Design: retrospective registry cohort with time-varying exposure, inverse probability of treatment weighting and restricted mean survival time
Population: 10,660 patients with systemic sclerosis in EUSTAR, 8,637 (81.0%) proton pump inhibitor-exposed, mean follow-up 5.7 years
Primary outcome: all-cause mortality and interstitial lung disease progression by lung function
Effect: mortality hazard ratio 1.88 (95% CI 1.40 to 2.54) with minimal 5-year restricted mean survival difference; no reduction in clinically meaningful forced vital capacity decline

Ten thousand six hundred and sixty patients in the EUSTAR systemic sclerosis registry with at least two visits were analysed, 8,637 (81.0%) of them exposed to a proton pump inhibitor over a mean follow-up of 5.7 ± 4.3 years. Exposure was modelled as time-varying, with inverse probability of treatment weighting to handle confounding by indication, and restricted mean survival time used to express absolute effects.

The exposed patients had more severe multisystem disease — more gastrointestinal, pulmonary and vascular involvement. After adjustment, proton pump inhibitor exposure was associated with higher all-cause mortality, hazard ratio 1.88 (95% CI 1.40 to 2.54), but the restricted mean survival time difference at five years was minimal, and the authors note that with cause-specific mortality unavailable this most likely reflects disease severity rather than drug harm. For interstitial lung disease, there was no reduction in clinically meaningful forced vital capacity decline; a modest attenuation of carbon monoxide diffusing capacity decline was seen, with small absolute differences.

Both halves of that need stating plainly, because both are misread. The mortality association is almost certainly confounding by indication that weighting could not remove — patients get a proton pump inhibitor because their disease is worse — and the minimal absolute survival difference is the clue. It is not a reason to stop the drug in a patient with reflux.

The other half is that the hope of a disease-modifying effect on lung fibrosis, through reducing microaspiration, does not hold up. Forced vital capacity decline was unaffected. Proton pump inhibitors in scleroderma are a symptomatic treatment for a symptom that is often miserable, and that is the whole of their justification. The authors' own conclusion is that the findings favour continued generalised use for symptomatic reflux.

Continue proton pump inhibitors for symptomatic reflux in systemic sclerosis; the mortality signal is confounding by severity
Stop justifying them as protection against interstitial lung disease progression — forced vital capacity decline was unaffected
Use the indication to set the dose: treat the reflux, do not escalate for lung reasons
Review the usual long-term proton pump inhibitor considerations — magnesium, B12, bone, enteric infection — since these patients take them for years
Treat a patient newly started on one as a marker of more severe disease, and look for what prompted it

Don't overread it

The mortality association almost certainly reflects who gets prescribed a proton pump inhibitor rather than what the drug does.

The statistics, in plain English

A hazard ratio of 1.88 with a minimal restricted mean survival time difference at five years is the pattern of confounding, not of harm: a large relative effect that translates into almost no absolute difference usually means the groups differ in ways the model has not captured. Inverse probability of treatment weighting can only adjust for what was measured, and disease severity in scleroderma is multidimensional. The absence of an effect on forced vital capacity decline, in 10,660 patients, is a well-powered negative finding and the more trustworthy half of the paper.