Benefit of High-flow Nasal Cannula on Persistent Dyspnea in Interstitial Lung Disease: Randomized Multicentric Cross-over Trial

The study will be a non-blinded, two-arm, crossover (2 periods of 2 weeks each), randomised controlled superiority multicentre trial comparing the effect of HFNT on persistent dyspnoea in intersititial lung disease patients with chronic respiratory failure as compared to LTOT over a two-week period. The oxygen flow rate with HFNT will be the same as for usual care. The HFNT will be used during the night and during the 3-minute chair rise test, it can be used during the day depending on the patient's needs. During the rest of the time, the LTOT will be used as in usual care. Participants will be randomised to receive (1) 2 weeks of LTOT then 2 weeks of LTOT+HFNT or (2) 2 weeks of LTOT+HFNT then 2 weeks of LTOT. The investigators will determine the effect of HFNT on quality of life and dyspnoea. The investigators will also study the effect of HFNT on secondary outcomes listed below. Each assessment will be performed at the end of each 2-week period. Polysomnography will be optional. The study will be conducted in 42 patients with LTOT suffering from persistent breathlessness.

With a total sample of 38 patients and a crossover design, if the real difference on the Saint George's Respiratory Questionnaire (SGRQ) is 6, the standard deviation of the matched difference is 12.5 and the significance threshold is 5%, a two-tailed Student's t test will have a power of 80% to conclude that the difference is significantly different from 0 (Calculated using PASS 14.0.14, Analysis of a cross-over design using difference).These hypotheses are based on the values observed in the article by Nagata et al. (PMID: 29283682), in patients with Chronic Obstructive Pulmonary Diseaes of the same severity, which reported an improvement in the SGRQ-s of -10.8 (95% CI: -15.3; -6.3, i.e. SD of 12.4) with HFNT administered at night for 6 weeks versus LTOT Another study (PMID: 31308647) reported an improvement of -11.9 (CI95% -17.2; -6.6) after an introduction of HFNT. As the minicmal clinical improvement difference for SGRQ is 4, a difference of 6 is considered was both realistic and clinically relevant. A Student's t test allows a conservative approach compared to the use of a linear mixed-effects model retained for the analysis of the primary endpoint.

To take account of reduced precision due to possible loss of follow-up or study withdrawals, the sample was increased by 10%, i.e. 42 patients: 21 in the HFNT+LTOT then LTOT sequence and 21 in the LTOT then HFNT+LTOT sequence.