Epidemiologists can’t help but be dismayed about all the ways there are to lose study subjects. We can never avoid that gap between the reference population we identify and the final study groups we assemble, despite our most resourceful efforts. As authors ourselves, the editors are sympathetic with the impulse to report response proportions in the best possible light. But matters are made worse when we obscure the gap. In this issue, Olson et al1 suggest a format for more fully reporting the steps taken to recruit a study population, and the response proportions at each step.
Full disclosure of the components of response and nonresponse—including the reasons for excluding otherwise eligible participants—is beneficial in the big picture. Not all types of loss are created equal. Self-selection and low response can lead to bias or loss of generalizability, whereas loss of subjects at certain other stages can be less damaging. More complete information provides a better basis for assessing study results, and also for comparing results across studies.
We encourage our authors to provide full and candid details of response proportions. Olson et al1 propose a table format for expressing these details. Flow diagrams are also an option. If this additional information lengthens a manuscript beyond the journal’s available page space, the details can be published in the electronic version of the journal—which raises another issue.
In the long run, the electronic publication of scientific journals may turn out to be the most accessible and permanent repository of our work. Although most of us prefer to leaf rather than to scroll through our journals, librarians and other information specialists widely regard the electronic version of our published work as the authoritative version. Our journal’s page space remains limited, but our electronic space does not. We encourage authors to take advantage of this resource. Authors who wish to provide supplementary materials for the electronic version (including perhaps the fine details of response proportions) should include these as appropriately labeled appendices in their submitted manuscripts.
Dale P. Sandler
1. Olson SH, Voigt LF, Begg CB, Weiss NS. Reporting participation in case-control studies. Epidemiology 2002; 13: 123–126.