Sponsored drug trials show more favourable outcomes
EDITOR - BMJ editor Richard Smith expresses concern about the influence of conflict of interest on publications1. Our recent finding within a Cochrane review2 supports the concern that drug company involvement in clinical trials affects outcome. The searches undertaken for the systematic review of clozapine vs. 'typical' antipsychotic drugs for those with schizophrenia identified 29 different randomised studies involving 2490 participants. Sixteen of the trials reported some kind of connection with the manufacturer of the compound. These studies were assumed to be 'sponsored studies'. Peto odds ratios (OR) and 95% confidence intervals (CI) were calculated for the primary outcomes of relapse, clinical improvement, and leaving the study early. The odds of relapsing were statistically significantly in favour of clozapine, in comparison to the older generation antipsychotics in the sponsored trials (OR 0.5 CI 0.3-0.7, 13 trials, 980 people). Non-sponsored studies reported statistically equivocal findings (OR 0.4 CI 0.1-1.4, 10 trials, 783 people). Both sponsored and non-sponsored studies suggested that clozapine mediates a clinically important improvement in those with schizophrenia when compared to the older drugs but again the former were more positive than trials not clearly supported by industry (random effects OR 0.4 CI 0.2-0.7, 9 trials, 1126 people; random effects OR 0.3 CI 0.1-0.7, 8 trials, 743 people, respectively). Sponsored studies also reported that significantly less people taking clozapine left the study early when compared to drugs such as chlorpromazine and haloperidol (OR 0.5 CI 0.4-0.7, 14 trials, 1245 people). Non-sponsored trials (12 trials, 950 people) did not show this degree of effect (OR 0.6 CI 0.3-1.2).
The observation that drug industry sponsorship causes more favourable outcomes is of concern. Our finding underpins how important transparency regarding drug company sponsorship is in assessing trial outcomes. Those undertaking drug treatment meta-analyses should investigate for significant sponsorship-bias by use of sensitivity analysis. It is also of concern that licensing authority decisions are mainly based on trials performed by industry and therefore do not have information from independent researchers. Of course the problem of conflicts of interest may be even greater outside of industry where trialists can have very considerable investment in their own particular brand of community care package or psychological intervention.
Kristian Wahlbeck, Editor
Cochrane Schizophrenia Group, Department of Psychiatry, University of Helsinki, PB 320, FIN-00029 HUCH, Finland
Rapid Response:
Sponsored drug trials show more favourable outcomes
EDITOR - BMJ editor Richard Smith expresses concern about the influence of conflict of interest on publications1. Our recent finding within a Cochrane review2 supports the concern that drug company involvement in clinical trials affects outcome. The searches undertaken for the systematic review of clozapine vs. 'typical' antipsychotic drugs for those with schizophrenia identified 29 different randomised studies involving 2490 participants. Sixteen of the trials reported some kind of connection with the manufacturer of the compound. These studies were assumed to be 'sponsored studies'. Peto odds ratios (OR) and 95% confidence intervals (CI) were calculated for the primary outcomes of relapse, clinical improvement, and leaving the study early. The odds of relapsing were statistically significantly in favour of clozapine, in comparison to the older generation antipsychotics in the sponsored trials (OR 0.5 CI 0.3-0.7, 13 trials, 980 people). Non-sponsored studies reported statistically equivocal findings (OR 0.4 CI 0.1-1.4, 10 trials, 783 people). Both sponsored and non-sponsored studies suggested that clozapine mediates a clinically important improvement in those with schizophrenia when compared to the older drugs but again the former were more positive than trials not clearly supported by industry (random effects OR 0.4 CI 0.2-0.7, 9 trials, 1126 people; random effects OR 0.3 CI 0.1-0.7, 8 trials, 743 people, respectively). Sponsored studies also reported that significantly less people taking clozapine left the study early when compared to drugs such as chlorpromazine and haloperidol (OR 0.5 CI 0.4-0.7, 14 trials, 1245 people). Non-sponsored trials (12 trials, 950 people) did not show this degree of effect (OR 0.6 CI 0.3-1.2).
The observation that drug industry sponsorship causes more favourable outcomes is of concern. Our finding underpins how important transparency regarding drug company sponsorship is in assessing trial outcomes. Those undertaking drug treatment meta-analyses should investigate for significant sponsorship-bias by use of sensitivity analysis. It is also of concern that licensing authority decisions are mainly based on trials performed by industry and therefore do not have information from independent researchers. Of course the problem of conflicts of interest may be even greater outside of industry where trialists can have very considerable investment in their own particular brand of community care package or psychological intervention.
Kristian Wahlbeck, Editor
Cochrane Schizophrenia Group, Department of Psychiatry, University of Helsinki, PB 320, FIN-00029 HUCH, Finland
Clive Adams, Co-ordinator
Cochrane Schizophrenia Group, Summertown Pavilion, Oxford OX2 7LG
1. Smith R. Beyond conflict of interest. Transparency is the key. BMJ 1998;317:291-2. (1 August.)
2. Wahlbeck K, Cheine M, Essali MA. Clozapine vs. 'typical' neuroleptic medication for schizophrenia (accepted). In: The Cochrane Library, Issue 4, 1998. Oxford: Update Software.
Competing interests: No competing interests