Table 3.
Relation between source of funding and methodological quality of research
| Study | Criteria used to assess methodological quality of research | Results |
|---|---|---|
| Cho et al14 | 22 item validated scoring system | Study design in drug company sponsored clinical trials better than in research where no stated sponsorship (P=0.04) |
| Clifford et al15 | 5 item validated scoring system (Jadad) plus component (individual items on Jadad scale and adequacy of concealment) approach | No difference by funding source for adequacy of allocation concealment (P=0.377); no difference by funding source for overall/composite score on Jadad scale (P=0.143) |
| Davidson16 | Sample size, blinding | For all trials higher rate of blinding for ones with industry sponsorship (67.5% v 41.8%, P=0.01); for trials investigating medications no difference in blinding (P=0.46); for sample size no difference between clinical trials supported by drug companies and those with other sources of funding or where funding not stated |
| Djulbegovic et al19 | 5 item validated scoring system (Jadad) | No difference in quality scores between randomised controlled trials funded solely by industry (mean 3.3 (SD 1.4); median: 3.5) and trials supported by public sources (mean 2 (SD 0.96); median: 2) (P=0.308) |
| Djulbegovic et al20 | 5 item validated scoring system (Jadad) | Randomised controlled trials funded solely or partly by industry had trend towards higher quality scores (mean 2.94 (SD 1.3); median: 3) than trials supported by government or other non-profit organisations (mean 2.4 (SD 0.8); median: 2) (P=0.06) |
| Jadad et al25 | 7 point validated scoring system (Guyatt and Oxman) | 6/6 industry funded systematic reviews and meta-analyses had serious flaws versus 34/44 non-industry funded reviews |
| Kjaergard et al28 | 5 point validated scale including: concealment of allocation, generation of allocation sequence, double blinding, dropouts/withdrawals, sample size | Clinical trials funded by either drug or device industry had higher quality than trials with no external funding (P<0.001); quality of publicly funded trials same as trials funded by drug or device industry (P=0.68) |
| Knox et al29 | 9 item scale developed for this study, including clinical design, generalisable data sources, statistical tests of significance performed on appropriate outcomes, statement regarding perspective, description of costs of the main included resources, description of time horizon, description of source of total costs differences, discussion of limitations, comparisons with other published studies | Drug company sponsored pharmacoeconomic analyses less likely to formally report on study generalisability, but were more likely to provide information on the key components of the methods section than were non-profit sponsored analyses |
| Liebeskind et al31 | 100 point scale addressing 5 aspects of trial design and reporting: randomisation, outcome, inclusion/exclusion criteria, description of therapeutic regimen, statistical analysis | Clinical trials with corporate support had better quality than trials with non-profit support (mean 73.1 (95% Cl 3.9) v 53.4 (9.8); P<0.0001) |
| Mandelkern32 | Presence or absence of placebo control | 5/16 industry funded clinical trials had placebo controls compared with 3/16 non-industry funded trials |
| Massie et al33 | Not stated | Higher proportion of industry funded clinical trials were adequately controlled and designed than were trials with other sources of funding (71% v 33%, P<0.01) |
| Neumann et al34 | Adherence to recommended protocols for cost effectiveness studies (adequate description of alternatives, study perspective clearly stated, discounted both costs and QALYs if needed, incremental analyses performed correctly) plus quality as judged by readers (scale of 1 to 7) | No difference between industry and non-industry funded studies on any measure: adequate description of alternatives P=0.30; study perspective clearly stated P=0.98; discounted both costs and QALYs P=0.65; incremental analyses performed correctly P=0.73; quality as judged by readers P=0.49 |
| Rochon et al35 | Modified version of Chalmers score including 14 items: control appearance and/or regimen, randomisation, blinding, patients blinded, observers blinded to treatment and results, previous estimate of numbers, testing compliance, results of randomisation on pretreatment variables and inclusion in analysis, major end points, post-beta estimate, confidence limits, statistical analyses, withdrawals after randomisation, side effects discussion | No difference in quality score between industry only funded clinical trials and those funded by government or foundations (mean 36.9% (SD 17.6%) v 37.1% (17.8%), P=0.271) |
QALY=quality adjusted life year.