Containing over 28 million citations, PubMed offers a treasure trove of data from across the life-sciences. But honing down on what you’re looking for can be challenging. We explore what it has to offer on clinical trials – and how easy is it to find.

The PubMed database offers a wealth of information on clinical trials. A quick search by publication type revealing an overwhelming 844,661 clinical studies or 796,275 clinical trials, with a subset further categorized by phase:

  •      Phase I, 18,047 records
  •      Phase II, 29,121 records
  •      Phase III, 13,718 records
  •      Phase IV, 1,522 records

With so much information available, pulling out what’s wanted can be a limiting factor. Several tactics can be deployed, with the best strategy depending on what the search involves. We set out to explore some of these approaches using a case study.

The TACT2 clinical trial

TACT2 is a phase III, multi-center randomized clinical trial led by a team at The Institute of Cancer Research, London. It’s investigating whether adding the drug capecitabine to chemotherapy after surgery can improve treatment for women with early breast cancer.

In 2017, results from the trial were published and are available in PubMed. But how easy is it to find this paper by searching the database?

Searching with keywords

Using keywords and filters is one of the most straightforward approaches for interrogating PubMed and can be very successful, requiring only basic knowledge about what you’re looking for.

As an example, with our case study typing in ‘breast cancer clinical trial multi-center phase III’ gives 87 records, with the TACT2 paper in fourth place. Putting the drug name ‘capecitabine’ alongside the clinical trial filter, brings up 1,548 results, with the TACT2 paper at number 22. And searching by sponsor using ‘The Institute of Cancer Research AND breast cancer’ brings up 2,352 records, with the relevant publication listed at 44.

PubMed clinical search

Using PubMed clinical queries is a way to refine searches to only a subset of the database, limiting results to clinical citations. It also splits them into three categories – clinical studies, systematic reviews or medical genetics.

For example, using the keywords ‘breast cancer’ identifies 98,564 clinical studies, 9,015 systematic reviews and 76,661 medical genetics results.

With clinicians in mind, there’s an additional option for refining results down into five clinical study categories: etiology, therapy, diagnosis, prognosis or clinical prediction guide – as well as two filters for scope: narrow (more precise, relevant citations but higher volume) or broad (more citations, but many may not be as relevant).

Keeping ‘therapy’ and ‘broad’ and using the search term of ‘breast cancer multi-center clinical trial phase III’ gives 90 records, with the TACT2 paper in fifth place.

Search by trial registry number

Since the International Committee of Medical Journal Editors (ICMJE)  stipulated that all clinical trials must be publicly registered as a requirement for publication of their results, this has increased the likelihood of trial data being available in PubMed.

So a powerful way of finding results from a specific trial is to use its clinical registry information, with those acceptable for PubMed to include in its journal citations listed in MEDLINE Databank Sources.

Within PubMed, registry numbers are included in the secondary source ID (SI) field and searchable using the [si] tag.

Using TACT2’s ClinicalTrials.gov (NCT00301925) identifier instantly brings up the relevant paper. However, its ISRCTN number (ISRCTN68068041) drew a blank, suggesting that not all registry data is included on a publication record.

Tapping into clinical trial data on PubMed

There’s an abundance of information on clinical trials on PubMed, with the main challenge working out the best search strategy to find it. 

A quick and easy approach to find the results from a specific clinical trial is to search using its clinical registry information. But for searches that are wider in scope, careful consideration is needed to choose the most relevant keywords and filters. Searching via PubMed Clinical queries rather than the comprehensive PubMed database is useful, enabling more focused searches and providing useful extra filters to further refine results.

 

Industry teams use BrackenData Analytics to analyze data on their competitors' clinical trials. Signup to get a personal demonstration of how to use BrackenData Analytics to leverage PubMed data:

 

 
Name *
Name
 
 

Comment