PUBLICATION BIAS AND PASSIVE SMOKE RESEARCH: INTERVIEWS WITH INVESTIGATORS

Anastasia L Misakian and Lisa A Bero
Institute for Health Policy Studies, University of California, San Francisco, 1388 Sutter St, San Francisco, CA 94109, USA

Objective: To determine whether editor and researcher bias prevents the publication of data concluding that passive smoke is not harmful to health (negative data).

Design: Semistructured telephone interviews of principal investigators (PIs) for original research projects on the health effects of passive smoke (excluding biochemical, mechanistic, genetic, and policy studies). Research was identified by contacting 12 federal and state agencies, 9 tobacco industry-affiliated organizations, and 67 private organizations. Data were collected on the extent, direction, and publication status of any unpublished data; reasons for not publishing; and publications resulting from the project.

Results: Sixty-five PIs (response rate = 83%) were interviewed about 84 research projects. Twenty-seven (32%) projects were published, 24 (29%) were partially published, and 33 (39%) were unpublished. The most common reasons for having unpublished data were ongoing data collection and/or analysis (38/57, 67%); ongoing manuscript preparation (9/57, 16%); and priority (passive smoke was a minor component of a larger project; 8/57, 14%). Two unpublished studies (1 with positive and 1 with mixed findings) were rejected by a journal; both have been resubmitted. The proportion of negative findings was similar in the published (28%) and unpublished (28%) data. There was no significant difference in the time to publication for negative and positive studies.

Conclusions: There is no publication bias against negative studies on the health effects of passive smoke. These findings do not support the tobacco industry's claim that risk assessments are incomplete due to an underrepresentation of negative data in the published literature.

Return to Session Information