The ENT Trials Register contains reports of randomised controlled clinical trials and quasi-randomised controlled trials (RCTs/qRCTs) related to the scope of the Group. The Register was established in 1998 and contains 27,000 reports of trials in ENT/head and neck cancer (as of May 2018). The majority of these are submitted for inclusion in the Cochrane Central Register of Controlled Trials (CENTRAL) on a monthly basis.
Bibliographic databases and trial registers are searched to identify reports of trials for inclusion in the ENT Trials Register. The core list of resources for electronic searches includes:
|CENTRAL (Cochrane Register of Studies (CRS))||Monthly|
|Web of Science (Web of Knowledge)||Monthly|
|CAB Abstracts (Ovid)||Quarterly|
|LILACS (Latin American and Caribbean Health Science Information database)||Annually|
|ICTRP (World Health Organization (WHO) International Clinical Trials Registry Platform)||Annually|
In the past we also included searches of:
|Embase (Ovid)||2015||Embase (Ovid) is no longer searched to populate the ENT Trials Register. Work done by Cochrane centrally to search Embase for RCTs and qRCTs, supported by crowd screening (read further details here), is deemed to be sufficient to identify all the relevant trials in ENT and no additional searches are required. This decision was based on a study conducted by the Cochrane ENT Information Specialist in 2014. The ENT Trials Register is populated with Embase records using searches of CENTRAL.|
|CINAHL (EBSCO)||2017||CINAHL (EBSCO) is also no longer searched to populate the ENT Trials Register. A search of CINAHL covering the whole of 2016, conducted by the Cochrane ENT Information Specialist, identified no unique references other than those identified by other sources.|
|BIOSIS Previews||2012||Cochrane ENT no longer have access.|
|ClinicalTrials.gov||2017||Work done by Cochrane centrally to search ClinicalTrials.gov for RCTs and qRCTs, supported by crowd screening (read further details here), is deemed to be sufficient to identify relevant trials in ENT and no additional searches are required.|
|KoreaMed||2015||Work done by Cochrane centrally to search KoreaMed for RCTs and qRCTs, supported by crowd screening (read further details here), is deemed to be sufficient to identify relevant trials in ENT and no additional searches are required.|
|IndMed||2013||Only one unique trial was identified between 2008 and 2013.|
|PakMediNet||2013||Similar to IndMed.|
|CNKI Google Scholar)||2015||It is not practical to search CNKI regularly because of the limitations of the search interface. Trials identified as part of searches for reviews are routinely added to the Cochrane ENT Trials Register.|
|Google Scholar||N/A||It is not practical to search Google Scholar regularly because of the limitations of the search interface. Trials identified as part of searches for reviews are routinely added to the Cochrane ENT Trials Register.|
The search strategies (available here) are based on the subject scope of Cochrane ENT. They have been adapted for use across all bibliographic databases. RCT search filters are used in combination with the subject strategies for databases in order to identify reports of trials. The filters are adaptations of the highly sensitive search strategy designed by Cochrane for identifying randomised controlled trials and controlled clinical trials (as described in the Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0, Box 6.4.b. (available from www.cochrane-handbook.org)). Regular testing and refining of subject strategies and search filters is undertaken to optimise the retrieval of trial reports relating to the scope of Cochrane ENT.
Over time some search strategies have been modified. Details of the previous versions and the strategies for the databases no longer searched are available here.
Search results are uploaded to the Cochrane Register of Studies (CRS), where the results are deduplicated. Records are then run through a machine learning tool available in CRS, called the classifier. The classifier assigns a probability (from 0 to 100) to each citation of it being a true RCT as well as a probability that it is relevant to the scope of our review group. For citations that are assigned a probability score of less than 10 of being a true RCT, the machine learning classifier currently has a specificity/recall of 99.987%. Further details are here.
The Cochrane Information Specialist then screens those records with a greater than 10% likelihood of being a RCT and relevant to ENT. The Information Specialist also screens those records without an abstract (because the classifier does not perform as well on these records). The records with scores below 10% are put to one side and not screened. This process reduces the number of records needing to be screened by between 50% and 70%.
The records are screened for inclusion in the Cochrane ENT Register and the unique records are added to CENTRAL.
Historically we have undertaken a number of handsearching activities, including conference proceedings and hard copies of journals. The details of these trials were added to the Cochrane ENT Trials Register. Details of these searches are available here.
The abstracts of conference proceedings are now widely available through Embase and Web of Science, and Cochrane ENT no longer has regular access to the abstract books of conference proceedings.
If you would like to get involved with handsearching activity please contact us.
Please contact us if you would like access to the Cochrane ENT Trials Register.
RCTs and rRCTs identified through non-systematic searches of Google Scholar, CNKI, Indmed, PakMediNet and KoreaMed, performed as part of the searches for reviews, are added to the Cochrane ENT Register and CENTRAL.
In 2018, all search results for reviews are deduplicated against the current Cochrane ENT Register and run through the machine classifier. All unique references with scores above 10, or with no abstract, are screened and added to the Cochrane ENT Register.
Since March 2018, we have piloted an additional search of CENTRAL. We identify all new references added to CENTRAL (without any search terms) in the last month that are not currently part of the Cochrane ENT Register. The Information Specialist screens those records with a higher than 10% probability of being a RCT and a greater than 20% probability of being relevant to ENT. This identifies just over 100 additional records to screen each month.
As technology streamlines the process for identifying RCT and qRCTs relevant to the scope of Cochrane ENT, we will continue to experiment with additional ways of identifying unique trials.