More Than a Code: Getting the Epidemiology of Fibrotic Interstitial Lung Disease Right

Fibrotic interstitial lung diseases (ILDs) are progressive, life-shortening conditions associated with substantial morbidity and limited treatment options (1). Given their clinical severity and diagnostic complexity, fibrotic ILDs remain difficult to diagnose and manage. This complexity has also hindered the design of large-scale, methodologically rigorous epidemiologic studies, which are essential to inform clinical care and guide future research. Prior estimates of incidence and prevalence have varied widely across settings, often because of nonstandardized case definitions and frequent misclassification of ILD subtypes (2, 3).

Although epidemiologic research has lagged, clinical practice has moved rapidly toward greater diagnostic precision. Advances in molecular phenotyping and the emergence of omics-driven classification schemes have begun to reshape our understanding of fibrotic lung disease, whereas the recognition of progressive pulmonary fibrosis as a distinct clinical phenotype reflects the shift toward more nuanced disease stratification (1). Yet, population-level data remain poorly aligned with these evolving frameworks. Combined pulmonary fibrosis and emphysema (CPFE) exemplifies this disconnect: a syndrome increasingly recognized in clinical practice but still inadequately defined in epidemiologic research. Patients with CPFE face particularly poor outcomes and present unique diagnostic and therapeutic challenges (4); yet, fundamental questions about its burden and natural history remain unanswered. These gaps underscore the need for robust, population-based studies to better characterize the full spectrum of fibrotic ILDs and inform future research and care delivery.

Administrative health data, particularly diagnostic codes from the International Classification of Diseases, 9th and 10th editions (ICD-9 and ICD-10), represent a powerful resource for addressing these knowledge gaps. When carefully validated, such data enable population-based analyses that can provide valuable insights into disease incidence, geographic variation, and longitudinal trends in diagnosis and outcomes, complementing clinical registries and broadening our understanding of fibrotic ILDs, particularly when prospective studies are impractical. However, the reliability of claims-based research in fibrotic ILD is limited by variability in coding practices and uncertainty surrounding the accuracy of diagnostic algorithms (2, 3, 5). These challenges highlight the need for standardized case definitions and robust validation strategies to strengthen the precision of administrative data in ILD research.

In this issue and the previous issue of AnnalsATS, two well-conducted studies take important steps toward addressing these limitations. Although methodologically distinct, both leverage ICD-coded administrative data to characterize disease burden and outcomes in fibrotic ILD. Together, they illustrate both the promise and ongoing pitfalls of using administrative data in ILD epidemiology.

The first study, conducted by Johannson and colleagues (pp. 1009–1017), used a nested case-control design with a large administrative dataset to examine the incidence, prevalence, and mortality of fibrotic ILD in Alberta, Canada, a province with a universal health insurance system (6). They first established an ILD cohort within the Canadian Registry for Pulmonary Fibrosis using clinically confirmed cases as the disease reference standard. Age- and sex-matched control subjects without disease were randomly selected from the general population. To identify an optimal case definition, the authors tested three ICD-based definitions with varying diagnostic stringency against the reference cohort. The “broad” definition (requiring one or more ICD codes or claims for ILD) demonstrated the best performance characteristics, balancing sensitivity and specificity. Applying this definition, the authors reported an age-standardized ILD incidence of 107.9 per 100,000 and a point prevalence of 789.7 per 100,000 in 2019 among at-risk adults, estimates that are substantially higher than previous studies from other countries (2). Although prevalence increased significantly over the study period, the authors observed a decline in all-cause mortality for fibrotic ILDs and the idiopathic pulmonary fibrosis subgroup between 2011 and 2018. One possible explanation for the observation of improved survival is related to the introduction of the antifibrotic medications during this period, although the survival benefit of these treatments continues to be debated (7).

Although Johannson and colleagues used administrative data to define the broader burden of fibrotic ILD in a general provincial population, in this issue of AnnalsATS, Guidot and colleagues (pp. 1493–1503) applied a similar approach to characterize the specific fibrotic ILD subtype of CPFE within a high-risk veteran population (8). Veterans experience a higher burden of fibrotic lung disease than the general population, a disparity likely driven by demographic and behavioral factors such as older age, high smoking prevalence, and male predominance (9, 10), although emerging data also suggest their unique military exposures may also play a role (11). Drawing on the rich U.S. Department of Veterans Affairs clinical database, the authors implemented a structured case identification approach based on ICD-9 codes, followed by detailed chart review of clinical notes, computed tomography reports, and pulmonary function test data. A subset of computed tomography images was also independently reviewed by a chest radiologist, demonstrating strong agreement with the original reports. The authors observed an annual CPFE prevalence ranging from 71 to 100 per 100,000 veterans and an incidence from 16 to 30 per 100,000. Affected individuals were predominantly male, had high smoking rates, and showed a characteristic pulmonary function test pattern of preserved forced vital capacity with reduced diffusing capacity of the lung for carbon monoxide, consistent with the “pseudonormalized” pattern described in prior literature (4).

Taken together, both studies deserve recognition for their rigorous and innovative use of administrative data to identify patients with fibrotic ILD and its subtypes. Johannson and colleagues developed a well-structured algorithm within a nested case-control study design, enhancing internal validity and establishing a novel framework for large-scale epidemiologic research. Their systematic validation of three ICD-based definitions, reporting sensitivity, specificity, positive predictive value, and negative predictive value, strengthens confidence in their estimates and should serve as a model for future studies using ICD codes. Meanwhile, Guidot and colleagues advanced case identification for an understudied ILD subtype by integrating imaging and physiologic data, another methodology that should become standard practice in future epidemiologic research when feasible.

Despite their strengths, these studies also remind us of the inherent limitations of ICD-based cohort studies in fibrotic ILD research. Both were conducted in populations essentially covered by universal health insurance systems and reported incidence and prevalence estimates at the higher end of previously published ranges, raising questions about their generalizability to more heterogeneous populations, particularly because both were regional rather than national in scope. Misclassification remains a central concern, as highlighted in the large number of CPFE cases excluded (27%) after careful chart review. Prior studies reinforce this challenge: Vu and colleagues found that 90% of patients assigned an ICD code for postinflammatory pulmonary fibrosis did not, in fact, have a true ILD (12), and Ley and colleagues reported a positive predictive value of only 42.2% for an ICD-based IPF algorithm (13). These findings highlight the disconnect that can exist between coded diagnoses and clinical reality, particularly for complex diseases such as fibrotic ILDs.

In summary, both studies provide important contributions to the evolving field of ILD epidemiology, offering valuable insights and methodologic models for future research. As the use of administrative data continues to grow, routine validation of ICD-based algorithms must become standard practice, ideally incorporating linkage to electronic medical records when feasible and transparent reporting of diagnostic performance metrics. Reliable estimates of incidence and prevalence (particularly for understudied subtypes such as CPFE) also require harmonized case definitions and greater consistency across studies. To advance the field, the development and widespread adoption of standardized guidelines for ILD cohort construction and epidemiologic research that incorporate these best practices are not merely timely but essential.