth Methods. 2018;9(4):551-560.

Reviewer(s): 

Monika Mierzwinski-Urban

Melissa Severn

Full Reference: 

Sutton, A.J. Galvan De La Cruz, M.C., Leaviss, J. Booth, A. Searching for trial protocols: a comparison of methods. Res Synth Methods. 2018;9(4):551-560.

Short description: 

The purpose of this study was to compare different retrieval methods to identify trial protocols in a systematic review by searching the following sources: published trial reports, journal websites, Google search engine, trial registers, bibliographic databases, and contact via authors. Out of 74 RCTs reviewed, 57% (n=42) of trial protocols were identified when all six resources were searched. The authors found that searching by trial registration number (where available) is the most effective method to identify trial protocols.

The authors also found that for protocols of trial results published prior to 2005, the only effective method of retrieval is to contact authors. No single source was found effective in identifying all protocols post-2005, however, the authors recommended a following prioritization of retrieval methods: Checking the trial publication for protocol details should be conducted first since it is the most efficient method of retrieval (20% of protocols were identified, 1% unique). Trial registers were recommended to be searched as a second stage since they were found to be most comprehensive method of retrieval (51% of protocols were identified, 7% unique) (starting with the search in ICTRP and followed by searches in clinicaltrials.gov, and ISRCTN). If the trial registry number was not identified, the authors suggested to conduct a basic search of trial registries by condition and intervention.  Contacting primary authors should be conducted as a third stage of retrieval since it was also found to be an effective method of retrieval (24% of protocols were identified, 12% unique). Journal website, Internet, or database searching were not found to be efficient and/or effective method of retrieval as they provided moderate to low yield (3%, 31%, and 12%, respectively), were time consuming to search, and did not identify any unique protocols.

Limitations stated by the author(s): 

This study was based on a single systematic review focusing on behavioural interventions, therefore, the authors suggested that to aid generalizability of the findings:

• additional sources may need to be searched for systematic reviews of other interventions (e.g. drug interventions)

•  searching similar studies in different topic areas was also recommended

Only the primary author was contacted, therefore, contacting other authors listed on the trial publications, could potentially increase the response rate from the authors and improve the identification of additional protocols. Also, by sending reminder email to non-responding primary authors could potentially increase the response rate from the authors and increase the rate of identified protocols.

Limitations stated by the reviewer(s): 

In addition to searching CENTRAL and MEDLINE, searching other bibliographic databases such as Embase may yield identification of additional protocols.

Since search results retrieved via Google search engine are not based solely on relevance, it would be warranted to incorporate additional strategies for Internet searching/screening, for

example:

•  Screen more than 20 results via Google search engine

•  Focus Google searches by different regions/countries

•  Conduct the same searches using other search engine(s)

Contacting the sponsor(s) of the trial or conducting targeted searching in other sources such as websites of regulatory agencies or conference abstracts may also help to identify additional trial protocols.

Study Type: 

Single study

Related Chapters: 

Tags: 

• Ongoing trials

• Search engines