Does declaration of competing interests affect readers' perceptions? A randomised trial

BMJ 2002; 325 doi: http://dx.doi.org/10.1136/bmj.325.7377.1391 (Published 14 December 2002)
Cite this as: BMJ 2002;325:1391.1
  1. Samena Chaudhry, Clegg scholara,
  2. Sara Schroter, research fellow (sschroter{at}bmj.com)a,
  3. Richard Smith, editora,
  4. Julie Morris, head of medical statisticsb
  1. a BMJ Editorial, BMA House, London WC1H 9JR
  2. b Medical Statistics Department, Education and Research Centre, Wythenshawe Hospital, Wythenshawe, Manchester M23 9LT
  1. Correspondence to: S Schroter
  • Accepted 4 November 2002

Conflict of interest has been defined as a set of conditions in which professional judgment concerning a primary interest (such as patient welfare or the validity of research) can be influenced by a secondary interest (such as financial gain).1 Despite increasing evidence that conflict of interest influences authors' conclusions,24 we found no research into the effect on readers' perceptions of research. We studied whether a declaration of financial competing interest influences readers' perceptions of the interest, importance, relevance, validity, and believability of a study.

Participants, methods, and results

We randomly selected 300 BMJ readers from the BMA's membership database, which contains individual reader characteristics. All readers were sent a short report indicating that the impact of pain from herpes zoster on patients' daily functioning may be substantial.5 Readers in group 1 were sent a version of this paper with different named authors to the original and with a declaration that they were employees of a fictitious company and potentially held stock options in the company. Readers in group 2 were sent a version of the same paper with the same named authors as in group 1 but with a statement that these authors were from an ambulatory care centre and had no competing interest. Readers were asked to rate the study in terms of interest, importance, relevance, validity, and believability on 5 point Likert scales (1=extremely uninteresting to 5=extremely interesting). We estimated that 86 readers were needed in each group to achieve a power of 90% to detect a difference in scores between the groups of 0.5 units on the 5 point scale. We used a simple two sample t test with the conventional 5% significance level and assumed a common standard deviation of 1.0.

Three questionnaires were returned by family members as the addressee was considered unfit to participate. Six were sent to the wrong address. We excluded one more as the recipient was an institution. In total, 170/290 (59%) questionnaires were returned (86 in group 1, 84 in group 2). Non-responders were significantly younger (mean age 40.7 years (SD 13.9, range 19-93 years)) than responders (44.7 (SD 15.5, 20-82) years in group 1 and 44.8 (16.9, 19-86) years in group 2) (t=−2.3, df=276.2; P=0.022). There was no significant difference in the proportion of men between non-responders (65%) and responders (58% in group 1; 64% in group 2) (χ2=0.44, df=1; P=0.51).

Independent t tests showed that readers in group 1 thought the study was significantly less interesting, important, relevant, valid, and believable than readers in group 2 (table). Within each group, scores for each item were poorly correlated to age (r0.22), and there were no significant differences in scores by sex.

View this table:

Score* distributions for readers' perceptions of papers with and without declaration of competing interests. Figures are percentage of readers

Comment

Declaration of competing interests may have a significant effect on readers' perceptions of the scientific credibility of published medical research. BMJ readers reported that data showing the impact of pain from herpes were less interesting, important, relevant, valid and believable when the authors were employees of a fictitious pharmaceutical company compared with an ambulatory care centre.

Our study has several weaknesses. The response rate was low and responders were older than non-responders. However, within each group there was no relation between age or sex and scores. Readers knew they were being studied and thus responses might not reflect the way they usually evaluate a manuscript. We used only one manuscript with one competing interest statement. It is possible that the unfamiliar name of the fictitious company might have influenced readers' responses. Future research should explore different types of competing interest statements with different manuscripts and samples of readers.

Acknowledgments

We thank the authors of the original manuscript for allowing us to use it and the readers for taking part in our study.

Contributors: RS initiated and suggested design for the study. SC and SS conducted the study. SS performed the data analysis, and JM provided statistical advice. All authors interpreted the results and wrote the manuscript. SS is guarantor for the study.

Footnotes

  • Editorial by Smith

  • Funding No additional funding.

  • Competing interests RS is the editor of the BMJ and is responsible for devising its policy on competing interests. He is paid a fixed salary and will not be affected financially by the success or failure of the policy on competing interests. This study was peer reviewed in the normal way, except that RS played no part in the process.

References

THIS WEEK'S POLL