Statistical Reporting Quality in Dental Research: A Comparative Analysis by Journal Impact and Study Design

Background: Determine the statistical report quality of articles published in dental journals according to the level of impact of the journal and the study design. Methods: A random sample of original articles published in four journals of General Dentistry indexed in Scopus was evaluated, two high-impact (quartile 1) and two medium-impact (quartile 3). The study design was recorded as observational, experimental, or other. The quality of each report was assessed through seven statistical error indicators. Two statisticians independently evaluated the articles and then contrasted their results; in the case of discrepancies in the ratings, they reached an agreement to establish a consensual criterion. Chi-square, Fisher’s exact, Mann-Whitney U , and Kruskal-Wallis tests were applied with post hoc Dunn-Bonferroni correction. Results: The presence of at least one statistical error was high (85.6%), with 123 (83.7%) and 44 (91.7%) articles in high- and medium-impact journals, respectively. Two statistical errors were more common (32.3%). We found a relationship between the presence of some errors ( P = 0.009) and the number of statistical errors ( P = 0.011) with the study design. Conclusions: There is a low-quality statistical report of the evaluated dental articles, as errors are very common; among these errors, inaccurate descriptions of the P values and a lack of justification for the sample size were the most common. The observational studies were also of higher quality than the experimental studies.