The State of Data Sharing in Plastic Surgery: An Analysis of Journal Practices and Author Adherence

Abstract

Background:

Data sharing is crucial for transparency and reproducibility in research. We aim to evaluate data sharing practices in plastic surgery research, focusing on data sharing statements (DSSs), identifying key trends, and assessing follow-through.

Methods:

We conducted a systematic review of the top 7 plastic surgery journals, selecting original research articles published from 2018 to 2023. Data extraction was performed in a masked duplicate manner, capturing DSS presence, funding sources, study design, and publication year. Trends in DSS inclusion over time, study designs, and journals were analyzed. A thematic analysis was conducted on DSS content. Corresponding authors of studies that stated data were available upon request were contacted to assess follow-through.

Results:

Our review included 727 articles, with only 1.51% (11 out of 727) including DSS. DSS prevalence varied, with the highest in Aesthetic Plastic Surgery (4.42%, 5 out of 113) and the lowest in Plastic and Reconstructive Surgery (0.43%, 1 out of 235). Clinical trials were the most likely to include DSS (1.92%, 9 out of 469), followed by cohort studies (1.16%, 2 out of 173). Private funding and certain publishers were negatively associated with DSS inclusion, whereas journal impact factors showed a positive correlation. Responses from authors were not received, indicating a gap between stated intentions and actual practices.

Conclusions:

Data sharing practices in plastic surgery journals are inconsistent. Although DSS inclusion is encouraged, many studies still lack DSS, and the quality varies. Clearer mandates and enforcement are needed for effective data sharing.

Takeaways

Question: Our study examines the prevalence of data sharing statements (DSSs) in the top 7 plastic surgery journals.

Findings: Our analysis of 727 articles found that only 1.51% included a DSS, highlighting discrepancies between stated intent and practice. Springer and Elsevier publications demonstrated higher DSS inclusion rates, indicating potential publisher influence, whereas barriers included patient privacy concerns and ambiguous sharing conditions.

Meaning: Despite open science advocacy, only 1.51% of articles include a DSS, highlighting barriers to transparency and reproducibility. Standardized policies and improved practices are needed to enhance plastic surgery research.

INTRODUCTION

In the realm of research, transparent and accessible data sharing is crucial for scientific progress. The White House Office of Science and Technology Policy declared 2023 the “Year of Open Science,” emphasizing the importance of making publicly funded research data accessible and promoting equitable and secure scientific practices.¹ A data sharing statement (DSS) in research publications specifies how raw data will be made available, enhancing transparency and the reliability of clinical research, and highlighting the need for reproducible research in this growing field.²

Many medical journals have prioritized achieving reproducible and high-quality research to further advance their specialties. As of July 1, 2018, the International Committee of Medical Journal Editors³ declared that manuscripts submitted to its journals must include a statement on data sharing; however, data sharing itself is not mandated.⁴ This illustrates an effort toward publicly accessible data, therefore facilitating study validation and enabling further analyses. Additionally, the National Academy of Medicine reports that sharing clinical research enhances the understanding of clinical trial outcomes by encouraging independent researchers to conduct new analyses and form new conclusions.⁵ Despite reports and efforts made to promote the use of data sharing, a gap between DSS and actual data availability still exists.⁶

Although data sharing is essential for the advancement of research knowledge, several barriers hinder transparent reporting, such as the privacy of patient identity. Plastic and reconstructive surgery have integrated photography into their routine practice, which enhances a provider’s ability to track subtle changes over time.^7,8 This implementation increases the risk of litigation and financial liability if personal health information were to be exploited.^7,8 Additionally, data providers may object to researchers using their data for secondary analysis due to a lack of incentive to share their data.⁹ Furthermore, some authors may initially include a DSS, but upon subsequent review, the data are often not actually made available. In 2021, a study found that merely 2% of meta-analyses actually made their data accessible.² These discrepancies undermine the credibility of the study findings, casting doubt on the transparency and reproducibility of the research.

The aim of our systematic review was to evaluate the prevalence of DSS in the top journals of plastic surgery. Additionally, we assessed the authors’ willingness to share their data in publications that include such statements. This study hypothesized that the prevalence of DSSs in plastic surgery literature is low and that DSS inclusion is associated with specific study characteristics, including funding source, journal impact factor, and study design. We used inferential statistical methods to evaluate these factors as potential predictors of DSS inclusion. We hope to identify factors influencing DSS presence in plastic surgery research and provide recommendations to enhance transparency.

METHODS

Study Design, Ethics, Reporting, and Reproducibility

A systematic review was performed to search for, select, and extract data from primary studies. We then conducted a cross-sectional analysis. The Oklahoma State University Center for Health Sciences institutional review board (study number: 2024115) approved this study. Our protocol (all versions), scripts, data, and other artifacts have been uploaded to Open Science Framework¹⁰ to facilitate open science practices, transparency, and reproducibility.

Search Strategy

On June 11, 2024, the top 7 journals were chosen from Google Scholar Metrics (Plastic & Reconstructive Surgery subcategory) based on their 2023 journal h5-index. Journals that fell outside the traditional scope of plastic surgery were excluded. To prevent subspecialty overrepresentation, only the highest impact journal per subspecialty was included. A MEDLINE (PubMed) search using journal ISSNs identified clinical studies, ensuring relevance. To capture recent developments, the search was limited to articles published between January 1, 2018, and December 31, 2023.

Training

Each team member underwent training on the study objectives, inclusion/exclusion criteria, general methodology, data extraction procedures and variables, and the use of an online systematic review platform, Rayyan (https://www.rayyan.ai/). We utilized Rayyan for blinded article screening. Rayyan supports efficient review processes by offering features that enhance inter-rater reliability, such as duplication detection and inclusion/exclusion categorization. Although Rayyan itself does not undergo traditional clinical validation, its reliability has been documented in systematic review literature as enhancing the screening stage without compromising accuracy.¹¹ Afterwards, 2 authors (K.K. and J.M.) conducted a calibration exercise, extracting data from 5 articles to ensure consistency and reliability. These extractions were reviewed to assess agreement and maintain accuracy in data collection.

Screening/Eligibility Criteria

Included articles were original research covering plastic surgery procedures or outcomes, including clinical trials, cohort studies, cross-sectional analysis, case-control studies, case series, and survey studies. Articles had to be in one of the top 7 journals, provide new primary data (qualitative or quantitative), and be published during the specified time period. Excluded were commentaries, reader responses, review articles, editorials, erratum/corrigendum, single-person case reports, systematic reviews, nonplastic surgery procedures, and studies without novel primary data. Two authors (K.K. and J.M.) independently screened articles and resolved disagreements with a third author (A.H.).

Primary Data Extraction

Two authors (K.K. and J.M.) independently extracted the following elements using a pilot-tested Google form: bibliographic details (name of journal, article electronic publication date, and PMID), repository details (database/repository name, data DOI, and website URL), funding source (government, private, industry, university, hospital, not funded, self-funded, and not listed), study design type (see screening/eligibility criteria), study topic (treatment/intervention or diagnostic/screening), journal data policies, article accessibility, presence and details of DSS (verbatim extract of the DSS), corresponding author email, data accessibility status (open access, unavailable, restricted, and embargoed), data format (raw, processed, and both), and data file type (.sav, .xlsx, and .csv). Although PubMed contains both electronic and print publication dates, we selectively extracted the electronic publication data, resulting in the inclusion of some 2017 studies. Discrepancies were resolved with another author (A.H.) to ensure reliable and accurate extraction.

Primary Data Analysis

A descriptive analysis using R (version R 4.3.1) in RStudio (version 2023.09.1+494) analyzed DSS prevalence by journal, study design, publication year, study topic, funding source, and accessibility (open access versus not open access). Annual DSS trends were also assessed. Journals with fewer than 40 publications were excluded due to insufficient data for trend analysis.

A hierarchical logistic regression analyzed DSS inclusion, addressing variability at the journal level. Because none of the journals shared a publisher, nesting by publisher was unnecessary. Categorical variables were converted to factors, and the journal impact factor was treated numerically. Samples with fewer than 30 observations were excluded to adhere to the central limit theorem.

Fixed effects for study design, impact factor, funding source, and accessibility were selected to evaluate their direct influence on DSS inclusion. This approach allowed us to isolate and examine the specific contributions of these predictors to the outcome. Random effects for journals accounted for the hierarchical data structure and modeling variability associated with differences among journals. This allowed us to identify the impact of fixed effects and understand how journal groupings influenced DSS use.

We assessed multicollinearity using variance inflation factors. This ensures our predictors did not compromise the accuracy of the calculated estimates by being highly correlated.

To improve the interpretability of our logistic regression analysis for readers, we simplified the presentation of our model results. We focused on reporting the influence of specific factors, such as study design, journal impact factor, funding source, and accessibility, on the likelihood of including a DSS. The intercept, representing the baseline likelihood of DSS inclusion in the absence of any predictor effect, is retained in the table but simplified for clarity. This approach facilitates a better understanding of how each predictor contributes to DSS inclusion.

Qualitative Analysis of DSS Themes

A theme analysis was performed to examine the variety of data statements presented within extracted articles. In conducting the thematic analysis of DSSs, we utilized artificial intelligence (AI)-assisted technology (ChatGPT-4o, OpenAI) to support the identification of common themes within the statements. AI was used to generate initial thematic groupings, which were subsequently reviewed and refined by members of the research team. Although AI provided an efficient means of processing data, the thematic outputs were rigorously checked to ensure alignment with study objectives and to account for any contextual subtleties that AI may overlook. This approach balanced efficiency with the accuracy necessary for qualitative analysis. ChatGPT-4o (https://openai.com/chatgpt/) was used to analyze the DSS verbatim from the pilot study sample, and 10 themes were established. These themes were used to describe the DSS in each study.

Verifying Data Accessibility

To assess the frequency of data made available upon request and the simplicity of accessing said data, we emailed corresponding authors with 2 questions: (1) Are you able and willing to share the de-identified data necessary for replication of your primary outcome analysis? and (2) If so, approximately when would you be prepared to send it for use in replication? The corresponding authors were informed that this email and their responses would assess the DSS practicality and follow-through. (See table, Supplemental Digital Content 1, which displays general study characteristics, http://links.lww.com/PRSGO/E13.) (See table, Supplemental Digital Content 2, which displays DSS by journal, http://links.lww.com/PRSGO/E14.) (See table, Supplemental Digital Content 3, which displays DSS theme analysis, http://links.lww.com/PRSGO/E15.) They were asked to respond but not provide data. If no response was received, a follow-up email was sent 1 week later.

RESULTS

Search Results

Our search string initially returned 814 original articles. After title and abstract screening, 759 articles were selected for extraction. We excluded 32 studies due to their low journal sample size. The remaining 727 articles underwent full-text data extraction, revealing that 715 articles did not include a DSS. Of the remaining 12 articles, 1 was excluded because its DSS was ineligible, directing readers to ClinicalTrials.gov where no data were available (Fig. 1).

Fig. 1.

PRISMA diagram. PRISMA, preferred reporting items for systematic reviews and meta-analyses.

General Characteristics of the Included Studies

Our analysis revealed that Plastic and Reconstructive Surgery generated the highest number of articles, accounting for 32.32% (235 out of 727) of our sample. In contrast, Ophthalmic Plastic & Reconstructive Surgery produced the fewest, with only 5.64% (41 out of 727) of our sample. “Clinical trials” dominated the study types, representing 64.51% (469 out of 727) of the articles, whereas “qualitative” studies were least common, comprising just 0.14% (1 out of 727) of the sample. We observed a general downward trend in publication numbers from 2021 to 2023, which may be attributed to differences in PubMed indexing. Funding sources varied considerably across the studies. Few studies received funding from various sources: industry (88 out of 727; 12.10%), government (56 out of 727; 7.70%), multiple (41 out of 727; 5.64%), private (36 out of 727; 4.95%), university (25 out of 727; 3.44%), and hospital (16 out of 727; 2.20%). However, the majority of studies were either not funded (291 out of 727; 40.03%) or had no funding information listed (174 out of 727; 23.93%). Notably, most articles were not open access (557 out of 727; 76.62%) (Supplemental Digital Content 1, http://links.lww.com/PRSGO/E13).

Trends in Data Sharing by Journal

Although the overall prevalence of DSSs was low, some journals included more DSSs than others. Aesthetic Plastic Surgery had the highest inclusion rate, with 5 out of 113 articles (4.42%) featuring a DSS. This was followed by the Journal of Plastic, Reconstructive & Aesthetic Surgery, with 3 out of 124 articles (2.42%), the Aesthetic Surgery Journal, with 2 out of 110 articles (1.82%), and the Plastic and Reconstructive Surgery, with 1 out of 235 articles (0.43%). In contrast, the Annals of Plastic Surgery, Journal of Cosmetic and Laser Therapy, and Ophthalmic Plastic & Reconstructive Surgery did not feature any DSSs. Notably, there was an increase in the percentage of DSSs from 2020 through 2022, followed by a decline in 2023 (Supplemental Digital Content 2, http://links.lww.com/PRSGO/E14).

Data Sharing by Study Design

Neither clinical trials nor cohort studies exceeded 10% in data sharing from 2017 to 2023. Among the study designs, clinical trials were the most common, comprising 64.51% (469 out of 727) and cohort studies were the least common, comprising 23.80% (173 out of 727) of our sample. Cross-sectional studies accounted for 5.78% (42 out of 727), whereas case reports/case series made up 4.68% (34 out of 727). DSSs were found exclusively in clinical trials (1.92%, 9 out of 469) and cohort studies (1.16%, 2 out of 173) (Fig. 2).

Fig. 2.

DSS by study design.

Data Sharing Themes

Authors acting as the gatekeeper role, controlling access to the data, occurred in 90.91% (10 out of 11) of articles with DSS. Conditional data availability was found in 81.82% (9 out of 11) of DSS. Privacy concerns appeared in 18.18% (2 out of 11) of the articles. Additionally, conditional timing for access and specified availability dates were noted in 18.18% (2 out of 11) of articles, indicating that the timing of data availability often depends on publication dates or specific milestones. Structured access platform and access mechanism and corporate ownership each accounted for 9.09% (1 out of 11). The remaining themes were not observed in our study (Supplemental Digital Content 3, http://links.lww.com/PRSGO/E15).

Factors Influencing the Presence of DSSs

Various factors influence the presence of DSSs within plastic surgery journals. Our hierarchical logistic regression analysis revealed a statistically significant low baseline probability of including DSSs in plastic surgery articles (estimate = −4.04, P = 0.008). Neither article access (P = 0.20) nor impact factor (P = 0.74) significantly predicted DSS inclusion. Publisher influence was notable: articles published by Springer (mean effect = 0.60) and Elsevier (mean effect = 0.34) were more likely to include a DSS. In contrast, articles from Oxford University Press (mean effect = −0.12), Taylor & Francis (mean effect = −0.19), and Lippincott Williams & Wilkins (mean effect = −0.43) were less likely to include a DSS (Table 1).

Table 1.

Hierarchical Logistic Regression Analysis With Random Effects Summary

Fixed Variable Entity	Estimate	Standard Error	Z	P	Interpretation
Intercept	−4.039	1.521	−2.655	0.008*	An estimate of approximately −4.0% with a P value of less than 0.05 indicates a significantly low likelihood of DSS presence at baseline
Impact factor	−0.208	0.633	−0.328	0.743	With a P value of less than 0.05, no significant relationship between journal impact factor and DSS presence was identified despite the estimate being approximately −0.2
Article access
Open access	1.035	0.805	1.286	0.198	With a P value of less than 0.05, no significant difference in DSS presence was identified between open access articles and nonopen access articles despite the estimate being approximately 1.0
Publisher:Journal sd__(Intercept)	0.000				An intercept of approximately 0.0 indicates little to no variability in DSS presence likelihood between journals within the same publishers
Publisher sd__(Intercept)	0.599				An intercept of approximately 0.6 indicates notable variability in DSS presence likelihood between publishers
Random effect entity			Mean effect
Publisher
Springer				0.595	For our study, the mean effect quantifies the association of DSS presence with each entity when compared with the mean, which is set to 0. Thus, greater positive mean effect values indicate a higher association with DSS presence and greater negative mean effect values indicate a lower association with DSS presence. Mean effects at or near 0 indicate little to no variability in DSS presence compared with the mean. Thus, our random effects summary indicates notable variability between different publishers, whereas minimal variability was noted at the journal level
Elsevier				0.340
Oxford University Press				−0.120
Taylor & Francis				−0.189
Lippincott Williams & Wilkins				−0.434
Journal (Publisher)
Aesthetic Plastic Surgery (Springer)				0.000
Journal of Plastic, Reconstructive & Aesthetic Surgery (Elsevier)				0.000
Aesthetic Surgery Journal (Oxford University Press)				−0.000
Ophthalmic Plastic & Reconstructive Surgery (Lippincott Williams & Wilkins)				−0.000
Annals of Plastic Surgery (Lippincott Williams & Wilkins)				−0.000
Plastic and Reconstructive Surgery (Lippincott Williams & Wilkins)				−0.000
Journal of Cosmetic and Laser Therapy (Taylor & Francis)				−0.000

1. Interpretation of estimates: positive estimates indicate an increased likelihood of having a DSS. Negative estimates indicate a decreased likelihood of having a DSS.

2. Random effects: intercepts for journals and publishers were included as random effects to account for variability across these entities.

3. Model details: a hierarchical logistic regression model with random effects was used to control for clustering at the journal and publisher levels.

^*Significance levels: P < 0.01.

Willingness to Share Data Upon Request

To assess DSS follow-through, we emailed the corresponding authors of the 11 articles claiming data availability but received no responses within 2 weeks.

DISCUSSION

Despite increasing advocacy for open science, less than 2% of plastic surgery research articles included DSSs, underscoring barriers to transparency. DSS inclusion varied across publishers, indicating that some are more likely to include them. Theme analysis revealed that approximately 90% of DSSs displayed a “gatekeeper role” but none mentioned public repository use. This discussion will relate our findings to existing research, identify data sharing obstacles in plastic surgery, and recommend improvement strategies.

Our study found a low prevalence of DSSs in plastic surgery research, consistent with prior literature. For example, Ascha et al¹² reported that approximately 3% of plastic surgery journals included DSSs. In our study, only 1.51% contained DSSs, indicating a decline in data sharing practices in this field. One major barrier is compliance with Health Insurance Portability and Accountability Act,¹³ particularly regarding photograph de-identification¹⁴ for patient privacy. Strategies such as digitally altering features have been proposed to improve anonymity.¹⁴ Informed consent practices also pose challenges, with research showing inconsistencies in obtaining prior photographic consent from patients.⁷ This can lead to incomplete data and compromise data quality. Funding disparities further impact data sharing,¹⁵ with many plastic surgery studies being unfunded compared with other medical specialties.¹⁶ To improve data sharing in plastic surgery research, we recommend implementing stringent data sharing policies, enhancing researcher training on data management and sharing, and encouraging journals to mandate DSS inclusion to promote transparency and reproducibility.

Our analysis indicated that articles published by Springer and Elsevier are more likely to include DSSs compared with other publishers. A review of the top 7 plastic surgery journals and their publishers revealed that while data sharing is encouraged, it is not required, if mentioned at all. For example, Springer strongly encourages data sharing without mandating a DSS, whereas Elsevier recommends checking individual journal policies, such as the Journal of Plastic, Reconstructive & Aesthetic Surgery, which encourages, but does not mandate data sharing.¹⁷ Inconsistencies in data sharing policies can hinder adherence, creating uncertainty for researchers. For instance, 1 journal “requires authors to include a data availability statement in their published articles,” but later adds, “the Data Availability Statement will be published in the final article only if the corresponding author chooses to make the data available.”¹⁸ The limited DSS use in our study likely reflects this lack of mandatory requirements. Stricter requirements are necessary for data sharing and the inclusion of DSSs to increase their prevalence in plastic surgery research.

Our theme analysis revealed that approximately 90% used a gatekeeper role, necessitating a reasonable request for data access—a subjective term that varies by author, risking inconsistent sharing and reduced transparency. To assess DSS follow-through, we contacted corresponding authors from the 11 articles claiming data availability upon request but received no responses within 2 weeks. This finding is worse than previous research by Gabelica et al,¹⁹ who found only 4% of authors in high-ranking anesthesiology journals shared their data despite having a DSS. Similarly, Gabelica et al²⁰ reported low data sharing compliance even in journals with strong policies, highlighting a persistent gap between intent and practice.

Adopting the Panton Principles of Open Science,²¹ the Findability, Accessibility, Interoperability, and Reuse principles,²² and the Transparency and Openness Promotion guidelines²³ can improve data sharing in plastic surgery. The Panton Principles promote free, accessible research, enhancing transparency and reproducibility. The Findability, Accessibility, Interoperability, and Reuse principles ensure that data are findable, accessible, interoperable, and reusable, thus maximizing research value. The Transparency and Openness Promotion guidelines provide comprehensive standards for transparency and reproducibility that can foster better practices in data sharing. Leveraging these frameworks can help overcome barriers such as privacy concerns and inconsistent data availability, fostering a collaborative and open research environment that improves research quality and patient care outcomes in plastic surgery.

Our study’s strengths include adherence to preferred reporting items for systematic reviews and meta-analyses guidelines, protocol registration on the Open Science Framework for transparency, and a priori development with documented amendments. We used masked duplicate data extraction to minimize bias, ensuring reliable findings. However, limiting articles to 2018 through 2023 may affect generalizability, and the short 2-week author verification period could have reduced response rates, impacting data comprehensiveness. The low response rate from corresponding authors may have also been partially due to emails being filtered into spam folders. Although we used institutional email addresses and clear subject lines to reduce the likelihood of this issue, we cannot entirely rule out the possibility of emails not reaching the intended recipients. Despite these limitations, our methodology establishes a comprehensive basis for examining data sharing practices in plastic surgery research.

Our analysis of the top 7 plastic surgery journals found that DSSs appear in only 1.51% of publications. Despite journal encouragement, privacy concerns and inconsistent data availability requirements contribute to this low prevalence. Standardized data sharing policies, improved patient photograph de-identification, and better-informed consent are crucial for enhancing transparency and reproducibility, ultimately advancing research and medical decision-making.

DISCLOSURES

Dr. Vassar reports receipt of funding from the National Institute on Drug Abuse, the National Institute on Alcohol Abuse and Alcoholism, the US Office of Research Integrity, Oklahoma Center for Advancement of Science and Technology, and internal grants from Oklahoma State University Center for Health Sciences—all outside of the present work. Dr. Ito Ford reports receipt of funding from the Center for Integrative Research on Childhood Adversity, the Oklahoma Shared Clinical and Translational Resources, and internal grants from Oklahoma State University and Oklahoma State University Center for Health Sciences—all outside of the present work. The other authors have no financial interest to declare in relation to the content of this article.