Our Communications Manager, Katie M. Kanagawa, interviewed Epidemiology and Population Health Postdoctoral Research Fellow, Titilola Falasinnu, about her research on increasing ancestral diversity in clinical trials (especially in lupus) and ways to measure pain using electronic health records (EHRs).
Can you start by telling us a bit about yourself? How did you get here (to Stanford Epidemiology & Population Health)? Was there something in particular that attracted you to the fields of science, health and disease?
I’m a classically trained epidemiologist in that I worked as an HIV surveillance epidemiologist at the health department in Washington DC right after my masters degree. My doctoral work was conducted at the British Columbia Centers for Disease Control (BCCDC) where I developed clinical prediction rules for screening HIV and sexually transmitted infections for the province. Part of my work at University of British Columbia included contributing to clinical and provincial interventions aimed at sexual minority groups, with a focus on ways to avoid unintended consequences of further marginalizing individuals that have these identities. However, my personal experiences with lupus contributing to the death and reduced quality of life of several friends and family members spurred my interest in the disease. This is what led me to the Simard Lab at Stanford. My work at Stanford spans outcomes, clinical trials and pain research in rheumatology.
I understand you have worked on increasing ancestral diversity in clinical trials (particularly in lupus). Can you please give us an overview of this research? What is the important problem you are working to solve and how have you approached solving it?
One of the most fascinating discoveries during my postdoc was that black patients comprise 43% of surveillance estimates of lupus but comprise only 14% of clinical trial enrollees. We replicated this finding in rheumatoid arthritis (another autoimmune rheumatic disease) and discovered even more stark disparities in this condition. Black patients comprise 13% of national prevalence estimates of rheumatoid arthritis but make up only 3% of trial enrollees. Hispanic patients comprised 18% of surveillance estimates but make up only 4% of trial enrollees.
The under-representation of minorities in trials is not a problem that is unique to rheumatology. These problems are evident in oncology and many other medical subspecialities. We do not know the extent to which clinicians are actively approaching racial minorities to participate in trials and there is an abundance of training opportunities provided to the clinical workforce to mitigate implicit bias in encounters that may lead to participation in trials. Reflecting on Marcella Alsan’s seminal work at Stanford on clinician/patient racial concordance and the effect of diversity in the physician workforce on improving the demand for preventive care in Black men, it is obvious that we need a more diverse trialist workforce. For example, ~1% of rheumatologists in the US are black.
We have to acknowledge that the inclusion of ethnic minorities in trials adds complexity and cost. However, historical precedents illustrate the even higher costs for neglecting diversity in biomedical research. For example, the earliest iterations of the HPV vaccines did not have some of the most prevalent subtypes of the virus found in Black women. Subsequent HPV vaccines now include some of these subtypes.
I suggest three ways to increase ancestral diversity in trials (not just rheumatology trials). First, is that we need to learn from our colleagues in the social sciences, especially participatory research, an approach that involves handing power from the researcher to research participants. I propose a democratizing framework to enhance diversity in the research pipeline. This would involve a deliberate engagement of communities of color across the life cycle of drug development from biomedical research, drug discovery, clinical trials and post-approval surveillance. For example, partnering with researchers at Historically Black Colleges and Universities can facilitate this engagement as many of these institutions are trusted in the Black community. Second, it is also important to ensure that racial minorities are retained in trials. There are peer mentoring programs such as the patient navigation programs that originated from oncology trials. Patient navigators are lay community health workers hired to educate patients about trials and provide individualized support for patients enrolled in trials. One study found that 75% of black patients who received patient navigation support completed the trial compared to 38% of those who did not receive patient navigation support. Finally, the lack of diversity may become a generalizability problem, especially if certain subtypes of disease are found to be more prevalent in racial/ethnic minorities. For this reason, I suggest that leveraging disease registries to identify potential trial subjects for screening for eligibility and enrollment can be a way to ensure equity.
What has driven you to pursue this research? What have you loved most, or found the most rewarding, about it?
I think my work in HIV research has prepared me for my future in lupus research. There are a lot of parallels between HIV and lupus, in terms of how these conditions and their outcomes disproportionately impact racial/ethnic minorities. It’s fascinating how the challenge of achieving equity in health outcomes seem to be a recurring them in these conditions. This is why I take pride in my personal connection to lupus and the most rewarding part of being a lupus researcher is knowing that I’m asking research questions that I know reflect the concerns of women of color like me.
You have also been exploring your emerging interest in ways to measure pain using electronic health records (EHRs). Can you please tell us a bit about that new research project and what is drawing you in this new direction?
Up to 100 million Americans live with ongoing pain, costing $635 billion annually. In 2016, the National Institutes of Health published the National Pain Strategy as a roadmap to reduce the burden of chronic pain in the United States. They called for “steps to increase the precision of information about chronic pain prevalence…to enable evaluation of population-level interventions and identification of emerging needs,” particularly in the leveraging of electronic health records (EHR). My goal is to use computational phenotyping, a clinical data science method that leverages data driven methods, to subtype and characterize patient conditions from heterogeneous EHR data, with a particular focus on lupus. The methods developed during this project will be exemplar for future disease-specific chronic pain research outside of lupus.
Why is this an important area of research to study at this particular point in time?
Living with a chronic disease such as lupus confers multiple challenges. Pain is a frequent self-reported symptom in lupus and is often one of the first symptoms of the disease. Pain takes many forms in lupus: musculoskeletal, headache, abdominal pain, pleuritis, secondary fibromyalgia and Raynaud’s phenomenon. Recently, 32% of patients with lupus reported pain and/or swelling as most negatively impacting their lives. Despite treatment advances, pain remains the most prominent, unaddressed patient complaint. Evidence suggests that more research is needed to address the problem of pain in lupus. For example, we found that only 0.2% of published studies in lupus examine pain compared to 9% of osteoarthritis and 1% of rheumatoid arthritis studies.
To my knowledge, no studies have used multiple EHR components, e.g., combining structured and unstructured fields for creating chronic pain algorithms that have broad utility for future studies. The uses of the chronic pain computational algorithm include: (1) Estimate incidence, prevalence and trends in administrative databases/EHR, (2) provide baseline measure to characterize pain trajectories in administrative databases, (3) Understand the epidemiology of other chronic overlapping pain conditions, (4) Facilitate cost modeling and health utilization projections, and (5) Characterize treatment response and effectiveness.
What do you hope to accomplish with this new line of research? What larger impacts do you hope to make, scientifically and perhaps societally speaking?
The bigger goal of this new line of research is to understand risk factors for the development of chronic pain in lupus using EHR and translate that knowledge into effective strategies to reduce pain and suffering in lupus.