Response to Kelly Near and Ann Duesing's letter: "Informatics Training for Public Health Practitioners".

Kelly Near and Ann Duesing have performed a great service by attempting to replicate the authors’ randomized controlled trial on the effects of training on information-seeking behavior among public health professionals [1]. They have served the profession even further by reporting in their letter the challenges that they encountered in replicating our study and for reporting their interesting yet divergent results [2]. We had recommended further replication of our study, because we asked important questions that trended toward the affirmative for the two hypotheses tested but could not be confirmed at the level of statistical significance. While Near and Duesing were able to use our practical advice in replicating our study, their letter suggests that differences in their population in southwestern Virginia and unexpected problems with administering a consistent incentive system for participants probably explain the differences between the two studies. We found that public health health care practitioners viewed PubMed as useful for answering their native questions. As reported elsewhere [3,4], our trainees anonymously highly valued their PubMed training in both immediate and follow-up evaluations two months later. We were intrigued that Near and Duesing did not make the same observation, which might have been due to the involvement of different populations. This additional discrepancy between the two studies also deserves further investigation with public health practitioner populations elsewhere. Evidence-based practice guidelines [5] indicate that our study currently offers the best evidence to support existing library or informatics training programs. Clearly, readers of the Journal of the Medical Library Association need to continue to offer training programs, unless later studies convincingly suggest otherwise. Our study suggests support for the two hypotheses. To test these hypotheses successfully, a future study will need to recruit enough participants to reach statistical significance, while providing all participants with equal incentives to submit their questions. Our study suggests that without incentives to submit questions, participants will not take the time away from their many work-related demands to fully record their questions. Near and Duesing’s study could not achieve equal status for incentives because of the decentralized local health system in southwestern Virginia. In contrast, a system capable of providing a consistent incentive and administrative system could ensure the existence of conditions conducive for a randomized controlled trial. The National Association of County and City Health Officials profiles of local health departments suggest that the states of Arkansas, Delaware, Florida, Mississippi, South Carolina, and Vermont have strong state-centric authority over local health departments [6]. These states possibly offer the potential for replication because these states could likely ensure consistent administration of incentives, although we encourage others to test our hypotheses in other contexts. Unified, large, urban-area public health departments represent other plausible environments for rigorously replicating our study successfully. We will gladly share our detailed study protocols and any pragmatic advice, as we did for Near and Duesing, with any interested colleagues to ensure that any replications conducted elsewhere proceed smoothly.