Resumen de Long-term impact of an editorial intervention to improve paper transparency and reproducibility

Marta Vilaró Pacheco

• This PhD thesis is concerned with evaluating the long-term effects of interventions in the editorial peer review process of a biomedical journal. The growing need to increase the value of research and avoid waste motivates this work. We focus on evaluating the long-term impact on the number of citations (NC) of articles introduced in trials, as this will allow us to evaluate the effects of adding either a methodological expert (statistician) or reporting guidelines (RG) during the peer review process.

  In 2004 and 2009, two randomized trials were published in Medicina Clínica (Elsevier, Barcelona), in which these interventions were added during the editorial process. They showed a slightly positive e¿ect on paper quality, as assessed by the Manuscript Quality Assessment Instrument (MQAI) of Goodman et al. In this work, we explore the effect on NC by collecting from Web of Science (WoS) the NC of each article that had been randomized in the two previous studies.

  This thesis presents different ways of analyzing count data involving time-to-event, first from a perspective of count data models and then from a perspective of recurrent events methods.

  Results show that including a methodological reviewer (for example, a senior statistician) who is dedicated to looking for missing RG items increases the NC by 40% (95% CI: 1% to 94%) when considering that citations are independent within articles, like count models. When considering the within-article correlation of citations using the frailty gamma model, we find that including a methodological reviewer during the peer-review process increases the possibility of receiving a citation by 41% (95% CI: 0% to 102%). Independently of the method used to analyze NC, and with different assumptions, the consistency of those results gives robustness to the findings.

  Therefore, as measured by the NC, with this work we show that randomized studies of using interventions in the peer review process to improve scientific impact are feasible. Our proof of concept study opens the door for the development of confirmatory trials.