Searching trials registries as part of systematic reviews

Photo J Clapton (c) 2021

A (selective) overview of York Health Economics Consortium’s course, presented online during November and December 2021. Disclaimer: this was a very rich two-session short course, and there was far more content than can be summarised here.

Why search trials registries?

  • For a more comprehensive picture of research findings: only 50% of research is published in journal articles
  • To explore the research pipeline (given time lag in formal publication)
  • To avoid research duplication
  • To keep up to date in a field/topic

Features of trial registries

  • There is still no global, universal, legal requirement to register clinical trials and publish trial results, although the situation has improved in the last 20 years.
    In some instances, there are funder and country-specific requirements to register trials
  • Registry entries are not published citation records but databases of trial activity:
    Implications for types of data field, likelihood of similar entries, and exportability
  • Registries vary by nature of coverage:
    by region/country, topic, industry, or company-specific
  • Numerous, including registries of registries. Coverage is incomplete
  • Sometimes registries provide access to results, but may be a different analysis of data (hence different results) compared with published versions

Which are the principal search registries?

  • NIH’s Clinicaltrials.gov.
    Maintainedby NLM at NIH, includes results, information from researchers, also includes other study designs
  • WHO’s ICTRP Search Portal (International Clinical Trials Registry Platform).
    Portal maintained by WHO from other registers. Not comprehensive

Need to search both to get a more complete picture

What other sources are there for research findings?

  • Journal articles
  • Clinical study reports
  • Individual patient data repositories
  • Pharma company websites
  • Clinical study reports and regulatory agency sources: EMA and FDA
  • HTA sources
  • Trial initiatives such as Crossmark
  • Dimensions – trials records and preprints
  • Aggregated databases, e.g. CENTRAL
  • Preprint sources, e.g. MedRxiv

What are trial registry search interface shortcomings?

  • Poor interfaces, lack of proximity or brackets or Boolean operators
  • Often cannot do multiple or long searches
  • Although trial registry data is ‘richer’ than in bibliographic databases, it tends to contain more noise (e.g. exclusion criteria)
  • Check register help pages or Julie and Carol’s website
  • Registers tend to be small; may facilitate a sensitive search
  • Unlikely to be able to translate Medline strategy

What’s the best way to search trial registries?

  • Check for help within the interface, and from the resource compiled by Julie Glanville and Carol Le Febvre https://sites.google.com/a/york.ac.uk/yhectrialsregisters/
  • Choose the most precise concept first, with the fewest records (e.g. intervention). Carry out quick tests
  • Aim to translate all the freetext terms; do multiple simple searches. Terms are likely to be at a technical level.

For example, for Clinicaltrials.gov,

  • Use single search terms
  • Use the generic drug name
  • Condition-related search term
  • Has a term mapping algorithm for synonyms – can’t be turned off, but details does tell you which terms it also searched for. Further down – alternative terms
  • No hyphens or truncation
  • Simple, advanced, expert searches available – can use field tags or API (Application Programming Interface) – can run search as a url and export as xml (there is a link in clinicaltrials.gov)
  • Download options quite sophisticated
  • There is limited research into how to search CT.gov; no information on its reliability

When starting a systematic review

  • Explore whether there is an analysis of the trial registration pattern in your field
  • See this overview of research in searching for unpublished data 2018 Isojarvi, J, Wood, H, Lefebvre, C, Glanville, J. Challenges of identifying unpublished data from clinical trials: Getting the best out of clinical trials registers and other novel sources. Res Syn Meth. 2018; 9: 561– 578. https://doi.org/10.1002/jrsm.1294
  • Key registers are CT.gov, ICTRP. For rich data – go to parent database, for speedy searching go to portals e.g. ICTRP, ScanMedicine (new portal)

How can we export records?

Different interfaces offer different options, but generally export to a citation-based reference management programme will not be complete due to lack of match of the data fields.

Many resources have limited or no download functionality.

Some do provide structured output (E.g. Clinicaltrials.gov)

Options may include:

  • A tab or comma delimited format, which can go into a spreadsheet
  • RIS/Bibtex to go into bibliographic management software (or develop a filter)
  • Xml file export

If no structured output is possible, options are:

  • Get research team to screen online
  • Cut and paste into word
  • Used web-based tools to take screenshots of entries, e.g. Onenote, Evernote, Zotero – build and share a notebook

Impact needs to be built into project planning.

How should we report searches of trial registries?

Documenting is the process of recording what you’ve done, whereas reporting is the external communication of the process. Ideally, the search process would be clear and reproducible, but this is difficult when dealing with sources such as trial registries.

Documenting needs to be proportionate to make best use of time yet communicate the essence of the approach.

Key elements would include search date, source, period covered (or inception), title, link, ID number if relevant, date entered for record.

Search terms may be relevant or search approach, but this aspect could be difficult to explain as likely to be iterative.

Overall

Trial registration is growing, as are the number of registries.

Such registries represent important sources of data that might be lacking in the published literature. However, searching and extracting findings from them continue to be challenging.

%d bloggers like this:
search previous next tag category expand menu location phone mail time cart zoom edit close