Association of Particulate Matter Exposure With Lung Function and Mortality Among Patients With Fibrotic Interstitial Lung Disease

Abstract

ImportanceParticulate matter 2.5 μm or less in diameter (PM2.5) is associated with adverse outcomes for patients with idiopathic pulmonary fibrosis, but its association with other fibrotic interstitial lung diseases (fILDs) and the association of PM2.5 composition with adverse outcomes remain unclear.ObjectiveTo investigate the association of PM2.5 exposure with mortality and lung function among patients with fILD.Design, Setting, and ParticipantsIn this multicenter, international, prospective cohort study, patients were enrolled in the Simmons Center for Interstitial Lung Disease Registry at the University of Pittsburgh in Pittsburgh, Pennsylvania; 42 sites of the Pulmonary Fibrosis Foundation Registry; and 8 sites of the Canadian Registry for Pulmonary Fibrosis. A total of 6683 patients with fILD were included (Simmons, 1424; Pulmonary Fibrosis Foundation, 1870; and Canadian Registry for Pulmonary Fibrosis, 3389). Data were analyzed from June 1, 2021, to August 2, 2022.ExposuresExposure to PM2.5 and its constituents was estimated with hybrid models, combining satellite-derived aerosol optical depth with chemical transport models and ground-based PM2.5 measurements.Main Outcomes and MeasuresMultivariable linear regression was used to test associations of exposures 5 years before enrollment with baseline forced vital capacity and diffusion capacity for carbon monoxide. Multivariable Cox models were used to test associations of exposure in the 5 years before censoring with mortality, and linear mixed models were used to test associations of exposure with a decrease in lung function. Multiconstituent analyses were performed with quantile-based g-computation. Cohort effect estimates were meta-analyzed. Models were adjusted for age, sex, smoking history, race, a socioeconomic variable, and site (only for Pulmonary Fibrosis Foundation and Canadian Registry for Pulmonary Fibrosis cohorts).ResultsMedian follow-up across the 3 cohorts was 2.9 years (IQR, 1.5-4.5 years), with death for 28% of patients and lung transplant for 10% of patients. Of the 6683 patients in the cohort, 3653 were men (55%), 205 were Black (3.1%), and 5609 were White (84.0%). Median (IQR) age at enrollment across all cohorts was 66 (58-73) years. A PM2.5 exposure of 8 μg/m3 or more was associated with a hazard ratio for mortality of 4.40 (95% CI, 3.51-5.51) in the Simmons cohort, 1.71 (95% CI, 1.32-2.21) in the Pulmonary Fibrosis Foundation cohort, and 1.45 (95% CI, 1.18-1.79) in the Canadian Registry for Pulmonary Fibrosis cohort. Increasing exposure to sulfate, nitrate, and ammonium PM2.5 constituents was associated with increased mortality across all cohorts, and multiconstituent models demonstrated that these constituents tended to be associated with the most adverse outcomes with regard to mortality and baseline lung function. Meta-analyses revealed consistent associations of exposure to sulfate and ammonium with mortality and with the rate of decrease in forced vital capacity and diffusion capacity of carbon monoxide and an association of increasing levels of PM2.5 multiconstituent mixture with all outcomes.Conclusions and RelevanceThis cohort study found that exposure to PM2.5 was associated with baseline severity, disease progression, and mortality among patients with fILD and that sulfate, ammonium, and nitrate constituents were associated with the most harm, highlighting the need for reductions in human-derived sources of pollution.

Notice of Retraction and Replacement. Goobie et al. Association of Particulate Matter Exposure With Lung Function and Mortality Among Patients With Fibrotic Interstitial Lung Disease. JAMA Intern Med. 2022;182(12):1248-1259

“…To the Editor On behalf of our coauthors, we write to explain errors in the article “Association of Particulate Matter Exposure With Lung Function and Mortality Among Patients With Fibrotic Interstitial Lung Disease” that was published online first on October 17, 2022, and in the December 2022 issue of JAMA Internal Medicine , which required a reanalysis. The study investigated the association of exposure of particulate matter 2.5 μm or less in diameter (PM 2.5 ) with mortality and lung function among patients with fibrotic interstitial lung disease in 3 North American cohorts: Simmons Center for Interstitial Lung Disease Registry at the University of Pittsburgh in Pittsburgh, Pennsylvania (Simmons cohort); 42 sites of the Pulmonary Fibrosis Foundation Registry (PFF cohort); and 8 sites of the Canadian Registry for Pulmonary Fibrosis (CARE-PF cohort).…”

“…This reanalysis was prompted by discussions with an environmental epidemiology colleague, who after publication raised the possibility of immortal time bias affecting the results with our original 5-year exposure average anchored to the date of censoring . Our reanalysis contains a new exposure approach for all mortality or transplant outcome models, which incorporates a 5-year time-varying exposure for each patient enrolled in the study from the time of enrollment until censoring.…”

“…Our original finding that a PM 2.5 exposure of 8 μg/m 3 or more was associated with statistically significant hazard ratios for mortality or transplant in all 3 cohorts is no longer correct. With the reanalysis, we now find this association in the Simmons cohort, but not the other 2 cohorts . The new meta-analysis of the combined 3 cohorts continues to show a statistically significant hazard ratio for mortality or transplant for each 1-μg/m 3 increase in 5-year time-varying PM 2.5 .…”

“…As a result of the new analyses and corrected findings, our original article has been retracted and replaced with new analyses and corrections affecting the Abstract, text of main article, Tables, Figures, and Supplement 1. Supplements 2 and 3 have been added to show the original article with errors highlighted and the corrected article with corrections highlighted.…”

“…The time-varying approach was implemented by assigning each patient’s enrollment date as “day 0” and censorship date as the number of days following enrollment (eg, someone enrolled for 5 years would have a censor date on “day 1826”) . We subsequently produced multiple 5-year exposure assignments for each patient, whereby for each 30-day period a patient was enrolled in the study, there was an exposure assignment based on the average of PM 2.5 and of each constituent exposure in the 5 years preceding that 30-day segment, so someone enrolled for 5 years (1826 days) would have 60 rows for PM 2.5 exposure and an additional 60 for each unique constituent exposure assignment.…”