This study analyzed seven prominent forensic science journals, selected based on their recognition in a recent publication [19]. The journals included Journal of Forensic Sciences (JFS; Impact Factor [IF]: 1.5), Forensic Science International (FSI; IF: 2.2), Journal of Forensic Odonto-Stomatology (JOFS; IF: 0.947), Journal of Forensic and Legal Medicine (J Forensic Leg Med; IF: 1.2), International Journal of Legal Medicine (IJLM; IF: 2.2), Legal Medicine (Leg Med; IF: 1.3), and Egyptian Journal of Forensic Sciences (Egypt J Forensic Sci; IF: 1.3). The analysis focused on the number of forensic odontology publications within these journals over two distinct five-year intervals: 2003–2007 and 2019–2023.

The choice of the two five-year periods, 2003–2007 and 2019–2023, for analyzing FO publications is based on several important factors. Firstly, we aim to analyze changes in self-citation patterns within FO over an extended period. The selected timeframes represent distinct phases in forensic science publishing, allowing for a comparative analysis of trends before and after notable developments in publishing practices, including the proliferation of digital databases, the implementation of open-access models, and shifts in citation behaviors. The 2003–2007 interval offers insights into FO literature during the early adoption of digital publishing, while the 2019–2023 interval captures contemporary trends influenced by the extensive use of online platforms, advanced citation tools, and enhanced international collaboration.

Secondly, these timeframes were selected to complement a recent study [19] that analyzed trends in FO publications from 2000 to 2015. The study identified a notable increase in FO research output as the field advanced, particularly in more recent years. The limited volume of FO research during the 2003–2007 period is especially relevant, highlighting the discipline’s developmental trajectory. Comparing this earlier phase with the 2019–2023 period provides valuable insights into shifts in self-citation rates and publication trends. The 2003–2007 interval serves as a baseline for FO publication activity, while the 2019–2023 timeframe represents the contemporary state of the field. Variations in publication volume between these periods reflect the growth of FO and its increasing impact on the broader forensic science literature.

The selection of FO publications was carried out in three stages. In the first stage, two investigators, both qualified forensic odontologists with ten years of experience (first author, NA and last author, SBB), independently accessed all journal issues from January to December from 2003 to 2007 and 2019 to 2023 through institutional subscriptions. They manually searched for original research articles, reviews, and case reports related to FO, screening the articles based on their titles and abstracts. Articles outside these categories were excluded from the analysis. In the second stage, one week after the initial screening, the selected publications underwent further manual review to confirm their relevance to FO. In cases of uncertainty, the two investigators communicated to reach a consensus on whether to include or exclude the publication. During the third stage, a more rigorous selection was applied to assess whether publications engaged in self-citation practices, using an approach similar to Livas et al. (2021) [14].

The following information was extracted from each included article for analysis: the name of the journal, article title, number of authors, names of first and last authors, author rank, type of study, the topic of the study, total number of citations, number of self-citations, self-citation rate (SCR), gender, region and country of the most self-citing author, and lastly, the institutional information.

To facilitate data analysis, regions were categorized into seven groups: Asia, Africa, North America, South America, Oceania, and “others” (a combination of continents). Articles were also classified by topic, including dental age estimation, dental sex estimation, both sex and age estimation, human dental identification, bite mark analysis, forensic odontology practices, and an “Other” category, which covered areas such as dental trauma, neglect, artificial intelligence, malpractice, odontometrics, and professional liability. Additionally, articles were grouped by the number of authors: 1–3, 4–5, 6–7, and 8 or more. The author rank refers to the ranking of the authors on the number of self-citations. It is categorized as first, last, or first/ last. The first/ last refers to articles where an equal number of self-citations were observed for the first and last authors. Information regarding gender, country, and institutional status was recorded based on self-citations. In each article, the self-citations of both the first and last authors were counted to calculate the SCR, which is the percentage of an author’s self-citations relative to the total citations in the reference list. Authors with the highest SCR had their data entered into the analysis. If both authors had the same SCR, the first author’s information was used for data entry and analysis. Both FO investigators (NA & SBB) carried out the data screening. Any discrepancies were resolved through discussion until they reached a consensus.

In cases where a scientific paper indicated that the first two authors contributed equally, only the first listed author was considered in the evaluation. The collected data were then entered into a Microsoft Excel spreadsheet (Microsoft Corporation, Redmond, VA, USA) for further analysis. One month after the initial registration in the Excel spreadsheet, articles that included self-citations were reviewed to verify the accuracy of the recorded information.

Descriptive statistical analyses were performed to explore potential associations between self-citations and various predictor variables. In addition, the selected publications were classified into two categories based on the publication periods: 2003–2007 and 2019–2023. A comparative analysis across these periods was conducted using Pearson’s chi-square test for categorical variables and the Mann-Whitney U-test for numerical data.

A negative binomial regression model was used to examine the associations between self-citations and various predictors. The categorical predictors included journal, study type, article topic, number of authors, the rank of the most self-citing author, origin, and gender, while the numerical predictor was the total number of citations. The author’s rank, origin, and gender were collected and analyzed only for articles with at least one self-citation. Consequently, the number of articles included in each univariate analysis varied, leading to different sample sizes across analyses. The distribution of the outcome variable, self-citations, was assessed using the Kolmogorov-Smirnov test, which indicated that it did not follow a Poisson distribution and showed overdispersion, justifying the use of a negative binomial regression model. Significant predictors identified in the initial univariate analysis were then entered into a multivariable negative binomial model. The significance of each predictor was assessed using likelihood ratio tests.

The box plot visualization was created in RStudio 2024.04.2 + 764 utilizing the ggplot2 package. All statistical analyses were performed using SPSS Statistics 26.0 (IBM, USA), with a significance level set at 5% (p < 0.05).