Data sharing implementation in top 10 ophthalmology journals in 2021

Introduction

The International Committee of Medical Journal Editors (ICMJE) proposed the implementation of deidentified individual participant data (IPD) sharing in their fellow journals in 2017.1 This policy has already been included in the ICMJE Uniform Requirements for Manuscripts Submitted to Biomedical Journals (URM),2 and this was also another big step for ICMJE to create a better research environment following the implementation of clinical trial registration into ICMJE journals in 2005.3 The idea of IPD sharing is based on the principle that participants in clinical trials put themselves at risk for the benefit of others, therefore, these priceless data generated from each participant should be used to get the most benefit out of them. Data sharing would lead to benefits for patients, progress in science and rational use of healthcare resources.4

The benefits of IPD sharing were quite evident. First, it should help increase confidence and trust in the results and conclusions from clinical trials through better transparency of the data. Second, it should encourage independent analysis and confirmation of results from a third party who is interested in the same research field. Third, development of new research should be easier through datasets mining. Finally, IPD sharing would help decrease repetition of clinical trials since tremendous data would be available without having to conduct new research.5

More importantly, IPD sharing was not only supported by ICMJE, but also by most clinical trialists and study participants as they thought that its benefits outweighed the drawbacks. Consequently, they were willing to share the data with others.6 7

Implementing a new policy is not an easy task. Since the announcement of the previous clinical trial registration policy, there were still some clinical trials and journals which did not strictly follow the policy, even though these works were started and published years after the announcement of the policy.8 9 Following the implementation of IPD sharing policy by the ICMJE in 2017, there were some publications from various journals which did not endorse and follow the new plan.

From a previous study, the percentage of endorsed and implemented ICMJE clinical trial registration policy varies across different disciplines.9 Thus, we expect that a similar trend might be observed in IPD sharing policy. The aim of this study is to examine the number of published clinical trials in 2021 among the 10 highest journal impact factor (JIF) ophthalmology journals that endorsed the IPD sharing policy.

Methodology

Ten ophthalmology journals with the highest ranking according to the 2020 JIF were identified. The endorsement of the IPD sharing policy was determined by gathering information from the journal instruction to authors section, which were accessed before 1 April 2022. Then, all articles from these journals published in 2021 were included into this study. We excluded the following types of articles: editorials, letters to editor, case reports, conference abstracts, pictures or illustrations, and preprints from analysis.

The study protocol of these publications was assessed by two assessors whether they met with the WHO definition of clinical trials which consists of the following: (1) prospectively assigned human participants to one or more health-related interventions, (2) health-related interventions and (3) health-related outcomes.10 If there was a discrepancy between the assessors, a consensus was reached by a discussion between both assessors. The publications that met the definition were then assessed for their clinical trial registration status. We considered ‘registered’ if the clinical trial registry identifier number was shown in the manuscript, or the study registration with the same protocol was found in the WHO International Clinical Trials Registry Platform11 if the clinical trial registry identifier number was not available in the manuscript in the first place.

To evaluate the implementation of IPD data sharing, we looked for statements about IPD sharing in the manuscript and in the clinical trial registry of each research. We classified articles that endorsed this policy when the IPD data sharing statement was present either in the manuscript or the clinical trial registry. The start and end dates of the sharing period and the methods of sharing among publications with data sharing statements were collected.

We measured the number of published clinical trials that implemented IPD sharing policy as our primary outcome. We also measured the number of published clinical trials that were registered in a clinical trial registry. The dataset of this study is publicly available via OSF platform.12 The details and flow of the study results were shown in figure 1.

Request permission

Results

We identified 10 highest ranking ophthalmology journals according to the 2020 JIF. The median of impact factors was 5.653. Among these journals, two journals only accept reviews or systematic reviews, and all other journals, which accept clinical trials, had endorsed IPD sharing policy by mentioning in their instruction for authors section.

From a total of 1852 original articles, 175 (9.45%) were classified as clinical trials, and among these clinical trials, 77 (44%) articles had been registered in a clinical trial registry as shown in table 1.

Interestingly, 86 (49.14%) publications included a statement about their IPD sharing policy either in the manuscript or the clinical trial registry. Of these, 42 publications declared that they were willing to share their IPD with others. In addition, there was a single journal that had all clinical trials registered in a clinical trial registry. However, there were two journals that all clinical trials included data sharing statements.

For the method of sharing IPD, 89.79% of corresponding authors stated that they would share the data with researchers on their request. The other 10.21% decided to upload their deidentified data into an online public database repository platform, such as pedig.net, or vivli.org. For the sharing period, the IPD from 37 publications were shared right after the publication of these articles. Whereas, three were shared after a certain date given by the corresponding authors within 6 months after publication date, however, there were two publications that did not define their sharing date. The ending date of the sharing period was stated in only two publications, while others did not specify the date.

Discussion

The 10 highest impact factor journals were used in this study because the impact factor of the publishing journal was a strong predictor of citations per year,13 therefore, these publications in high-impact factor journals were more likely to be read, and cited by others. We found that all of the journals included in our study endorsed the IPD sharing policy and clearly stated in their instruction to authors. Some journals also encouraged researchers to share these data as soon as possible.

We found that the number of publications with IPD sharing statements is higher than those with clinical trial registry. This might depend on journal publishing regulations in which some of the journals strictly required their publications to include the statement even if they did not declare their clinical trial registration. In terms of declaring IPD sharing statements, authors of publications submitted to strict journal regulations were more likely to adhere to the policy.9

Moreover, our study found that from the total of 42 clinical trials which were willing to share their IPD which was less than a quarter of the total published clinical trials, and less than half of the total clinical trials with IPD sharing statement. A recent review showed a similar trend on a discrepancy between the willingness to share data and the intention to actually share them.6 Most clinical trialists were optimistic and willing to share data, but the actual data sharing rates were between 10% and 46% of total published clinical studies, which was relatively low.6 14

Most authors would like to share IPD on request from other researchers. This sharing method could allow corresponding authors to have control over the data, and be able to withhold the data if the request is not applicable to the purpose of sharing.15 While most authors share on request, a few other clinical trialists used an online data repository platform for sharing IPD which is consistent with the results from a study by Danchev et al.14 Some platforms were specifically created for certain specialties, societies or countries, and each one had its own guidelines and tools for data storage. According to a study by Banzi et al, some data repository platforms cover all research areas, host IPD from clinical studies published in medical journals, and also link the published data from publishers into the platform. There were platforms that exclusively made for sharing clinical research data.16 These platforms could allow researchers to share and retrieve IPD easier, nevertheless, there were still some concerns, for example, potential data breaches, trust of secondary use, lack of knowledge or research requirement, and misinterpretation of the data which are all beyond control.17 Therefore, increasing the amount of dataset previews, linking to clinical trial registries, helping tools and governance on each platform from institutions or government sectors could improve the discoverability, utility and transparency of datasets on online repository platforms.17 18

From the results, we found that only 44% of the clinical trials were registered in a registration platform. Although the clinical trial registration policy has been declared by ICMJE since 2004,3 the rate of registration was still below half of all clinical trials. We also reached out to the corresponding authors of these clinical trials for more details and explanation about the missing clinical trial registration. Some of them did not consider their research as clinical trials, therefore, registration of these studies was not necessary from their perspective. This highlighted the fact that even though the clinical trial registration policy has been implemented for years, some clinical trials still fail to follow the policy.

The results of this study suggest that strict journal submission guidelines on IPD sharing policy can help increase the number of articles endorsing this policy. This is consistent with the results from a previous research on the clinical trial registration policy that journals that are followers of ICJME’s URM guidelines with strict submission guidelines had a greater opportunity to endorse and implement the policy.9

Our evaluation had some limitations. This study is a cross-sectional perspective of the current state on how IPD sharing policy works. The number of studies endorsing and implementing the policy are still growing every year, and the attitude among researchers seems to change over time in favour of IPD sharing.19 Therefore, an update analysis on this matter should be repeatedly assessed in the future. Moreover, our analysis only included publications in top ophthalmology journals which might not be able to be generalised to other research fields since the implementation varies from specialty to specialty. 9 Additionally, we did not extensively explore the researchers’ opinion and perspective on the IPD sharing policy, which could better explain the IPD sharing rate in this study. Future studies are necessary to understand the factors that influence researchers’ decisions on IPD sharing, therefore, they may be helpful in increasing the implementation of the IPD sharing policy. The protocol was not registered with a clinical trial registry, as it was not considered as a clinical study as defined by WHO and ICMJE. However, the methodology was meticulously described to ensure reproducibility by future investigators.

In conclusion, although the IPD sharing policy has been implemented since 2017, the number of IPD sharing among clinical trials published in high-impact factor ophthalmology journals was lower than half, and the number of trials that actually shared IPD was much lower. Besides, only a half of all clinical trials were registered in a clinical trial registry. To improve the rate of endorsement, strict journal submission guidelines and cooperation from researchers were necessary.