Using Clinical Study Reports versus published articles in a Cochrane Review update

ID: 

92

Session: 

Poster session 5

Date: 

Thursday 27 October 2016 - 10:30 to 11:00

Location: 

All authors in correct order:

Musini V1, Lawrence K1, Wright J1
1 Hypertension Review Group; University of British Columbia, Canada
Presenting author and contact person

Presenting author:

James Wright

Contact person:

Abstract text
Background: Cochrane Reviews typically include only journal-published randomized controlled trials. These publications often provide very little information on harms. The trial's corresponding clinical study reports (CSRs) are seldom included. This results in potential misrepresentation of efficacy and harm data that renders such meta-analyses potentially unreliable.

Objectives: To document advantages and disadvantages of including information from CSRs of trials meeting the inclusion criteria as opposed to including only published journal articles in a Cochrane Systematic Review.

Methods: Electronic databases are typically searched to identify primary studies that meet the inclusion criteria. However, for market approval purposes or for ongoing safety evaluation CSRs are required by regulatory authorities like the European Medicines Agency (EMA). For the 2016 update of a Cochrane Systematic Review, a formal request for all relevant CSRs was made to EMA under the Access to Documents Policy.

Results: The advantages of including CSRs include:
1. comprehensive information is available on study methods;
2. availability of numerical data with standard deviation instead of graphs in published articles;
3. availability of data of all secondary outcomes as stated in the protocol;
4. opportunity for accurate assessment of risk of bias of each included study;
5. provision of detailed information for all-cause mortality, non-fatal serious adverse events and specific adverse events as opposed to very limited information in the published article.
Disadvantages include:
1. CSRs are often more than a thousand pages in length;
2. it is difficult to identify the CSR as each included study had three different identifiers;
3. time consuming as requested CSRs were obtained in batches based on date of request and order in the queue;
4. not all CSRs were available for all included studies (five published trials were not registered in clinical trials.gov).

Conclusions: Including CSRs of all included studies, when available, leads to a more comprehensive analysis and interpretation of benefits and harms of a drug therapy.