Evidence Live 2015: Analysis of US phase 3 ClinicalTrials.gov records completed before January 1st, 2011 (n=5051; time frame: 2002 to 2014)

Abstract accepted for a 15 minute oral presentation (10 + 5) at Evidence Live 2015.

---

Introduction (background)

There has been a growing concern about the selective reporting of clinical trial results (i.e., publication bias). The problem of unpublished and misreported clinical trials was the main reason behind the creation of campaigns such as AllTrials and The BMJ Open Data Campaign. The birth of clinical trial registers (e.g., ClinicalTrials.gov, EudraCT, ISCRTN, among others) more than one decade ago allowed for the first time the possibility to identify unpublished clinical trials in public databases.

Aims

The aim of this study was to answer the following question: are US clinical trial data sufficiently shared?

Methods

Results disclosure of US phase 3 completed ClinicalTrials.gov registries (n=5051; registration dates: January 1st, 2002 to January 1st, 2014) was carried out by data mining (i.e., expert search queries) in journal article databases: PubMed, Embase, Google Scholar, and EBSCO Discovery Service. Other variables such as locations, time difference between the registration date and start date, and results disclosure at ClinicalTrials.gov were also analyzed.

Results, raw data and comments related to these analysis are available in the following BMJ rapid responses

Re: The US requirement to deposit trial data within a year is unworkable. http://www.bmj.com/content/347/bmj.f6449/rr/690626

Response to J. Castellani (PhRMA): An ounce of data (i.e., 64740 data values). http://www.bmj.com/content/347/bmj.f1881/rr/762606

Zombie statistics strikes again
http://www.bmj.com/content/347/bmj.f1880/rr/763200

Supplementary explanation of research methods is available via figshare:

http://dx.doi.org/10.6084/m9.figshare.1121675

Results

Over half ClinicalTrials.gov study registers - completed before January 1st, 2011 (n=2957) - are unpublished (50.8%).

Over half of these study records were retrospectively registered.

Over half of these registries have not disclosed their results at ClinicalTrials.gov.

Conclusions

US phase 3 clinical trials data are not sufficiently shared.