News · Longevity & Aging
An old antibiotic beat a newer drug at preventing COPD flare-ups, 17% fewer
Two maintenance drugs for chronic lung disease have never been compared head to head. An analysis built to imitate that missing trial, across 15,100 matched patients, found a clear winner.
- Patients starting azithromycin had 17% fewer serious lung flare-ups than those starting roflumilast.
- About 15 people would need treating with one instead of the other to prevent one flare-up.
- The two drugs have never been compared directly in a trial.
- Long-term antibiotic use carries costs this analysis does not weigh.
- 7,550 matched pairs from US insurance records, with nobody randomly assigned.
Chronic obstructive pulmonary disease is measured in flare-ups. Not in lung function tests, from the patient’s point of view, but in the episodes where breathing gets suddenly worse and does not recover for days or weeks.
They are the thing that lands people in hospital, and each one tends to leave lung function a little lower than before. Preventing them is most of what long-term treatment is for.
For patients who keep having them despite inhalers, two drugs are added. One is roflumilast, developed for exactly this. The other is azithromycin, an antibiotic from the 1980s taken continuously at low dose. Nobody has ever tested them against each other, so researchers writing in The BMJ built the comparison out of records.
Building the trial that was never run
The method matters more than usual here, because comparing two drugs in ordinary care is normally close to meaningless. Doctors choose, and the choice encodes what they know about the patient.
A target trial emulation attacks that by writing the protocol first. Eligibility is defined, the moment of initiation is fixed, follow-up starts there, and only then are the patients found in the data.
The analysis identified 7550 matched pairs of patients over 40 with active COPD who newly started one drug or the other, drawn from US insurance records covering thirteen years.
The result
Compared with roflumilast, azithromycin was associated with a 17% reduction in the hazard of a first moderate or severe flare-up.
Expressed the way a clinician would use it, that is a number needed to treat of 15: roughly fifteen patients given azithromycin instead of roflumilast to prevent one flare-up that would otherwise have happened.
Fifteen is a low number in preventive medicine. Statins for primary prevention sit in the hundreds. Against a background rate of 1.2 first flare-ups per person year, this is a large practical difference between two drugs that guidelines currently treat as broadly interchangeable add-ons.
The cost the comparison does not include
A 17% reduction in flare-ups is not the whole ledger, and the reason is in the drug’s own description.
Azithromycin is an antibiotic. Taking one continuously for months or years does things that a flare-up count does not capture: it selects for resistant bacteria in the person taking it and in the community around them, it can damage hearing, and it lengthens an interval in the heart’s electrical cycle in a way that occasionally matters.
Roflumilast has its own problems, mainly diarrhea, weight loss and mood effects, and they are among the reasons patients stop taking it.
Which drug wins depends on what is being counted. On the outcome measured here, azithromycin does. On antibiotic stewardship it plainly does not, and that consideration belongs to populations rather than to the patient in front of the doctor.
What COPD is doing to the lungs meanwhile
COPD makes it hard to breathe, and it is progressive: the airways and air sacs lose their elasticity, the walls between air sacs break down, and the airways make more mucus than they can clear.
Nothing available reverses that. Everything in the treatment list is about slowing the decline and preventing the acute episodes that accelerate it, which is why a difference in flare-up rates is the outcome worth arguing over.
Where the design gives way
Propensity matching balances what the records hold: age, sex, comorbidities, prior treatment, healthcare use. It cannot balance the unrecorded reasoning behind a prescription.
A patient with chronic bronchitis and heavy sputum production is the classic candidate for roflumilast, and that phenotype may also be the harder one to keep out of trouble. If prescribers were systematically sending sicker patients toward one drug, part of the gap is that, not the drug.
Insurance records also capture what was dispensed rather than what was taken, and they end when coverage does.
The honest reading
This is the best available evidence on a question that a randomized trial has never addressed, and it is not a randomized trial. The effect is large enough, and the design careful enough, that it deserves to influence the guidelines rather than sit in the literature.
What it should not do is move any individual patient, because the choice between these two drugs turns on more than the number of flare-ups, and the person best placed to weigh the rest of it is the one doing the prescribing.
People also ask
What did the study find?
Compared with roflumilast, azithromycin was associated with a 17% reduction in the hazard of a first moderate or severe COPD exacerbation (hazard ratio 0.83; 95% CI 0.79-0.87), with a number needed to treat of 15 (95% CI 12-21). The unadjusted incidence of a first exacerbation was 1.2 per person year in the cohort.
What is an exacerbation?
A flare-up: a sustained worsening of breathlessness, cough and sputum beyond normal day-to-day variation, usually needing steroids or antibiotics and sometimes hospital admission. Each one tends to leave lung function slightly worse than before.
What is a target trial emulation?
An analysis of existing records designed to imitate the trial nobody ran. Eligibility, the moment treatment starts and the follow-up window are all specified in advance as a trial protocol would, then applied to the data.
Why does it matter that they were never compared directly?
Because guidelines have to choose between them anyway. Where no head-to-head trial exists, prescribing is guided by indirect comparison and habit, and an emulation is the next best evidence available.
Why not just prescribe azithromycin then?
Because it is an antibiotic taken continuously for months or years. That carries risks of bacterial resistance, hearing loss and heart rhythm effects, none of which is the outcome measured here.
How reliable is a matched comparison?
Propensity matching balances the recorded characteristics of the two groups. It cannot balance what the records do not contain, and the reason a doctor chose one drug over the other for a particular patient is exactly the sort of thing that goes unrecorded.
Should anyone change treatment?
No. Both drugs are prescribed for specific reasons in specific patients, and the trade-offs run beyond flare-up counts. This is general information rather than medical advice.