We included in our search the MEDLINE classification for the following 8 study designs: randomized controlled trials (RCT), clinical trials, systematic reviews, case reports, cohort, cross-sectional, case-control, and comparative. A hierarchical filtering process was used in Ovid to ensure all study types were mutually exclusive and that each article was only indexed for one study type.

The search strategies were developed by the second author (MS).

We reviewed a random 5% sample of the records to assess classification accuracy for specialty and study type for all 3 bibliometric searches. In the neurology and neurology subspecialty searches, we also assessed the accuracy of articles being correctly classified as pediatric.

For comparison of neurology and other specialties as well as for comparison of study type in pediatric journals, we constructed models for the number of publications by year, journal, and subspecialty and for the number of publications by year, journal, and study type, respectively. In both cases, Poisson models were initially considered; however, due to the presence of overdispersion these models were replaced by quasi-Poisson models. The starting point were models including main effects (for year, journal, and subspecialty in the first comparison and for year, journal, and study type in the second comparison) as well as all two-way interactions. When the evidence for an interaction was weak (p > 0.10), that interaction was removed, and the reduced model was fitted. Likelihood ratio test was used to compare full and reduced models. Based on the final models, estimated marginal means by year, journal, and subspecialty and by year, journal, and study type, respectively were computed.

Next, contrasts between each subspecialty and neurology within journals (comparison of neurology and other specialties) were estimated. The contrasts were then averaged over years, and displayed as ratios, together with 95% confidence intervals adjusted for multiple testing using the multivariate t-distribution method (9).

The need for ethical approval was waived by the Children’s Hospital of Eastern Ontario Research Institute Ethics Board, as this was a purely bibliometric study.

The JCR 2017 report revealed that the five general pediatric journals with the 5 highest impact factors were: The Journal of the American Medical Association (JAMA) Pediatrics (10.8), Pediatrics (5.5), Archives of Disease in Childhood Fetal and Neonatal Edition (4.0), Journal of Pediatrics (3.7), and Pediatric Research (3.1) (Table 1). Note that JAMA Pediatrics continued Archives of Pediatrics and Adolescent Medicine, the name change occurred during our observation period. We searched both, combining results and will refer to this set as JAMA Pediatrics. The 4 selected specialties used for comparison with neurology were immunology, endocrinology, gastroenterology, and respirology.

Our bibliometric search yielded 4,239 publications, of which 1,501 were in neurology; see Table 1 for a summary of distribution of these 1,501 neurology publications in the top 5 general pediatric journals. Table 1 also summarizes the type of study design of neurology articles published in the top 5 general pediatric journals. The number of publications in the other 4 specialties was: immunology (715), endocrinology (394), gastroenterology (467), and respirology (1,162) (Table 2). The distribution of publications for each study type by specialty is provided in Table 2.

Figure 1 describes the estimated ratio of mean number of publications per year of subspecialty versus neurology articles. The length of the bars indicate 95% confidence intervals for the estimates. The position of the estimate (the dot in the middle of the bars) should be interpreted in relation to “1.” If the value is below 1, then in a given journal, the number of publications in given a specialty was lower than in the reference specialty (Neurology). If the estimate is above 1, then the number of publications in a given specialty was higher than in Neurology. In all journals considered, there was a greater number of publications in neurology than other specialties with the exception of the number of respiratory-focused publications in Archives of Disease in Childhood Fetal and Neonatal Edition, JAMA Pediatrics and Pediatric Research and the number of immunology-focused publications in JAMA Pediatrics being comparable to neurology.

According to the JCR 2017 report the five general neurology journals with the 5 highest impact factors were: Lancet Neurology (27.1), JAMA Neurology (11.5), Brain (10.8), Annals of Neurology (10.3), and Neurology (8.1).

When our bibliometric search was performed without a child limit, 5,278 records were identified. After restricting to pediatric articles our search produced 643 records, and thus the proportion of all articles that were indexed as pediatric was 12.2%. Table 3 outlines the distribution of pediatric neurology focused articles in the top five general neurology journals, as well as the study types.

According to the JCR 2017 report the five neurology subspecialty journals with the 5 highest impact factors were: Movement Disorders (8.3), Stroke (6.2), Multiple Sclerosis Journal (5.3), Epilepsia (5.1), and Journal of Stroke (4.8).

When our bibliometric search was performed without a child limit, 5,513 records were identified. After restricting to pediatric articles our search produced 685 records, and thus the proportion of all articles that were indexed as pediatric was 12.4%. Table 4 outlines the distribution of pediatric neurology focused articles in the top five subspecialty neurology journals, as well as the study types.

Of the 4,239 pediatric journal records, 212 (5%) were reviewed. Accuracy was 94.8% by specialty and 97.6% by study design. Of the 643 neurology journal records, 32 (5%) were reviewed. Accuracy was 100% by specialty and 93.8% by study design, and 93.9% of records were accurately classified as pediatric. Of the 685 neurology subspecialty journal records, 34 (5%) were reviewed. Accuracy was 100% by specialty and 100% by study design, and 100% of records were accurately classified as pediatric.

Inherent to any bibliometric approach, the major limitation of our analysis was that we did not examine individual article records, except those in the random sample we examined to verify the accuracy of indexing. Rather, we relied on indexing in MEDLINE for classifications. Nonetheless, as we previously mentioned, the accuracy of the 5% sample we manually verified was very high and thus we can be confident that the results achieved through bibliometric methods are a true reflection of the specific body of literature.

Our results provide some guidance to authors regarding where they may wish to consider submitting their pediatric neurology research. Prospective authors will be able to incorporate the results into their decision making, based on the intended audience, the perceived impact of the research and the study design utilized.