ENO Breathe vs Usual Care in COVID-19 Recovery

BACKGROUND:

ENO Breathe is a breathing and wellbeing programme developed specifically for people recovering from COVID-19, who are still suffering from breathlessness and associated anxiety. Delivered by ENO in collaboration with Imperial College Healthcare teams entirely online, the programme focuses on breathing re-training through singing. Initial evaluation of the project suggested health improvements related to participation, however, a properly conducted efficacy study is needed.

DESIGN: An assessor blind, parallel group, randomised controlled trial, to assess the effects of ENO Breathe compared to usual care on health status. The study will also evaluate the impact on respiratory symptoms, anxiety, depression, and breathlessness perception to investigate the mechanisms involved.

RESEARCH QUESION AND OUTCOME MEASURES:

The primary outcome will be change in health status, assessed using the RAND-36 tool (RAND SF-36), comparing ENO Breathe and Usual Care arms.

Additional endpoints will be: the COPD assessment test score (CAT), 0-100 Visual Analogue Scales (VAS) for breathlessness (at rest, walking around the house, climbing stairs, and running), Dyspnoea-12, anxiety (GAD-7), and depression (PHQ-9).

Assessments: The outcomes outlined above will be recorded at baseline prior to randomisation, and then repeated after 6 weeks.

POPULATION: Adult patients recovering from COVID-19 will be recruited from specialist COVID-19 clinics, where participants will have been fully assessed and investigated by referring healthcare professionals. Participants in the active arm will attend once weekly ENO online workshop sessions for 6 weeks, and have access to bespoke online digital resources, designed to support participants between sessions. The comparison arm will receive usual care. Randomization will be 1:1.

SAMPLE SIZE Sample size: Based on pilot data in people with COPD, where the standard deviation (SD) for change in SF-36 was 15 points, to identify a clinically relevant 10 point difference in SF-36 responses at a 0.05 level of significance with a 90% power would require 48 patients in each treatment arm. Allowing for 20% dropout, we will therefore recruit 120 patients.

Primary analysis will be on an intention to treat basis. Change in parameters will be assessed comparing ENO Breathe and Usual Care, using Generalised Linear Models to estimate treatment effects. In order to better understand changes in the RAND-36 "anchor measure", the investigator will relate this to changes in measures of anxiety, depression, breathlessness, and other respiratory symptoms.

An additional responder analysis will compare the proportion in each treatment arm achieving a clinically important (10%) improvement in RAND-36 scores.

Update 21 JUNE 2021

To calculate the sample size, pilot data from a singing-based intervention in COPD, (considered the most relevant data available at study inception) was used, as outlined above, where the researchers aimed to recruit 120 patients. Data subsequently became available from ENO Breathe participants that rook part in the program prior to the research study, in which the SD was 8 for change in SF-36 scores. We have also decided using an MCID of 5 for the SF-36 is more appropriate, as this is not specific to a particular medical condition. As such, a total sample of 108 participants is required, using 1 to 1 study arm allocation. Allowing for 30% dropout, which is more appropriate, 158 participants will be recruited. Additionally, as a higher than expected number of participants were considered, study withdrawals due to be unable to attend the session times they were allocated, or despite meeting inclusion criteria, 1-to-1 meeting with the session leaders highlighted issues related excessive fatigue precipitated by exertion, so were not deemed appropriate at that point in time. As such, the researchers will also conduct a modified intention to treat analysis including all participants who were randomised and were deemed appropriate to participate following their 1-to-1. This consideration had not been included in the original exclusion criteria, due to limited awareness of its relevance at that time. These changes have been made, and here documented, before any baseline data has been sent to the analysis team, and before any follow up data has been collected.