Clinical Trials Registries For Systematic Reviews – An Alternative Source For Unpublished Data

Abstract

Objectives: A common problem in systematic reviews are incomplete data extraction forms resulting in problems attempting evidence synthesis; we rarely have all the data for the endpoints of interest for all studies, and parameters that inform meta-analysis or connect networks are missing. Increased transparency in clinical trial reporting means this problem is slowly disappearing. From January 2015 the European Medicines Agency (EMA) will publish clinical study reports submitted with marketing-authorisation applications for human medicines. MethOds: We identified several data sources outside the primary publication. Standard data sources for systematic reviews of interventions include peer-reviewed publications, conference abstracts and clinical trial registries. Clinical study protocols are often published but are not identifiable through searches in online databases, therefore, to find these, systematic reviewers must visit the journal website. Manufacturer submissions to health regulators are also increasingly made available; these give detailed trial descriptions and results presented are more likely to be comprehensive. Results: In a recent example in Hepatitis-C we utilised several additional data sources in our evidence synthesis. Clinical trial protocols were used to identify definitions of endpoints included and to fulfil aspects of the critical appraisal. Fibrosis stage is an accepted treatment effect modifier in Hepatitis-C; our review therefore collected subgroup data for this. However, this was not readily available in peer-reviewed publications; we thus obtained data from EMA submission documents and UK and German reimbursement submissions. Other examples include a 2013 COPD systematic review which retrieved mortality data from the FDA website for three studies reporting cardiovascular-related death and for one study reporting overall death. cOnclusiOns: Systematic reviewers should be aware of additional data sources that are publically available. Whilst peer-reviewed data is preferential, incorporation of this grey literature into an evidence synthesis could lead to a more informed overview of clinical efficacy and thereby healthcare decision making.Objectives: Assess the potential relationships between Biomarkers of Potential Harm (BOPH), specifically WBC, Apo lipoprotein, C-reactive protein, HDL, LDL, total cholesterol, and biomarkers of cigarette smoke exposure (BOE), specifically serum cotinine, creatinine adjusted urinary total NNAL and 1-hydroxypyrene (1-OHP), using National Health and Nutrition Examination Survey (NHANES) data from 2007 till 2012. Secondary objective was to assess the relationship between BOPH and smoking status (past, current or never), and cigarette per day (CPD) use in current smokers. MethOds: Data were obtained from NHANES 2007 to 2012. The study sample included 17,293 respondents age 21 years and above who had answered questions on cigarette smoking and had complete laboratory values for their biomarkers measurement. The population was categorized as current (CS), past...

Obstacles to the reuse of study metadata in ClinicalTrials.gov

“…Clinical-trial registries such as ClinicalTrials.gov have become a crucial source of information for systematic reviews and other metanalyses. Recent studies recommend that systematic reviews include a search of clinical trial registries to identify relevant trials that are ongoing or unpublished 2 – 5 . Baudard et al .…”

Challenges of identifying unpublished data from clinical trials: Getting the best out of clinical trials registers and other novel sources

“…In their study of 443 completed Chinese trials registered in trials registers, Liu et al suggest that 35.2% of trials in the WHO ICTRP, 36.5% of trials in http://ClinicalTrials.gov, 36.3% in the Chinese Clinical Trial Registry (ChiCTR), 26% of trials in the ISRCTN Registry, and 55.6% in the Australian New Zealand Clinical Trials Registry provide results in subsequent journal publications. Trials register records may also contain data for outcomes or subgroups that cannot be found in published papers …”

“…Trials register records may also contain data for outcomes or subgroups that cannot be found in published papers. 15 It should, of course, be noted that registers still have a high non-compliance rate in terms of the publication of results. Anderson et al note that only 38.3% of trials that were likely to be subject to the FDAAA provisions to report results in ClinicalTrials.gov had reported them up to September 2013.…”

A technical review of three clinical trials register resources indicates where improvements to the search interfaces are needed

“…Clinical trials registers form an important part of the search for studies in systematic reviews of intervention effectiveness 1‐8 . Trials registers can be searched to identify newly registered, on‐going, or recently completed but unreported studies (i.e., unpublished studies), and to review registered study protocols, for example, to identify adaptations to the design of a trial or changes to the outcomes measured over‐time 1,3,4,7‐26 . They can also provide an alternative source of study data and study results for some studies.…”