Control-Alt-Delete: Why data integrity matters in medicine

At the start of the year, thousands of publicly accessible datasets and research papers related to such topics as chronic disease, environmental health, adolescent health, sexual and reproductive health and substance use were abruptly removed across several United States government agencies.

As scientists scrambled to salvage and archive this data for further study, the research community was dealt another blow with the announcement of a funding freeze at the National Institutes of Health, threatening the viability of both ongoing and proposed research projects.

In our roles as clinician-scientists in training, we observed these rapid developments with caution and concern, knowing that we likely would not be able to predict the full impact of these decisions on our nation’s health for years to come.

In fact, it seemed both chilling and timely that our issue of the AMA Journal of Ethics had gone live in January. Inspired by our shared coursework and research interests while pursuing post-lockdown graduate study at the London School of Hygiene and Tropical Medicine, we applied for an editorial fellowship to center a theme issue on the intersection of epidemiology and medicine – two distinct but interrelated disciplines.

Both disciplines share the aim of improving health and well-being. Medicine centers on personalized healthcare for individuals and families, whereas epidemiology encompasses the study of the health of populations. Essential to both fields is the iterative process of collecting, analyzing and utilizing data – then doing it all again.

But what happens when data is missing or flawed? We know that when populations are excluded from or underrepresented in research, we end up drawing conclusions that can’t be universally applied – a classic example being the higher risk for missed or delayed diagnosis of myocardial infarction in women due to symptoms that are considered “atypical.”

Individuals and communities who aren’t well-represented in research also tend to be those whom health systems disproportionately and inequitably fail to reach: those who are uninsured or do not have a job or transportation or are experiencing food insecurity; those who live with chronic illnesses or disabilities; or those who experience language and cultural barriers when attempting to access care. Data, therefore, reflect wider social and structural inequities, biasing how evidence is applied in clinical practice: specifically, guidelines and formulas that draw upon epidemiological research can sometimes stem from and propagate social and institutional biases, which can lead to distrust of medicine, scientists, and the healthcare system at large.

For these reasons, the disappearance of data sets and the throttling of research funding should galvanize not only the scientific community but also every one of us; we and our loved ones are affected by large-scale clinical and epidemiological research every day. To the 2 in 5 people who could develop cancer or the roughly 6 in 10 people over 44 who could experience a cardiovascular event within their lifetime – their treatment and prevention-focused care plans are informed by key studies and datasets supported by federal research funding. To the almost 4 in 10 people covered by at least one of the federal healthcare plans, anonymized Medicare, Medicaid and Veterans Affairs data provide the basis for future population health studies that aim to help understand and improve people’s health and care.

Aside from communicating with our elected legislators, we can all advocate for data integrity by discussing these issues with friends, family and colleagues or by volunteering to participate in research. Ultimately, we emphasize the need to build and maintain trust and respect among individuals whose experiences are represented by data, clinicians caring for patients and physician-scientists who use patients’ data.

By Emily L. Graul, M.D./Ph.D. student at Emory University in the Medical Scientist Training Program, and Dr. Christopher K. Wong, internal medicine and pediatrics resident at Baylor College of Medicine. 

Any opinions, conclusions, and recommendations expressed in this article are those of the authors and do not represent the views of Baylor College of Medicine.