In order to identify SRs, one reviewer (CRB) extracted data regarding population, interventions, comparisons and outcomes (inclusion criteria) and evaluated the methodological quality of the included trials.
Currently, brief protocol summaries, including aims, primary outcomes, inclusion and exclusion criteria, duration of the trial, and planned intervention, of all clinical trials are available on the internet.
Study strengths are the 30-year follow-up on objective outcomes, inclusion of several objectively measured risk factors for IHD and mortality and exclusion of workers with pre-existing cardiovascular disease at baseline.
In regard to motor function outcomes, inclusion of fish intake as a confounder with a realistic reliability ratio of 43% doubled the mercury effect compared with results unadjusted for fish intake.
During the start of this evaluation, several problems related to the classification of outcomes, inclusion of post-hoc analyses and follow-up studies of RCTs in the study sample, and assessment of the role of the funding source were encountered.
The interventions under study and trial design characteristics, such as definition of outcomes, inclusion criteria, risk factors and statistical considerations can cause substantial differences when pooling data, and may constitute the major causes of heterogeneity found in the current meta-analysis.
There are several strengths of our meta-analysis including comprehensive search strategy, multiple clinically important outcomes, inclusion of non-English trials, duplicate abstraction, a priori subgroup and sensitivity analyses, obtaining missing data from authors, and adherence to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [ 39].
In assessing study performance, four areas will be considered: (1) subjects: how recruited to the study; (2) intervention: adequate description of the intervention used for the treatment of patients; (3) data analysis: details of how the data were collected and analysed; (4) outcomes: inclusion of individual subject data or a statistical summary, details of missing results, drop outs.
It discusses sources of variability, including the goal of the CAD system, learning methodology, study population, design, outcome measures, inclusion of radiologists, and study quality.
