Reviews/Anti CGRP Antibodies/Real-world CGRP antibody effectiveness
Real World · CGRP-RW-02 · v2
Borderline: within 4 points of the Likely/Uncertain boundary, so a re-grade could land either side.
Many independent studies followed migraine patients given anti-CGRP antibodies in ordinary clinical practice, and most found the drugs clearly reduced headache and migraine days, echoing what the original trials reported. Some investigators said so directly, describing their results as similar to trials [8], confirming trial findings [33], or even better than trials [3]. But the real-world studies had no comparison group taking a dummy treatment, whereas the trial results subtract out the response seen on placebo [23], so the two are not measuring quite the same thing. The real-world patients were also generally harder to treat, having already failed several other preventives [2][8], and how many patients benefited varied widely from study to study [17][30][3]. No study in this set lined a real-world group up directly against its matching trial. So the panel accepted that effectiveness points the same way as the trials but concluded the evidence does not show the results closely match in size.
A large, consistent body of uncontrolled before-after cohorts across many countries and all the licensed antibodies shows substantial reductions in migraine and headache days in routine care [2][4][7][15][16][32], with several teams framing their findings as concordant with trials [8][33] or designed to extend them [13]. Investigators in [3] reported the real-world effect as seeming more effective than in RCTs, with ≥50% responder rates of 76.5% in high-frequency episodic and 63.5% in chronic migraine. Against this, the randomised benchmark in the corpus [23] reports a placebo-subtracted reduction of 2.81 migraine days, a different quantity from the raw change-from-baseline the real-world cohorts report. Responder rates in the real-world cohorts range from 33.7% at three months [17] and a minority sustained long-term [30] up to the higher rates in [3] and [12]. The contested point: The division is between direction and magnitude: the cohorts agree the drugs work in practice, but whether their uncontrolled reductions genuinely 'mirror' placebo-subtracted trial effects is unsettled [23]. The dispersion of responder rates, from 33.7% [17] to the 60-77% range [3], and the absence of any head-to-head real-world-versus-trial comparison leave the quantitative match unestablished.
All 33 papers behind this claim with their DOIs and what each does to it, the panel’s arbitration, the evidence gap map and the open research questions. Free, and sent to your inbox.
We’ll also send you other claims from Anti CGRP Antibodies, one at a time, and tell you when a release or a pivotal trial changes a grade. Unsubscribe by replying.
One reviewer's only job is to attack the finding. The challenges below were weighed before the grade was set.
Metric mismatch: every real-world study here is an uncontrolled before-after cohort with no placebo arm, whereas RCT 'outcomes' are placebo-subtracted differences (e.g. ~2.8 MMD in [23]). Large RW reductions (−7 to −12 days) embed placebo response, regression to the mean and natural history, so equating them with trial effects to claim 'mirroring' compares different quantities.
Population non-equivalence: RW cohorts are systematically more refractory than pivotal trials (e.g. [8] failed five orals plus BoNT-A; [2] mean 8.4 prior failures; high medication-overuse prevalence). Concordance claimed across non-matched populations is an inferential overreach.
Where it remains uncertain. Real-world studies are uncontrolled before-after designs in more-refractory patients, so their reductions are not the same quantity as the placebo-subtracted trial benchmark [23], and responder rates span 33.7% [17] to over 60% [3].
Aleatory
≥50% responder rates vary widely across cohorts, from 33.7% at three months [17] and a minority long-term [30] to 63.5-76.5% [3], reflecting genuine patient-to-patient and setting variability.
Epistemic
The corpus contains no study that directly and quantitatively compares a real-world cohort with its corresponding trial; concordance rests on investigators' qualitative conclusions [8][33].
Inferential
Uncontrolled change-from-baseline reductions are inferred to equal placebo-subtracted trial effects [23], but they embed placebo response, regression to the mean and natural history.
The load-bearing papers from a corpus of 33, each classified by what it does to the claim. Every DOI is taken from the corpus the claim was graded on.
Direct author statement of RW-RCT concordance for erenumab in refractory CM (N=139).
“In this treatment-refractory CM population, erenumab showed efficacy in a real-world setting similar to data from clinical trials.”
Panel auditInternal validity: low · scope of inference: Erenumab effectiveness in chronic migraine that failed five oral prophylactics plus onabotulinumtoxinA.Key limitations: retrospective · no concurrent comparator · qualitative trial comparison
52-week RW study (N=300) framed as extending trial findings; 71.4% achieved ≥30% MMD reduction at weeks 9–12.
“To extend the results from clinical trials, we assessed the real-world efficacy and safety of erenumab in patients with chronic migraine from the outpatient clinic at the Danish Headache Center.”
Panel auditInternal validity: medium · scope of inference: Long-term erenumab effectiveness and safety in chronic migraine at a Danish centre.Key limitations: no comparator · high 52-week attrition · single centre
RW responder rate (~33.7%) at the lower end relative to pivotal trials — tempers strict 'mirroring'.
“One-third of patients with EM and CM achieved ≥50% MMD reduction after 3 months of erenumab treatment.”
Panel auditInternal validity: low · scope of inference: Erenumab real-world effectiveness in Canadian patients with episodic or chronic migraine and prior treatment failures.Key limitations: open-label · no control group · small sample
RW galcanezumab cohort (N=163) reporting effect that exceeds rather than mirrors RCTs — concordance in direction, not magnitude.
“Galcanezumab in real life was safe, well tolerated and seemed more effective than in RCTs.”
Panel auditInternal validity: low · scope of inference: Galcanezumab effectiveness in high-frequency episodic and chronic migraine in Italian headache centres.Key limitations: no concurrent trial comparator · narrative not tested comparison · observational
4 of 33 papers shown. The full review lists all of them, with the panel’s arbitration →
Claims are graded at each quarterly release, and re-graded between releases only when a pivotal trial lands. Every change is a new version of the claim.
The corpus consistently confirms effectiveness in the direction of trial findings, but methodological indirectness, population mismatch and response-rate heterogeneity prevent a higher-confidence verdict that real-world results truly mirror placebo-controlled trial outcomes.
Release 2026-q4 re-graded every claim from its corpus under protocol v0.19. An earlier release's grade is shown for the record; it came from an earlier protocol and is not a movement.
Assumptions that carry weight but are not themselves established.
The panel records the gap that would most move the claim, and the study that would close it.
No placebo-anchored real-world comparison. Without a placebo arm the real-world reductions cannot be converted into the placebo-subtracted quantity the trials report [23], so magnitude concordance cannot be tested. The panel rates this the highest-importance open gap for the claim, and names the study design that would close it: A real-world study incorporating a concurrent or historical placebo-equivalent control, or a formal placebo-adjustment model applied to pooled cohorts.
Open question: When placebo response is accounted for, does the real-world effect size match the placebo-subtracted trial effect? Why it is still open: Every real-world study is uncontrolled before-after [2][4][7], while the trial benchmark is placebo-subtracted [23], so the two quantities have never been equated within the corpus.
Priors is a sceptical first reader, not peer review, and not medical advice.
Release 2026-q4 · graded 21 Oct 2026 · protocol v0.19