This study is the first to describe post-9/11 ILD incidence, prevalence, demographics, and radiographic features in a closed occupational cohort followed longitudinally over 21 years after an intense occupational/environmental exposure (WTC). We found an age- and sex-standardized post-9/11 prevalence rate of 252.5/100,000 persons, more than two-fold greater than most reports [23, 31], despite excluding sarcoid-related ILD from our primary analyses. Most importantly, half of the ILD cases developed PPF, higher than some previous reports [24].

Incidence and prevalence data for ILD and PPF are limited, and comparisons are difficult due to methodologic differences including clinic- vs. population-based, cross-sectional vs. longitudinal, demographics (including age at diagnosis), rate standardization, diagnostic confidence (self-report, claims-based, medical record confirmation), diagnostic criteria and coding, inclusion/exclusion of specific ILD types (e.g., sarcoidosis or IPF), and follow-up time. The largest study by Olson et al. [23] included sarcoidosis – which has granulomatous inflammation as its mechanism of fibrosis, quite different from other causes of fibrosing ILD [32]. By using only claims data, all sarcoidosis cases with any type of lung involvement were included, as claims data could not identify only those with ILD. Our data show how claims data alone can influence results. Including all sarcoidosis cases with any type of lung involvement increased the age- and sex-standardized prevalence rate for ILD in our study by nearly two-fold to 444.1/100,000 persons, and for PPF by 5% to 85.0/100,000 persons (Table 2 and Online Resource 5). These differences help to explain the wide range of reported prevalence and incidence rates for ILD and PPF, with prevalence rates ranging from 6-118/100,000 persons and 3–70/100,000 persons, respectively [23, 31, 33,34,35]. Crude age-specific prevalence rates were also different than US rates post-9/11 ILD prevalence rates were 95.4/100,000 persons (age 50–59 years) and 630.9/100,000 persons (age 60–69 years) compared with the same ages in the US, 72/100,000 persons and 162/100,000 persons, respectively [36].

The majority of occupational/environmental ILD have long latency periods (often decades) between exposures and disease [37]. Therefore, it is not surprising that we found the average time between WTC-exposure and ILD diagnosis to be 13 years. With continued follow-up this average time will certainly increase.

We found older age on 9/11 predicted ILD. Also similar to other studies, smoking status was a risk factor for ILD, with ever-smokers having over 2-fold greater rate of developing ILD compared with never-smokers [38, 39]. The presence of GERD prior to ILD diagnosis increased the rate of ILD nearly 5-fold. The pathophysiological relationship between GERD and ILD remains controversial, but one possibility is that recurrent micro-aspiration leads to chronic inflammation and fibrosis [40, 41]. In a cohort without WTC-exposure, GERD was a significant contributor to ILD in those who had GERD prior to an ILD (non-IPF) diagnosis [42].

We found no published studies with lung function data prior to the development of ILD or PPF. We hypothesized that lung function prior to diagnosis would be predictive of the development of ILD. First post-WTC-exposure FVC was lower in those who developed ILD compared with the whole cohort. However, after adjusting for confounders, our analyses did not confirm lung function or change in lung function (regardless of when measured) as a significant risk factor for the development of ILD. This important finding, if confirmed in other cohorts, would indicate that lung function within the normal range, or even lung function decreases shortly after major respirable exposures, would not be a factor for identifying those at risk for the development of ILD. The absence of such an association would place an unexpected and significant limitation on the ability for clinicians to identify those who might benefit from closer surveillance and early anti-inflammatory and/or anti-fibrotic treatment after exposure or ILD diagnosis.

Many of the toxic components identified at the WTC-site have been associated with ILD in non-WTC settings [9,10,11]. Years later, WTC studies have demonstrated persistent inflammation [43, 44], a known prodrome for pathogenic fibrotic response [45]. Together with early case reports of ILD and PPF among WTC-exposed participants, our finding of an increased prevalence of ILD among FDNY-WTC-responders compared with a general population (age- and sex-standardized) is not unexpected. Yet, we were unable to identify a WTC-exposure-response gradient for the development of ILD. A prior study from a different WTC-cohort (WTC Health Registry) found a similar incidence rate (36.7/100,000 person-years), but with a WTC-exposure-response gradient [14]. The WTC Health Registry cohort includes not only rescue/recovery workers, but also those who lived and worked in the surrounding area south of Canal Street. Their study had further notable differences from ours: ILD was based on self-reports obtained from a survey without medical record confirmation, used a different definition of WTC-exposure intensity, and included different types of WTC-exposed individuals who were less likely to have intense dust cloud exposure. The inability to demonstrate a dose-response effect in our study may be due to these differences as well as the long latency period between exposure and disease, complexity of the host inflammatory response, low incidence rate for ILD in midlife populations, and other non-WTC-exposures that occur during firefighting. Individual pre-9/11 firefighting exposure data were not available; however, when we used service years (an imperfect proxy), we found no significant association with ILD or PPF.

It is difficult to directly compare survival rates between our study and prior studies due to differences in multiple factors including methodology, age at diagnosis, ethnicity, type of fibrotic lung disease, follow-up time, comorbidities, healthy worker effect, and treatment availability. The median survival of IPF has been reported to range from 2 to 5 years [46, 47]. The actual range of survival for any individual IPF patient can vary, with up to 25% of patients living beyond 10 years, especially when diagnosed earlier [48,49,50,51,52]. Following ILD diagnosis, we observed an average follow-up of 8.5 years, longer than generally reported.

A limitation of our study is that CTs were obtained only in those with clinical indications – symptoms, abnormal spirometry, or abnormal chest radiograph. This results in the inclusion of more ever-smokers. Complete incidence and prevalence rates require CT-screening of asymptomatic non-smokers, a limitation we share with other published studies. However, given the unique nature of our cohort with annual monitoring and free diagnosis/treatment, we believe few cases would have been missed. Compared with other studies, this may have led to higher incidence and prevalence rates as well as earlier diagnosis. Another limitation is that we may be missing post-diagnosis FVC measurements among participants who were too ill to produce reliable spirometry/PFTs or who died before a scheduled spirometry. This would likely result in overestimation of the average post-diagnosis lung function values and result in misclassification of some persons with PPF as non-PPF. Likewise, post-diagnosis follow-up may not be comparable to prior studies as our diagnosis date was defined as the first CT with ILD findings [15]. An additional limitation, similar to most published studies, is that CTs and PFTs were not done at a single facility, and CT interpretations were qualitative clinical readings rather than quantitative analyses. However, all were done at highly regarded facilities; the requirement for repeat confirmatory scans provided added confidence; and our process mimics real-world clinical practice. Furthermore, our definition for ILD excluded isolated interstitial lung abnormalities (i.e., isolated, unilateral, or ground glass opacities without bilateral reticular findings). Other limitations include incomplete data on underlying conditions predisposing ILD, such as connective tissue disorders, other than sarcoidosis, or family histories. Furthermore, though consistent with current guidelines, diagnosis did not rely on histopathology, which was only available for eight participants (7/8 from lung transplant specimens), and notably all confirmatory of the fibrosing ILD diagnosis by CT. Lastly, a lack of socioeconomic, racial, ethnic, and sex diversity may limit the generalizability of our findings.

Major strengths include the prospective longitudinal design and the requirement for ILD confirmation on repeat CT. The FDNY-WTC cohort is a single center cohort with pre-9/11 health data confirming that all ILD cases occurred post-9/11. The cohort has a high retention rate because of labor-management support, free monitoring and treatment, and financial compensation provided by the WTC Victims Fund [53]. Third, the ability to include over two decades of follow-up of middle-aged participants allowed adequate time for ILD development, diagnosis, and progression. Fourth, annual monitoring for symptoms and lung function changes, with a low threshold for CT and referral to in-house pulmonologists, allowed for the identification and confirmation of each case and their progression without reliance on imprecise ICD-10/CPT diagnosis/claims codes as proxies. Finally, because we are the only study with longitudinal lung function measurements obtained before and after 9/11, we were able to examine the effect of lung function, with and without the influence of 9/11 exposure, on the development of ILD. This allowed us to demonstrate that lung function in any capacity was not a significant risk factor.

Our study summarizes characteristics of those with post-9/11 ILD in the FDNY-WTC cohort. Risk factors for the development of ILD were identical to non-WTC cohort studies – age, smoking status, and GERD. Although we did not observe a significant WTC-exposure-response gradient, we observed a two-fold greater prevalence of ILD, even after excluding sarcoidosis, compared with non-WTC studies. Half the ILD cases progressed to PPF, highlighting the unique nature of WTC exposure intensity. Lung function, pre- or post-exposure, was not a predictor of developing ILD or PPF. Five years after diagnosis, those who developed PPF had greater lung function decline than ILD without PPF. While healthy-worker effects cannot be discounted, longer survival than generally reported may also be due to this no-cost program removing any financial barriers to early diagnosis and treatment and possible differences in the pathophysiology of PPF vs. IPF. Our findings indicate that continued longitudinal follow-up is necessary to assess if additional ILD cases occur with or without PPF, if CTs alone are adequate to predict final outcomes, and if survival time increases with newer treatments.