Our most important finding is that only 7.7% (95% confidence interval: 4.1%-13.1%) of the 155 responding journals indicated they had a dual-use review policy. Of the potential explanatory variables we examined, the best predictor of having a dual-use policy was membership in the NPG. All 9 NPG journals had a dual-use policy, whereas only 3 of the remaining 146 journals did. This suggests that the odds of having a dual-use policy depended more on a publisher's rules than on an individual journal's rules. Not surprisingly, among the journals outside the NPG, those with previous experience reviewing dual-use research were more likely to have a dual-use policy than those without experience. When considered individually, higher journal impact factors were associated with greater odds of having a dual-use policy, which is consistent with recent claims that impact factor was the only variable among several (such as publisher and country) that was significantly associated with having a misconduct policy.7
In our study, however, the significance of journal impact factor as a predictor of having a dual-use policy disappeared after adjusting for publishing group membership.
The percentage of journals in our sample that had a dual-use policy (7.7%) was much lower than the percentage reported by van Aken and Hunger (25%).6
One likely explanation for the difference in results is that different populations were studied; they focused on top journals in biomedicine, which have high impact factors, while we did not limit our sample in this way. In their sample of 28 life science journals, van Aken and Hunger included 20 of the top journals in virology and biomedicine, which had an average impact factor of 16.7, as well as 5 Russian journals and 3 Chinese journals with no ratings for impact factor.6
In contrast, the average impact factor in our sample was 3.8. While it is important to understand how the top biomedical journals are responding to dual-use dilemmas in publication, it is also essential to have an understanding of how low- and mid-level journals are dealing with this issue, since these journals also review research that may pose a threat to biosecurity. Our research suggests that journals with high impact factors (ie, those in the top tier) are more likely to have dual-use policies than journals with low impact factors, though much of this may be explained by publishing group policies.
Nine journal editors described how they would respond if they encountered a paper that raised dual-use concerns, but our analysis did not count these journals as positive respondents because they did not have a formal (written) policy, and we decided to include only journals with formal policies. An interesting follow-up study would ask editors about their informal policies pertaining to dual-use research.
None of the journal policies that the editors shared with us included criteria for determining whether research should be published, not published, or edited for potentially harmful content. The policies only described a procedure for applying an additional level of review for manuscripts identified as raising dual-use concerns. A follow-up study might ask editors about the criteria they would use in decision making, or whether they would make decisions on a case-by-case basis.
Several of the editors had interesting responses to our survey. One editor said that despite the fact that her journal had published several papers that had dual-use implications, she and her colleagues had never heard of the term “dual-use” and had never considered implementing a policy to address such manuscripts. Another editor immediately changed the journal's policy to read, “All authors will conform to the National Science Advisory Board for Biosecurity (NSABB) guidelines for Dual-use Life Sciences Research.” Several editors stated that they would develop dual-use policies in the near future. Surprisingly, a few editors had negative reactions to the idea of a dual-use policy; in fact, one stated, “We do not employ two different tiers of peer review because of potential use of biotechnology by terrorists or criminals. This could be considered potentially as discrimination, bias or a political issue.” Another common reaction among editors was to state that it was unlikely their journals would ever receive a manuscript that would have dual-use implications and consequently there was no need to develop a dual-use policy.
These reactions from the editors suggest several reasons that so few biomedical journals have developed dual-use policies. First, many editors may simply not be aware of the importance of biosecurity issues in scientific publishing. Second, some editors may be aware of biosecurity issues but do not feel that these issues are relevant to their journals because they have not reviewed dual-use research and they do not believe they are likely to do so in the near future. Third, some editors may understand the need for a dual-use review policy but have simply not taken the time to develop one. Fourth, some editors may view a dual-use policy as an inappropriate restriction on freedom of inquiry and the dissemination of scientific knowledge. Additional research could be conducted to explore journal editors' ideas and attitudes concerning dual-use review and potential barriers to policy development.
A limitation of our study is the low response rate (39%). It is worth noting, however, that our response rate was not significantly lower than the response rates of other e-mail surveys, which tend to have lower response rates than telephone surveys.8
One possible explanation for our low response rate is that we conducted the study during the summer, when many academic institutions in the U.S. and Europe are on break. Some of the editors indicated that they would respond when they returned from vacation, but they never did. Another possible explanation for the low response rate is that some of our e-mails may have been rejected as spam or junk e-mail. Finally, editors may have been concerned about the potential legal liability implications of responding to our survey, although we stated in the e-mail that we would protect the confidentiality of the responses. We do not believe that any of the reasons for the low response rate undermined the validity of our data or biased our overall results.