A convenience sample of 100 randomized controlled trials (RCTs) was identified by hand-searching recent issues of five peer-reviewed, high impact factor general medical journals (Annals of Internal Medicine, British Medical Journal, Journal of the American Medical Association, The Lancet, The New England Journal of Medicine). Issues published between January 1999 and October 2000 were searched until 20 RCTs/journal were identified. To be eligible for inclusion, the RCT needed to be published as a full report. Interventions were restricted to pharmaceuticals; medical devices, surgical procedures and methods of medical management (e.g., lifestyle counseling) were excluded. No attempts were made to limit the selection to any particular RCT design, number of treatment arms, comparator (e.g., placebo, active control, alternate dosing, herbal therapy), study population or disease category.
Relevant data, including funding source(s) and primary outcome, were abstracted from each eligible RCT. Reporting quality was assessed using both a composite (overall score on the Jadad scale [19
]) and a component (individual items on Jadad scale and adequacy of allocation concealment [20
]) approach. The Jadad scale consists of a total of five items; two items relate to blinding, two items relate to randomization and one item assesses the description of withdrawals/drop-outs. When using the Jadad scale to score the quality of a trial report, each of the five items receives a "yes" or a "no," resulting in an overall/composite quality score that can range from 0 to 5; higher scores reflect better methodological quality [19
Allocation concealment was rated as adequate, inadequate or unclear in the manner proposed by Schulz et al [20
]. Allocation concealment refers to the process that prevents foreknowledge of treatment assignment and thus, shields those who enroll participants from being influenced by this knowledge. For example, a trial was rated as having "adequate" concealment if allocations were performed using central randomization; numbered/coded bottles/containers; serially numbered, opaque, sealed envelopes or if the formulations were prepared by a pharmacy. Allocation was classified as "inadequate" if assignments were made on an alternating basis or via reference to case record number or date of birth. Trials that received an "unclear" rating would have failed to provide sufficient information regarding the allocation process on which to base our decision. In all cases, reporting quality was evaluated by two independent, experienced reviewers (DM, TC). No formal training was conducted prior to evaluating the RCTs using either of the quality assessment scales since both raters have extensive experience using these methods. Moreover, since any disagreements in quality ratings were resolved by consensus, we did not undertake assessment of inter-rater reliability.
In order to examine the relationship between trial outcome, funding source and reporting quality, SPSS-PC software was used to conduct statistical analyses in the form of Fisher's exact test or ANOVAs, as appropriate. The odds, and corresponding 95% confidence intervals, of unclear allocation concealment, by funding source and trial outcome, are also presented.
Trials were classified according to their funding source(s) in a manner similar to that used by Rochon [17
], permitting comparison of trials across four levels of funding: entirely industry, entirely not-for-profit, mixed and not reported. A trial was classified as having "mixed" funding if it received support from at least one industry source and at least one not-for-profit source. Because this study was restricted to RCTs that examined pharmaceuticals, industry funding is synonymous with pharmaceutical company funding.
The primary outcome was defined as the one stated as such by the authors or, if there was no such statement, the one that was most clinically relevant (i.e., mortality over morbidity). If one outcome was not more clinically relevant than the others, the outcome contributing the most patients was used. On the basis of statistical interpretation of results, rather than reliance on authors' interpretations presented in discussion/conclusion sections, the outcome of trials was classified as favoured new treatment, favoured conventional treatment, neutral (i.e., non-significant) or unclear.