Imagine spending weeks—or even months—collecting data around the clock, only for it to sit unused after your study is published. It’s a scenario many researchers in circadian rhythm and mental health know all too well. Despite the potential for shared data to accelerate discoveries and reduce duplication, it’s still the exception rather than the rule. A recent assessment by the UK Circadian Mental Health Network (CMHN) found that only 8 out of 114 studies published in 2023 shared their data. And when data was shared, it was often locked away in hard-to-access formats, making it nearly impossible to reuse.
What’s Holding Us Back?
The challenges are real. To begin with, the data in our field is incredibly diverse, spanning everything from molecular and physiological measurements to clinical records and wearable sensor data. Because circadian data often involves time-series data, raw datasets can be very large, and the original pre-processed might not have generated the right variables or ‘features’ to answer the new research question. There are significant challenge managing and integrating these datasets, but even when the data is relatively straightforward—like simple survey data—it often goes unshared.
This stands in stark contrast to fields like molecular biology or genomics, where data sharing is far more common and ingrained in research practices, despite the complexity of the data. This highlights that data complexity alone cannot explain the lack of sharing in circadian research. Instead, it brings us to another major concern: ethical considerations. As one of CMHN’s principal investigators highlighted:
“We did not have consent from these participants to share their data. This was a small sample size, so there was a higher chance for reidentification. We also felt that the NHS ethics committee would not approve data sharing given the small sample size and clinical sample.”
As a result, researchers often feel more comfortable sharing data from animal models or volunteer studies than from clinical trials.
But perhaps the biggest barrier is cultural. Academic incentives still prioritise publications and citations over open data practices. Many researchers worry about losing their competitive edge or being “scooped” if they share their data too soon. As another PI pointed out:
“Clearer guidelines on data sharing are needed. I think journals will need to require it for there to be substantial change.”
People Are Willing to Discuss, But Challenges Remain
In recent efforts to access participant data for re-analysis in a pilot project, Sophie Faulkner, a postdoctoral researcher at the University of Manchester, encountered both encouraging and frustrating experiences. While some researchers were unresponsive, the majority were open to discussing data sharing after some follow-up. The barriers she encountered were often practical rather than ideological: a lack of resources or plans for providing data after studies had closed, data that had been lost, or epoch-level data that was never downloaded because it wasn’t needed for the original study’s aims.
“I was pleasantly surprised by how open and willing people were to discuss sharing data,”
Sophie noted.
“It feels like attitudes are definitely changing. However, it’s clear that many researchers aren’t thinking about data sharing from the early stages of their projects, which can create hurdles down the line.”
How Can We Fix This?
At the CMHN, we’re tackling these challenges head-on. We’ve started by offering data curation services to help researchers prepare their datasets for sharing. For example, we recently assisted Prof. Andrew Coogan’s team at Maynooth University in organizing and documenting their data, making it easier for others to access and use.
We’ve also launched training programs to help researchers adopt the FAIR principles (Findable, Accessible, Interoperable, and Reusable). So far, we’ve delivered two specialised courses on FAIR data in chronobiology and a general course for principal investigators (PIs) in the CMHN. Keep an eye on our mailing list for upcoming sessions and workshop!
The data we collect is more than just numbers—it’s the key to understanding some of the most complex questions in science. Let’s make sure it doesn’t go to waste
What Needs to Change? Our Recommendations
Recommendations for researchers | |
Use general repositories | Use platforms like GitHub and Zenodo for sharing data, especially in the absence of specialised repositories for specific data types. |
Future repository linking | Instead of vaguely stating future data availability (data will be open), include a specific repository link in the data availability statement where data will be hosted, such as Zenodo’s versioning system or GitHub repositories, to ensure accessibility and proper version control. |
FAIR training | Engage in training on FAIR principles and data management to improve data handling and sharing practices effectively. |
Anonymise patient data | Implement data anonymisation to allow sharing sensitive patient data to maintain privacy and comply with ethical standards. |
Ensure ethical data sharing | Seek ethical approval for data sharing to allow the sharing of full or partial datasets where possible, and secure appropriate participant consent. |
Add a licence | Sharing data without a clear licence complicates reuse. Open licences (CC-0 or CC-BY) are recommended. |
Share valuable metadata | Share metadata, research protocols and code, which are essential for replicability and transparency, even if data cannot be shared openly. |
Use journal policies | Read the journal’s data-sharing policy and evaluate article compliance, as a reviewer or editor. |
Recommendations for institutions | |
Specialised training and curation services | Provide targeted training and data curation services to help researchers effectively manage and share data within specific research domains. |
Institutional policies for data sharing | Develop policies that reward data sharing in recruitment and evaluation processes to encourage and recognise researchers who contribute to open data practices. |
Best practices templates | Publish exemplary cases that illustrate successful integration of data-sharing practices within informed consent, data management and funding proposals to guide and motivate researchers. |
Standardised README templates | Create and disseminate domain-specific README templates to ensure consistent and comprehensive data documentation across different research areas. |
Recommendations for publishers and funders | |
Integrate data verification in peer review | Evaluate the data-sharing statement and the data shared in the peer-review process for publication. This ensures that submitted data meet high standards of accessibility and usability. |
Reward prior data-sharing efforts | Consider researchers’ previous data-sharing practices when allocating funds. Offer specific grants and funding opportunities to researchers who have demonstrated a strong commitment to open data practices. |
Support for domain-specific repositories | Provide continuous funding and support for the development and maintenance of domain-specific data repositories. This will enable tailored data storage solutions that meet the unique requirements of circadian and mental health research data, as an example. |
CMHN, Circadian Mental Health Network; DMP, data management plan; FAIR, Findable, Accessible, Interoperable, Reusable; MRC, Medical Research Council; UKRI, UK Research and Innovation.
You can read a recently published article on FAIR data from Network members here: https://mentalhealth.bmj.com/content/27/1/e301333
Blog by Haya Deeb and Sophie Faulkner
Comments