Assessing Indices of Transparency and Reproducibility in Animal Models of Opioid Addiction

Abstract

Psychology’s reproducibility crisis has led to a reckoning of research practices in many fields. Moreover, several preclinical fields have come under scrutiny due to poor rates of treatment translation from animals to humans. This is true of the preclinical addiction field (Venniro et al., 2020). Ensuing investigation revealed that many of the same research design aspects that undermine reproducibility also threaten translation potential (Fergusson et al., 2019). We examined indices of transparency and reproducibility in animal models of opioid addiction from 2019 to 2023. In doing so, we aimed to understand whether efforts to improve reproducibility are relevant to this field. We measured the prevalence of transparency measures such as preregistration, registered reports, open data, and open code as well as compliance to the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines. We also measured reported rates of bias minimisation practices, sample size calculations and multiple corrections adjustments. Lastly, we estimated the accuracy of test statistic reporting. Appraising 247 articles revealed poor uptake of transparency measures, the ARRIVE guidelines, bias minimisation practices and sample size calculations. Adjustments for multiple comparisons was alone in being implemented in most articles (76.5%). Lastly, half of articles contained non-decision errors and 11% contained decision errors. We discuss the implications of these results and potential explanations as well as solutions for their improvement. Our study is the first of its kind in this field and demonstrates that attempts to improve reproducibility and, in turn, translation, are needed in the animal models of opioid addiction field.