For over a decade, the open science movement has been heralded for accelerating scientific discovery resulting in more rapid translation of research findings that are critical to public health (McKiernan et al. 2016). Open science efforts have led to the availability of large population neuroscience datasets that have revolutionized cognitive and clinical neuroscience (Uddin, Castellanos, and Menon 2024; Biswal et al. 2010), especially with respect to health disparities research that is critical for informing the development of interventions to promote health equity (Harnett, Merrill, and Fani 2024). Such big data use comes with the responsibility to ensure ethical conduct of data analysis and interpretation to prevent further stigmatization of historically marginalized groups (Laird 2021). As investigators, federal partners and journal editors, we have a responsibility to communicate expectations surrounding acceptable data use and discourage misuse of population neuroscience data.

Complex datasets like the Adolescent Brain Cognitive Development^SM (ABCD) Study underscore the need for caution in how human subjects' data are generated, modelled, analysed, interpreted and communicated to both fellow scientists and the community at large. Creators and users of large datasets are responsible for promoting data use that avoids harm to individuals or populations. Using the ABCD Study as an example, we provide recommendations to help ensure large datasets are used responsibly and do not further stigmatize, marginalize or otherwise harm minoritized groups.

The primary research goal of the ABCD Study is to determine how experiences during childhood interact to influence brain and cognitive development including social, behavioural and health outcomes (Jernigan, Brown, and Dowling 2018). ABCD Study data collection and sharing are collaborative efforts requiring expertise from multiple scientific disciplines and academic, industry and government institutions, as well as the communities under study. The ABCD dataset has the potential to produce valuable insights into cultural and environmental factors relevant to youth that may influence different health trajectories. However, when studying health disparities, researchers often broadly define social constructs such as race, ethnicity, gender and socioeconomic status, using them as proxies for other experiences. These constructs should be avoided as independent variables in isolation (Hoffman et al. 2022). Rather, models should incorporate contextualizing variables such as family, social and neighbourhood environments, school and recreational experiences, vocational environment, acculturation and perceived discrimination, among others, especially in predictive models that attempt to explain variability in behaviour, cognition and mental and physical health outcomes. The use of these constructs necessitates thoughtful consideration to prevent further stigmatizing or marginalizing youth. By way of example, attribution of differences in cognitive measures between children of different reported races is particularly sensitive. Interpretations may be incomplete and misleading when results do not consider and contextualize findings in both the communities from which the samples were drawn and historical context. With these additional considerations, authors can enhance the accuracy of scientific understanding and avoid the current legacy of trauma, discrimination-induced stress and marginalization that may contribute to or be the cause of those differences.

Professional reviews and published reports include recommendations supporting community-engagement to enhance the validity of constructs and measures and improve the integrity of analytic efforts as well as accuracy of communicated results. For example, the principles of Indigenous Data Sovereignty and Governance can guide use of public datasets for research related to American Indian and Alaska Native populations (White et al. 2023). Researchers are also advised to consider equity-based questions throughout the research development and analytic processes (Bodison et al. 2023), such as “Have you acknowledged any potential bias in measures/constructs (known or suspected)?” and “Have you been careful to contextualize variables, such as race/ethnicity/, gender and/or SES?”

The NIH has taken steps to ensure that researchers accessing the ABCD dataset are properly trained in responsible use of data. For upcoming data releases, training modules accompany the Data Use Certification (DUC) workflow and must be completed before access is granted. For 2024 data releases and beyond, the DUC includes language about compliance with human subjects' protection requirements. In particular, researchers must agree to conform to the ethical conduct of research and consider any potential psychological, social, economic and other potentially harmful impacts their research results could have on individuals, communities and society and take steps to minimize such impacts. Additionally, the ABCD Data Analysis, Informatics and Resource Center and Justice, Equity Diversity and Inclusion (JEDI) workgroups have articulated special considerations for the use of variables, particularly historically misused social constructs. The ABCD Wiki includes specific notes for use of individual variables and composite metrics and recommended practices to assist in the consideration of confounding and explanatory influences in effects under examination. These transparency efforts can improve the reproducibility of ABCD findings (Lopez et al. 2024).

With the increasing availability of large, multidimensional datasets, researchers bear additional responsibility for ethical measure development, data use, analysis, interpretation and communication of findings. While new strategies and techniques will continue to be developed, it is already the responsibility of the individual researcher to attend to best practice data use methods to minimize the perpetuation of stigma and harm to individuals, communities and society. We therefore urge researchers to attend to the recommendations provided here and elsewhere to help ensure fair, inclusive and just science.

Author Contributions

Sandra A. Brown: conceptualization, project administration. Hugh Garavan: conceptualization, supervision. Terry L. Jernigan: conceptualization, supervision. Susan F. Tapert: conceptualization, project administration, writing – original draft. Rebekah S. Huber: conceptualization, project administration, supervision, writing – original draft. Daniel Lopez: conceptualization, project administration, resources, supervision. Traci Murray: conceptualization, project administration, resources, supervision, writing – original draft. Gayathri Dowling: conceptualization, supervision. Elizabeth A. Hoffman: conceptualization, project administration, supervision. Lucina Q. Uddin: conceptualization, project administration, resources, supervision, writing – original draft.

Disclosure

The views and opinions expressed in this manuscript are those of the authors only and do not necessarily represent the views, official policy or position of ABCD consortium investigators, the NIH, the US Department of Health and Human Services or any of its affiliated institutions or agencies.

Conflicts of Interest

All authors report no biomedical financial interests or potential conflicts of interests.