Reporting quality in preclinical animal experimental research in 2009 and 2018: A nationwide systematic investigation

Abstract

Lack of translation and irreproducibility challenge preclinical animal research. Insufficient reporting methodologies to safeguard study quality is part of the reason. This nationwide study investigates the reporting prevalence of these methodologies and scrutinizes the reported information’s level of detail. Publications were from two time periods to convey any reporting progress and had at least one author affiliated to a Danish University. We retrieved all relevant animal experimental studies using a predefined research protocol and a systematic search. A random sampling of 250 studies from 2009 and 2018 led to 500 publications in total. Reporting of measures known to impact study results estimates were assessed. Part I discloses a simplified two-level scoring “yes/no” to identify the presence of reporting. Part II demonstrates an additional three-level scoring to analyze the reported information’s level of detail. Overall reporting prevalence is low, although minor improvements are noted. Reporting of randomization increased from 24.0% in 2009 to 40.8% in 2018, blinded experiment conduct from 2.4% to 4.4%, blinded outcome assessment from 23.6% to 38.0%, and sample size calculation from 3.2% to 14.0%. Poor reporting of details is striking with reporting of the random allocation method to groups being only 1.2% in 2009 and 6.0% in 2018. Reporting of sample size calculation method was 2.4% in 2009 and 7.6% in 2018. Only conflict-of-interest statements reporting increased from 37.6% in 2009 to 90.4%. Measures safeguarding study quality are poorly reported in publications affiliated with Danish research institutions. Only a modest improvement was noted during the period 2009–2018, and the lack of details urgently prompts institutional strategies to accelerate this. We suggest thorough teaching in designing, conducting and reporting animal studies. Education in systematic review methodology should be implemented in this training and will increase motivation and behavior working towards quality improvements in science.

Not All Porcine Intestinal Segments Are Equal in Terms of Breaking Force, but None Were Associated to Allometric Parameters

“…For the post hoc tests, we employed the Dunn-Bonferroni test to correct for alpha inflation in order to limit the potential of false-positive results from a statistical point of view. It is puzzling that conducting and reporting a sample size calculation at all is surprisingly uncommon; a recent systematic review found it to have only increased from 5.2% to 7.6% in 2018 [109], despite the methodological drawbacks of not doing so [110]. To have performed a sample size calculation does not, however, support the assumptions that we made, at least some level of transferability from dogs to pigs, but at least provides some support for the reliability of the results, given its assumptions were true, for which there are no data at present.…”

Advocating for Generalizability: Accepting Inherent Variability in Translation of Animal Research Outcomes

“…Similarly, before the implementation of requirements to report study design elements (e.g., randomization, blinding, sample size estimation) across three major journals in 2011, less than 33% of studies reported randomization, less than 47% reported blinding, and less than 6% reported sample size estimation. The number of articles reporting these factors increased significantly after journal interventions were implemented (21,22). Yet despite advances in required reporting by journals, fewer than 50% of published studies include rigor criteria such as randomization, blinding, and sample size estimation (23).…”

Reporting of 3Rs Approaches in Preclinical Animal Experimental Studies—A Nationwide Study

“…The literature search, random sampling, and data retrieval were described previously [19]. Briefly, the literature search was conducted in Medline (via PubMed) and Embase for all citations that referred to in vivo studies conducted on non-human vertebrates with one or more authors affiliated with at least one of five Danish universities of interest.…”

“…These 250 publications were selected based on the random sampling allocation sequence. The PRISMA flow diagram is available in Kousholt et al [19].…”