VA Big Data Science: A Model for Improved National Pandemic Response Present and Future
Background: The US Department of Veterans Affairs (VA) enterprise approach to research (VA Research) has built a data-sharing framework available to all research teams within VA. Combined with robust analytic systems and tools available for investigators, VA Research has produced actionable results during the COVID-19 pandemic. Big data science techniques applied to VA’s health care data demonstrate that medical research can be performed quickly and judiciously during nationwide health care emergencies.
Observations: We envision a common framework of data collection, management, and surveillance implemented in partnership with other health care agencies that would capture even broader, actionable, and timely observational data on populations, while providing opportunities for enhanced collaborative research across agencies. This model should be continued and expanded through the current COVID-19 and future pandemics.
Conclusions: Extending the achievements of VA Research in the COVID-19 pandemic to date, we advocate national goals of open science by working toward a synergistic national framework of anonymized, synchronized, shared health data that would provide researchers with potent tools to combat future public health crises.
A Role for VA Research in Efficacy
The US Food and Drug Administration (FDA) Reauthorization Act of 2017 requires that manufacturers submit evidence establishing a product’s benefits (effectiveness) outweigh its risks (safety) before it can be promoted and distributed.6 As such, the FDA has been obligated by external stakeholders and Congress to be more explicit and transparent about benefit-risk profile supporting its decisions on licensure. This process led to requiring more phase 4 postmarketing observational studies for safety and effectiveness.7 Although the FDA postlicensure system remains vigilant toward safety, effectiveness information is limited due to insufficient reporting (with exceptions of manufacturer studies for new indications or to exhibit superior comparative effectiveness). The agency typically relies on a static set of efficacy data generated prelicensure with a dynamic and evolving set of safety data accrued postlicensure to support its assessment that benefits outweigh risks.
For example, operating in near real time, postauthorization safety monitoring systems, led by the Centers for Disease Control and Prevention and other federal systems, identified a safety signal for thrombosis following the Janssen COVID-19 vaccination. Distribution was quickly paused, the safety signal was investigated, the magnitude of the risk was characterized, new language describing the risk and providing guidance regarding clinical management was included in labeling, and distribution was resumed, all within a few weeks. This remarkable success demonstrated how timely the safety system can operate to evaluate risk.
In contrast, the duration and extent of protection against COVID-19 variants are largely limited to the assessment of immune biomarker surrogates. Such clinical effectiveness data are urgently needed for the FDA’s Center for Biologics Evaluation and Research and Center for Drug Evaluation and Research to make accurate benefit-risk assessments and continue to conclude the balance is favorable. As we prepare for the next pandemic, we must consider plans for monitoring postauthorization/postlicensure effectiveness as well as safety in real time. VA Research is ideally situated for this task.
Published studies on effectiveness at the VA serve as a prototype and could lead the way to initiating those preparations.4,8-11 One of the striking features of the VA system that became apparent in the preparation of the mRNA vaccine study was the speed at which an enormous volume of COVID-19 testing data were produced. This enabled implementation of methodologically sound test-negative and case-control analysis. Analyses sufficiently powered to conclude mRNA vaccines were highly effective when used in real-world conditions among a diverse population from nearly every state and territory during a period in which multiple COVID-19 variants were already circulating.3 This is unique to the VA and would not be possible for any other US health care system. With planning, the VA system could produce product-specific, real-world evidence of effectiveness comparable to the timeliness and quality of the safety data currently produced to support regulatory benefit-risk assessments. For example, the VA conducted an effectiveness study of tixagevimab/cilgavimab for preventing COVID-19 during the initial Omicron surge, which is continually updated while Omicron circulates and repeatable for different subvariants.12
The FDA continues to collaborate with the VA on demonstration projects to evaluate the impact of available vaccines and treatment against COVID-19 variants. The VA has also initiated several large-scale sequencing programs for COVID-19 specimens that will support these efforts, including VA Science and Health Initiative to Combat Infectious and Emerging Life-Threatening Diseases (VA SHIELD), VA Sequencing for Research Clinical and Epidemiology (SeqFORCE), and VA Sequencing Collaborations United for Research and Epidemiology (SeqCURE).13,14 Successful proof-of-concept studies using these data could provide a template for VA and other medical systems/databases to report effectiveness in near real time.
