Pathways to Publication in Pediatric Hospital Medicine Educational Research

We performed this retrospective cohort study using abstract submission data from 4 years of PHM conferences. The annual PHM conference is alternately hosted by the American Academy of Pediatrics, Society of Hospital Medicine, and Academic Pediatric Association (APA).

We obtained data on educational research abstracts submitted to the 2014–2017 PHM conferences, including abstract title, primary investigator (PI) name, and presentation type (oral, poster, rejected). We excluded abstracts for which presentation was declined by the authors. Twenty-four months after the 2017 conference, we queried PubMed, MedEdPORTAL and Google Scholar for evidence of publication using the abstract title and PI’s name. We defined published abstracts as articles with a similar title and an author list including the abstract’s PI. We recorded the journal publication date and impact factor for the year before publication. If multiple publications were identified, we included the earliest dated publication. For journals published every 2 months, we recorded the date of publication as falling between the 2 months. Impact factors were found within a common impact factor database¹¹  or on a journal’s Web site. Any uncertainty regarding whether an abstract reached publication was resolved by discussion among study authors.

We described the number and proportion of published educational research abstracts. Odds of publication were compared by presentation type by using logistic regression models. We calculated the median time to publication (months) and median journal impact factor overall for each presentation type, determining statistical significance using the Kruskal-Wallis test. Using abstracts published within 24 months after the conference (n = 50; 89% of publications), we generated Kaplan-Meier survival curves for each presentation type to assess differences in time between publications. Finally, we identified the most common publishing journals for each presentation type. All analyses were performed by using SAS version 9.4 (SAS Institute, Inc, Cary, NC); P values ≤.05 were considered statistically significant. This research was deemed non–human subjects research by the study site’s institutional review board.

Of 173 educational research abstracts submitted to the 2014–2017 PHM conferences, 24 (14%) were selected for oral presentation, 116 (67%) were selected for poster presentation, and 33 (19%) were rejected (Table 1).

Fifty-six (32%) educational research abstracts were published; most had been selected for either oral or poster presentation (n = 49, 88%). Oral abstracts had threefold greater odds of publication than poster abstracts (odds ratio [OR] 3.2; 95% confidence interval [CI] 1.3–8.0; P = .011; Table 1) and fivefold greater odds of publication than rejected abstracts (OR 5.2; 95% CI 1.6–16.7; P = .003). There was no difference in the odds of publication between poster and rejected abstracts.

The median time to publication was 13.3 months (interquartile range [IQR] 4.5–21) with no significant difference between presentation types (Table 1). However, the cumulative number of publications at each month was statistically significantly different between presentation types as demonstrated by Kaplan-Meier survival curves (P = .005; Fig 1).

The median journal impact factor for all publications was 2.2 (IQR 1.0–2.7; Table 1). The median journal impact factor for published oral abstracts and published poster abstracts was >2.5 times and 2 times higher, respectively, than published rejected abstracts (P = .002; Table 1). Academic Pediatrics was the most common publishing journal for presented abstracts (Table 2). Hospital Pediatrics was the publishing journal for one oral presentation, seven poster presentations, and three rejected presentations.

One-third of educational research abstracts submitted to the 2014–2017 PHM conferences were subsequently published. We observed significantly higher odds of publication for oral abstracts compared to poster and rejected abstracts; however, the odds of publication for poster abstracts were similar to those of rejected abstracts. The median time to publication did not differ between presentation types, although cumulative differences at each month were significantly different. Oral and poster abstracts were published in journals with higher impact factors than rejected abstracts. These results suggest that abstract acceptance and presentation type may be early indicators of publication success.

Our results align with previous literature in PHM and general medical education research. Herrmann et al⁴  identified an overall publication rate of 36% within 30 months of the 2014 PHM conference.⁴  We show that the subset of medical education research from the same conference over a 4-year period had only a slightly lower publication rate (32%) when queried at a minimum of 24 months postconference. We also describe a shorter median time to publication (13 vs 17 months). This suggests that PHM conference abstract reviewers are screening educational research with similar rigor as other types of research, and selected abstracts are of similar quality if measured by the likelihood of publication. In previous studies, authors describe medical education research publication rates of 33% to 44%^6,8–10 ; however, these studies were heterogeneous in conference type and target population and excluded rejected abstracts. Finally, our results align with previous research in which authors report that the odds of publication differ on the basis of presentation type, with abstracts selected for oral presentation having higher odds of publication compared to poster and rejected abstracts.^9,6,12 

Several factors may influence the observed relationship between presentation type and publication outcomes. The opportunity for critical feedback after conference presentation may contribute additional considerations for manuscript preparation. However, it is unclear if the quality or amount of feedback differs by presentation type. Abstract selection for oral presentation may signal a study’s overall relevance to the field, which may further motivate authors to seek publication or submit to journals with a higher impact factor. Other potential influencers not assessed in this study include local institutional research infrastructures, years of research experience, and receipt of formal research training via fellowship or advanced degree programs. These confounders may help explain why the abstract selection process does not fully predict publication success of educational scholarship. More than 40% of oral abstracts never went on to publication, and just >20% of rejected abstracts did. This variability highlights the need for additional instruction in scholarship development and presentation.

PHM is a young subspecialty, and although there is significant growth in the number of fellowship opportunities,¹³  institutions likely have varying mechanisms for educational research mentorship. Fortunately, there are external opportunities to develop these skills, including the Educational Scholars Program, a 3-year national faculty development program sponsored by the APA,¹⁴  and the Advancing Pediatric Educator Excellence Teaching Program, a 2-year program cosponsored by the American Academy of Pediatrics Section on Hospital Medicine and the APA, which emphasizes teaching skills and promotes collaboration for future scholarship.¹⁵  Additionally, the Association of Pediatric Program Directors Longitudinal Educational Assessment Research Network extends collaborative research opportunities to program directors.¹⁶  A scholarly approach to clinical education is paramount to building an accessible educational research repository and advancing the field.

This study has several limitations. First, we conducted our literature search only 24 months after the 2017 PHM conference. This time frame likely underestimates the number of publications from that year, although 89% of publications from all years were published within 24 months. Second, our search strategy may not have identified all existing publications, especially if the article title was dissimilar to the abstract title. Study authors discussed the few abstracts with discordant titles with consensus building. Third, we were unable to obtain reviewer criteria for every PHM conference; thus, we cannot ensure consistency in reviewer guidelines or a blinded review process, which may influence the overall abstract selection process. Finally, the type of educational research (eg, curriculum development) may influence publication odds. We did not assess for differences, but this could be explored in the future.

Abstract submission to the national PHM conference is a reasonable first step to disseminating educational scholarship. In further research, authors could explore the characteristics around abstract selection and presentation that most strongly influence publication success.

We thank Evan Weber, MD, chief resident of the Children’s Mercy Hospital Pediatric Residency Program for her help with data collection.