Impact of adding data from clinical study reports to a systematic review of adverse events in cancer patients

Session: 

Oral session: Investigating bias (2)

Date: 

Monday 17 September 2018 - 14:50 to 15:00

Location: 

Carrick 2

All authors in correct order:

Dietz KC1, Simmonds M1, Phillips RS1, Wright K1, Dale H1, Stewart LA1
1 Centre for Reviews and Dissemination, University of York, UK
Presenting author and contact person

Presenting author:

Kristina Charlotte Dietz

Contact person:

Kristina Charlotte Dietz
Abstract text
Background: Journal articles reporting clinical trial results can over-estimate benefits, and under-report harms. Inclusion of data from non-journal sources, such as Clinical Study Reports (CSRs), submitted to regulatory agencies as part of the approval process may help alleviate this problem. Inclusion of regulatory sources generally results in more precise estimates of harms (narrower confidence intervals due to additional data), and can alter conclusions.

Objectives: The objectives of this systematic review and meta-analysis are three-fold:
1) the results will provide updated risk estimates of arterial and venous thromboembolic events associated with the sole or additive use of cetuximab or panitumumab relative to standard care in adult cancer patients with solid tumours based on ‘best available data’ (i.e. most complete data across journal- and regulatory sources);
2) the impact of including regulatory data sources will be described in quantitative and qualitative terms, considering metrics such as available sample size, precision of risk estimates, statistical significance, direction of effects, and clinical significance of effect sizes;
3) the feasibility of incorporating CSRs into evidence syntheses will be discussed in terms of time, resource requirements, and practicality. This discussion is hoped to inform general method and guidance development for the wider, more routine inclusion of regulatory sources in evidence syntheses.

Methods: We will calculate risk ratios with 95% confidence intervals by treatment arm (intervention, control) for each prespecified outcome. We will compare results based exclusively on journal sources with those using ‘best available data’ including from CSRs. We will compare the direction of effect, heterogeneity and precision of risk estimates, and the statistical and clinical conclusions that can be drawn from each and considered alongside the feasibility of accessing and using CSRs. The value of utilising CSRs in Cochrane Reviews will be discussed.

Sponsorship: The project is funded by a National Institute of Health Research Systematic Review Training Fellowship (NIHR, reference: 3484) and will be submitted as dissertation project for an MSc in Applied Health Research completed as part of the Training Fellowship. The protocol has been prospectively registered at https://tinyurl.com/yccd9zwn

Relevance to patients and consumers: 

Systematic reviews should provide patients and doctors with the best available evidence to support informed decisions about the potential benefits and risks of a clinical intervention. Current methods may not be optimal, because journal articles may selectively report clinical trial outcome data, particularly harms. This project explores the impact and practicality of adding clinical study reports (CSRs) as a source of data used in reviews. CSRs, which are submitted to regulatory bodies as part of the licencing process, may provide more complete data to weigh-up benefits and harms, but use would require extensions to current review methods.