Determining the degree of dissemination bias for health-related quality of life outcomes in antidepressant trials




Poster session 1


Sunday 16 September 2018 - 12:30 to 14:00

All authors in correct order:

Sharma T1, Rasmussen K1, Paludan-Müller A1, Gøtzsche P1
1 Nordic Cochrane Centre, Denmark
Presenting author and contact person

Presenting author:

Tarang Sharma

Contact person:

Abstract text
Dissemination bias distorts the true effectiveness of interventions, leading to unreliable results. Though health-related quality of life outcomes have been used in antidepressant trials, their results have rarely been published.

To determine the level of selective reporting, including publication bias, of the health-related quality of life outcomes of EQ-5D and SF-36 in antidepressant trials.

We used clinical study reports (CSRs) of five antidepressant drugs received from the European and UK regulators to identify all double-blind randomised, placebo-controlled trials for any indication that included EQ-5D or SF-36 as outcomes in their protocols. We contacted Eli Lilly, GlaxoSmithKline (GSK) and Pfizer for missing data and the publications for the included trials. We compared outcome data from the CSRs to what was available online and in the publications.

Our inclusion criteria allowed us to include 15 trials on 4717 people (where two trials had both the instruments). For five trials it was unclear which quality of life outcome was used. There was complete information for SF-36 for only three of eight trials from CSRs and for only two trials from publications and there was complete information for EQ-5D for four of nine trials from CSRs and from none of the publications. No complete information was available for SF-36 or for EQ-5D for any of the 15 trials through the data online.

If a systematic review has quality of life as an outcome, even using CSRs could be unreliable for getting a complete set of results. Getting access to the raw data including the case report forms may be the only reliable source of data.

Patient or healthcare consumer involvement:
No patients were involved in this review but we aim to involve them during the dissemination of our results.


Relevance to patients and consumers: 

Patients make decisions on whether to use a particular medicine based on the evidence of its benefits versus its harms. When the underlying evidence for decision-making is biased, it can distort the apparent effectiveness of interventions, giving rise to misleading conclusions. Determining the true impact of medications on the quality of life of patients requires working with clinical study reports (CSRs) such that more reliable information is available to patients.