COPD stability as a target: fewer than a quarter reach it, and those who do live longer

Design

Post hoc analyses of five randomised trials with Bayesian joint modelling

Population

Patients with COPD in IMPACT, FULFIL, MATINEE, METREX and METREO

Primary outcome

Achievement of composite disease stability and subsequent exacerbations and all-cause mortality

Effect

Stability in 22 per cent at week 52 on triple therapy; stability at week 28 associated with 45.7 per cent lower subsequent exacerbation risk and 51.7 per cent lower all-cause mortality risk

Post hoc analyses of five trials - IMPACT and FULFIL comparing triple with dual inhaled therapy, and MATINEE, METREX and METREO adding mepolizumab to triple therapy - tested a composite definition of disease stability: no moderate or severe exacerbation, and no worsening from baseline in either COPD Assessment Test score or FEV1.

Stability was uncommon. Twenty-two per cent achieved it on fluticasone furoate, umeclidinium and vilanterol at 52 weeks in IMPACT and 46 per cent at 24 weeks in FULFIL, and 18 per cent on mepolizumab added to triple therapy at 52 weeks. Triple therapy achieved it more often than dual, and mepolizumab more often than placebo. The prognostic finding is the striking one: patients stable at week 28 had a 45.7 per cent lower risk of a moderate or severe exacerbation after week 28 and a 51.7 per cent lower risk of death from any cause, compared with those who were not.

Hold two things separate. That stability predicts a better subsequent course is now well supported. That achieving stability causes the better course is not shown, and cannot be by a post hoc analysis of trials designed for other endpoints - patients who go 28 weeks without exacerbating, deteriorating symptomatically or losing lung function are, unsurprisingly, the ones with less aggressive disease. The clinical use is as a review question rather than a claim: at six months, has this patient exacerbated, has the CAT score worsened, has the FEV1 fallen? Three answers of no identify someone doing well; any yes identifies someone whose treatment deserves another look.

• Ask the three stability questions at six months: exacerbation, CAT worsening, FEV1 decline.
• Record a CAT score at baseline, or the comparison at review is impossible.
• Treat instability as a prompt to review treatment and adherence, not as a failure of the patient.
• Do not present stability as a treatment effect; the association with better outcomes is prognostic.
• Note these are post hoc analyses of industry trials with a definition constructed afterwards.

The statistics, in plain English

A 45.7 per cent reduction in exacerbation risk and 51.7 per cent in all-cause mortality are relative reductions between two groups defined by how they did in the preceding 28 weeks, not by randomisation - so this is prognosis, not treatment effect. A composite endpoint defined after the data existed can be constructed to separate groups cleanly, which is why the prognostic finding is best read as confirming what the components already tell you. The Bayesian posterior probability of stability, 25.6 per cent at week 52, is consistent with the observed 22 per cent.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.