A robust understanding of disease epidemiology, including geographic distribution, demographic, environmental, and occupational risk factors is foundational to ILD research. Understanding the population-level epidemiology of ILD has been challenging, in part because of limited generalizability of early studies, which relied on individual patient recruitment from tertiary care referral centers, oversampled severe disease, and underestimated population-level burden. Recent literature has leveraged claims data to study “real-world” epidemiology and demonstrated that the prevalence of ILD is high in the Veteran population (7). However, claims data are limited to diagnosis, procedures, medications, and costs. In contrast, EHR data can generate more nuanced “computable phenotypes” that incorporate granular structured information, such as lab test results, and unstructured data elements, such as clinical notes and imaging. These computable phenotypes can be used to study the natural history of disease and identify factors that predict outcomes. The VA has been an international leader in this field, with a robust EHR data warehouse dating back to the 1990s, and is uniquely poised to build on existing epidemiology research by leveraging the EHR to identify ILD patients for cohort development that can be used to study the population prevalence and geographic distribution of fibrotic lung diseases. Additionally, the VA’s national lung cancer screening program, which aims to conduct computed tomography scans for approximately 900,000 eligible Veterans, has the power to detect early ILD in an at-risk group. Paired with genetic and exposure data (6), this type of population-level analysis could lend important insight into novel ILD risk factors, their pathogenic mechanisms, and targets for prevention and intervention.

The treatment and disease trajectory of ILD varies. Idiopathic pulmonary fibrosis is generally considered the most severe type of ILD. However, a growing body of literature suggests that other ILDs may also progress (8). Identifying which patients are at risk for progressive disease at index diagnosis and developing a personalized approach to treatment have remained elusive because of the small sample sizes of prior cohorts with limited power for subphenotyping. Progress in this area will require integration of large quantities of clinical and “omics” data—such as genomics, proteomics, radiomics—and new machine learning algorithms for biological endotype discovery, phenotypic refinement, and clinical outcome prediction (9). The VA has the clinical and “omics” data and the sample size to support this work. Oncology research has demonstrated that this type of analysis can lead to targeted therapies and patient-level prognostication. Expanding this precision medicine approach to ILD holds the promise of identifying optimal treatment for those patients who will derive maximum benefit while withholding the same treatment from those who will not benefit or may even be harmed.

Randomized, double-blind, placebo-controlled trials are considered the gold standard for the evaluation of new therapeutics. However, traditional clinical trial methodology is costly, cumbersome, and often lacks generalizability beyond the narrowly defined eligible populations. For rare diseases such as ILD, limited clinical trial sample sizes that lack the power to detect changes in meaningful secondary endpoints pose an additional challenge. Pragmatic trials, such as the CleanUP idiopathic pulmonary fibrosis trial, have emerged as a method to more efficiently generate new evidence that is generalizable (10). Additionally, randomized, embedded, multifactorial, adaptive platforms (REMAPs) such as those that emerged during the severe acute respiratory syndrome coronavirus 2 (SARS CoV-2) pandemic can facilitate rapid evidence generation and test multiple interventions simultaneously. A global effort is currently underway to develop a REMAP program for ILD (11). EHR tools can be used for eligibility screening, recruitment, follow-up activities, and endpoint evaluation in these trials, thus decreasing inefficiencies without increasing risk to participants. The VA has already demonstrated that existing infrastructure can be used to conduct pragmatic trials in pain and posttraumatic stress disorder; it is, therefore, ideally situated to lead these multisite trials in ILD.

Physiologic metrics such as pulmonary function tests have historically been the primary endpoint in ILD research. However, a growing sentiment among physicians and patients alike recognizes that patient-reported outcome measures (PROMs) should complement physiologic endpoints (12). However, integrating PROMs, which have traditionally been measured through questionnaires, has been challenging. Engaging patients through web-based applications can be more efficient, comprehensive, and cost-effective. Additionally, data derived from home-based monitoring by means of wearable devices, activity trackers, and portable spirometry can provide objective data points to complement PROMs, identify clinically important changes occurring between visits, and improve patient engagement. The VA has already built a robust infrastructure to support mobile applications and remote monitoring technologies and has workflows that allow for real-time data sharing to ensure its effective use for patient care. During the peak of the SARS CoV-2 pandemic, Veterans logged into the VA video connect app more than 120,000 times a week, triggering a $1 billon commitment to further expand remote monitoring platforms in the VA. The VA telehealth platform’s continued success will also facilitate improved access to care by allowing Veterans living in remote areas to remain engaged in subspecialty care.

Advances in evidence do not produce health benefits for patients unless they are translated into practice. Recent studies have demonstrated that ILD diagnosis is often delayed and that uptake of novel therapeutics outside of subspecialty centers is low (13). Clinical data sets from the EHR represent a robust means by which to evaluate real-world care patterns and identify population-level gaps between what is known from research evidence and actual clinical practice (evidence-practice gap). The national scope of the VA’s healthcare system is ideally positioned to study these natural variations in healthcare delivery, uncover factors that influence quality of ILD care, and test care delivery interventions. Overcoming diagnostic delays and improving access are fundamental to the VA’s mission and will require innovations in identifying patients with ILD early through EHR algorithms (14) or new machine-learning technology. It will also require further investment in digital care infrastructure for patients living outside major academic catchment networks. Before the SARS CoV-2 pandemic, which saw rapid expansion of telehealth, the VA was already at the forefront of using digital health to transform how Veterans access high-quality care through both synchronous (e.g., video visits) and asynchronous communication (e.g., wearable devices) with care teams. Using these tools to study quality, safety, and outcomes and developing interventions to address care gaps will help bridge the current evidence-to-practice gap in ILD.