A: In my view, both are absolutely true – patient healthcare data aggregation allows for great breakthroughs and is also challenging from an ethical perspective in terms of privacy and patient choice. Combining information from health records for millions of people gives us insight into disease and disease processes in way that can allow us to predict and share, with some precision, what the disease trajectory might look like for an individual patient. At the same time, patients continue to express great concerns about data privacy. We need to be able to consider both issues simultaneously – our need for large numbers and the need to consider patient choice and control in the sharing of records.

A: Historical, societal, and medical traditions are often not our friend when we are looking to have people share their health data – especially vulnerable populations who are often at the highest risk for the health outcomes we are studying – for example heart disease, diabetes, poor birth outcomes. A history of exploitation and implicitly and explicitly racist and classist treatment of certain individuals (e.g. Black, poor, and immigrant people) in medical and medical research settings means that many people don’t trust the medical profession and medical research professionals. Until this tradition is addressed in a multi-factorial way I am not sure it is reasonable to even have the expectation that these groups would willingly share their data.

A: While we are making some headway in addressing historical traditions in medicine and medical research that have been both dismissive and exploitative (through training around issues like implicit bias) in my view we won’t really start to break down these barriers until we do two key things: 1) we must elevate and partner with other professionals across race/ethnicity and socioeconomic groupings in all clinical and research endeavors (whether focused specifically on disparities or more broadly on health). Only with inclusion of these individuals as partners and leaders can we begin to understand the full picture and bring whole populations into the fold in this journey) ; 2) we must include study participants – including those who share their medical records with us, as partners in our work (inclusion of community advisory boards in all planning processes is a good step along this road but it goes much further – commitment to sharing data with participants is another element of this, including patients and participants in leadership and decision making groups is another – asking participants and the affected populations what THEY think we should do and then DOING IT is another — paying participants for their time is another — the list goes on and on but is really nested in true partnering).

A: We have a prospective study underway that I think exemplifies good strategies for approaching research with vulnerable populations and how we might engage larger populations in investigations where data sharing is needed in order to advance discovery to interventions to improve health for those most in need and for everyone. The Supporting Our Ladies And Reducing Stress to Prevent Preterm Birth (SOLARS) study in Oakland, California, funded by the UCSF California Preterm Birth Initiative (UCSF PTBi-CA), is investigating the role of stress in the observed disparities in preterm birth (delivery before 37 completed weeks gestation) in Black and Latina women using a community-engaged, community-based approach for enrollment and follow-up with a focus on identifying low-income women. The study focuses initially on discovery with the ultimate goal of uncovering interventions and involves collection of survey data throughout pregnancy on stress, psychological well-being, and health as well as biospecimens and hospital record review and integration. When we had the idea for this study many other clinicians and researchers questioned whether vulnerable women would engage deeply in this work with us and contribute biospecimens across pregnancy but the community advisory board of the UCSF PTBi-CA felt strongly that because the study was led and staffed by women with lived experience and women of color and because it focused specifically on understanding the role of stress and birth outcomes specifically in Black and Latina women within their geography that women would want to participate and would even be excited to do. Our pilot that tested whether this was true was hugely successful – very vulnerable Black and Latina women engaged with the study, stayed in the study, and expressed enjoying participating. We are now working to enroll 500 women in the study and plan to go bigger – increasing the number of women and locations as the study progresses and as more funding is secured. I think this study continues to be successful because women see themselves in the study and the study teams and feel that they are contributing to the health of their communities – this seems to be absolutely key. They are our partners not our study subjects.

A: I think we may see some increase in enrollment into trials if we work to partner more with communities in the work we do. We likely need to shift our models of how and where we enroll participants starting with the communities that are affected most by whatever it is that is being researched. If it is an intervention or treatment focused on heart disease or cancer – talk to people with heart disease or about what they think – might recruitment in support groups be helpful, do they have connections to other groups in the community that might be supportive or might even allow recruitment in their offices? Shifting how and where we engage and again, partnering, is key in my view. We may have to actually begin to require some level of patient and community engagement in funding proposals and human subjects applications to push engagement and shift how research is done but it seems like true progress might require this and that such partnering would contribute to the generalizability of findings. Once partnering takes hold open sharing of information will likely feel less threatening because individuals are being included in decision making – that is what partnership is – shared engagement, shared benefits.

A: I think we have to shift this idea of needing to “incentivize” people to share their data to one where we really focus on how we can partner with people to understand better why they may not want to share and also work to include all kinds of people with all kinds of backgrounds in how we approach sharing. I think we need to talk about why and with whom sharing might happen and be open to people having some choice in how this happens and with whom. Sometimes people are okay sharing some kinds of data and not other types or sharing for certain reasons and with certain groups. We also need to be super mindful of who is reaping the benefits of sharing and making sure that those benefits reach those who contributed data – this may be in the form of, for example, making sure we pay people for their time when contributing data or in making sure the gain more information about their own health. An all or none proposition to sharing may lead to our never having data that is truly representative of our populations.

A: Yes, there may always be communities or populations that don’t want to share but we must not make assumptions about who these groups are. We need to engage people and groups as partners and be flexible around what sharing looks like.

A: I think consideration of personal technology is key to think about broader sharing of data. Of course there is the “sharing” that happens without a person’s knowledge or approval (or via an approval that was buried deep in long, rarely read approval when you downloaded an application on your phone for example) and then there is the sharing that is transparent and truly agreed to. I think shifting from a burying agreements to share and truly transparent sharing can only help in the long-run because it suggests that a person is valued and that their privacy and choice is prioritized.

A: This is a tough one. What I will say is that I hope it looks more like, for example, the 23andMe model where individuals are asked if they want to contribute their genetic data to certain research projects as they arise and if they say yes – great, and if they say no that is okay too. I think this kind of choice-making leads to more “yes” answers because it is clear that people are valued and their wishes and opinions are prioritized. In my opinion greater flexibility in sharing helps at multiple levels – better results, better generalizability and greater trust in the reasons and questions behind what we are asking people to share and how they, their families, and their communities will benefit.