What does “non-progression” mean?
A closer look at how trials describe spinal structural change in axial spondyloarthritis.
Completed descriptive pilot Internal checks No external validation
Choose your reading view
Does “no progression” mean nothing changed?
Not necessarily. A report may count a small increase in an imaging score as “no progression”. Our first pilot asks what the phrase means in the reports we inspected, and what a reader needs to know before interpreting it.
Why it matters
How someone feels, whether new structural damage develops, and whether an existing lesion reverses are different questions. An outcome label alone does not tell us which question a study answered.
What we found
Definitions differed. In matched groups of participants, changing the cutoff changed the percentage classified as non-progressors. That is a change in classification, not a new treatment effect.
What it cannot tell us
The pilot does not rank medicines or prove structural repair. It is a limited audit of accessible reports, not a complete account of all research or personal treatment advice.
Which definition sits behind the endpoint?
We audited populations, exact inequalities, readers and missing-data rules underlying approximately two-year spinal structural outcomes in accessible axial SpA trial reports published in 2018–2026.
Design
A bounded descriptive audit of nominal 96–112-week mSASSS or syndesmophyte outcomes from randomized parent trials and related reports. Actual image windows, estimator differences and unknown denominators remain explicit.
Main observation
Four parent families supported five matched within-cohort threshold contrasts. Definitions differ across contrasts; no effect was pooled and no treatments were ranked. MEASURE-1 and ENRADAS share a report and reading campaign.
Critical limits
Selection and access bias remain possible. Registry histories, some supplements and status checks are incomplete. Excluding SELECT-AXIS-1’s inferred denominator linkage leaves three families but only two source reading exercises. No lesion-specific repair endpoint was identified in the inspected material.
- 9
- parent trial families
- 46
- endpoint records
- 5
- contrasts from 4 families
Counts describe this bounded pilot. Families are not independent methodological replications.
One finding, read carefully
A different cutoff.
The same participants.
In the MEASURE-1 comparative reading dataset, the reported rate changes from 60.7% at score change ≤0 to 82.1% at ≤2. The 21.4 percentage-point difference subtracts rounded published rates. It measures a classification difference, not a treatment benefit.
See the finding, sources and caveatsSource-linked records and reproducible arithmetic are available with the report. Internal AI checks are part of the workflow; they are not external expert validation.