Mortality and diagnostic practice variation in interstitial lung disease admissions: insights from a multicentre UK cohort study

Background Interstitial lung diseases (ILDs) are a heterogeneous group of often progressive, unpredictable diseases. They frequently result in hospitalisations secondary to respiratory decompensation, termed ILD-related admissions. A proportion are due to an acute exacerbation of ILD (AEILD). All are associated with high mortality but are poorly characterised in real-world populations. Aim To evaluate mortality outcomes and associated risk factors following ILD-related hospital admissions, including AEILD. Methods We conducted a multicentre retrospective cohort study of primary International Classification of Diseases Version 10 coded admissions for ILD between 1 January 2017 and 31 December 2019 across 11 NHS hospitals in the North West of England. AEILD events were classified using clinical criteria: <30-day respiratory deterioration not secondary to cardiac failure, pulmonary embolism or pneumothorax. The AEILD subgroup was divided into those with CT confirmation (definite AEILD) and without CT confirmation (suspected AEILD). Primary outcome was time from admission to death. Statistical analyses included Kaplan-Meier and multivariate proportional hazards modelling. Results Of 938 ILD-related admissions, 54.5% met study AEILD criteria. Overall, cumulative all-cause mortality to 90-days post-discharge was 40.2%. For the AEILD cohort, cumulative all-cause mortality to 90-days post-discharge was 47.6%. Median survival of the AEILD cohort was 107 days (95% CI 87.0 to 141.0 days) and the other ILD-related admission cohort 241.0 days (95% CI 208.0 to 308.0 days), with a statistically significant difference in survival (p<0.0001). 37.6% (192/511) of AEILD events had CT confirmation. Within the AEILD subgroup, median survival was higher in the CT group (144 days vs 100 days, p=0.027). AEILD was independently associated with mortality in a multivariate model. Preadmission oxygen, age and neutrophilia were associated with mortality in both ILD-admission and AEILD 90-day all-cause mortality models. 13.9% of admissions had documented palliative care input. Conclusions Mortality associated with ILD-related admissions is high, with AEILD events independently associated with mortality. Findings highlight the need for improved education, access to palliative care and targeted AEILD research.