Publishing Your Education Work in the Journal of Graduate Medical Education

Abstract

of the Editorial Board There's nothing to writing. All you do is sit down at a typewriter and open a vein. Walter Wellesley ''Red'' Smith T he Journal of Graduate Medical Education (JGME) has lofty aims: provide a resource for graduate medical educators to learn about novel approaches as well as publish their own work and improve the quality of information guiding decisions about graduate medical education training. As one step to reach these goals, the editors have created general categories for submissions to JGME. These categories are expected to evolve over time as the needs of educators change. In this editorial, we outline the general categories, quality expectations for submissions, and common minefields to avoid. As always, we welcome your suggestions and comments (jgme@acgme.org). Original Research Currently, most submissions to the JGME fall into the ''Original Research'' category. Like other research, educational research needs a plausible intervention, outcomes that match the goals of the intervention, and methods that are likely to ascertain the outcomes fairly. 1 ''Original Research'' submissions must include a structured abstract with background/rationale, research question or objective, methods (including design, setting, participants, intervention, comparison group if applicable, outcomes, and analysis), results, major limitations, and conclusions. A well thought out research question or objective is critical and will lead to the most relevant methods. A focused question aligned with appropriate outcomes and analysis will also reduce the likelihood of finding spurious associations. The background should include what is known at present about the topic as well as what is unknown that the research study attempts to answer. Methods should be appropriate to examine the intervention or question and will conform to usual quality standards. 2-7 At times, education research does not allow ideal experimental methods (eg, blinding of subjects [trainees] to the research question, blinding of observers/data collectors to the research question, randomization of subjects, uncontaminated controls, previously validated outcome instruments, stable medical practice environment). It is essential that the necessary choice of less than optimal methods be discussed openly. Similarly, qualitative research should adhere to usual methodologic standards, with any deviations explained. 8,9 Objective outcomes, rather than subjective (eg, self-reported) measures, are more likely to persuade readers that the results are true. 10 In addition, outcomes relevant to the research question are essential. For example, if the intervention is designed to improve results on the in-training examination, then scores on the in-training examination are an appropriate outcome measure. However, if the intervention is geared toward increasing resident-patient discussions about smoking cessation, a knowledge test is not a suitable measure, but patient report of such discussions, with a valid instrument, would be relevant, as would amounts of tobacco use reported by patients. Other issues that merit attention include the timing of postintervention assessments, intraobserver reliability, and choice of comparison or control groups. In studies in which the research question tests whether an educational experience can improve resident performance, the performance optimally should be measured at some distance from the intervention, not immediately postintervention. In projects in which there is more than one data collector/ observer, comparisons among observers, preferably blinded to the study question, are important. Given the difficulties of randomizing trainees or educational experiences, historical controls (ie, past cohorts) frequently are chosen for comparison. In these situations, an analysis of the potential differences between the current and past cohorts' characteristics and environments should be included. Discussion of the limitations as well as the strengths of the project are key to the reader's full understanding of the work and do not detract from the likelihood of the paper being published. Similarly, statistical significance does not always indicate that the results should change our current practice. After considering the limitations of the study, the conclusions should answer the question ''Should these results change our educational practices and, if so, in what situations?'' In view of the diverse readership of this journal, jargon and terms specific to 1 discipline should be avoided. To be useful to most readers, extra care is needed to ensure that all aspects of the project are clear to most readers.