Appropriate Comparator Use for EMA-Approved Drugs Fell From 2011 to 2024

Only 58% of new drug indications approved by the European Medicines Agency (EMA) between 2011 and 2024 had at least one trial population tested against an appropriate comparator, according to a retrospective cohort study in eClinicalMedicine. The share fell from 78% of indications approved in 2011 to 2012 to 39% of those approved in 2023 to 2024.

The paper reports a shrinking share of drugs tested against an appropriate comparator and a rising share of single-arm trials.

How the study measured comparators

The team identified 487 drugs (524 indications) approved by the EMA between 1 January 2011 and 31 December 2024. After exclusions, 301 drugs (321 indications, 649 populations) remained. The reference point was Germany’s Federal Joint Committee (G-BA), which names the appropriate comparator for each population during benefit assessment. That comparator is generally an established therapy authorised and reimbursed in Germany, and best supportive care or watchful waiting can also qualify. The G-BA’s dupilumab decision for chronic spontaneous urticaria named omalizumab with second-generation H1 antihistamines as the appropriate comparator. Comparators can also vary by subgroup: the G-BA’s tislelizumab assessment treats docetaxel as appropriate only for patients with PD-L1 negative tumors.

Ad

Each population was sorted into one of four groups. A trial counted as using an appropriate comparator if its control arm was on the G-BA list for that population. A trial whose control arm was not on that list counted as inappropriate. A trial with no control arm was single-arm. A population with no patients in the submitted data was recorded as no data. A drug indication counted as appropriate if at least one of its populations met that standard. The authors used G-BA reports published up to 31 October 2025.

The 186 excluded drugs had no comparator determination in the G-BA record. That covered 107 orphan drugs (115 indications), seven reserve antibiotics (11 indications), 64 drugs (67 indications) with no assessment, six drugs (seven indications) whose manufacturer submitted no data, and two drugs (three indications) that were exempt. Orphan indications entered the cohort only when the G-BA later assessed them, which happens when annual revenue exceeds EUR 30 million (EUR 50 million before November 2022) or orphan status lapses.

Appropriate comparator use by approval period

The share with at least one appropriate comparator fell from 25 of 32 indications (78%) approved in 2011 to 2012 to 14 of 36 (39%) approved in 2023 to 2024. Over the same period, the share with only inappropriate comparators rose from 19% to 36%, and single-arm designs rose from 3% to 25%.

Source: Glaus et al., eClinicalMedicine (November 2026), Results section, proportions by approval period.

After adjusting for therapeutic area and approval pathway, indications approved in 2023 to 2024 had 76% lower estimated odds of at least one appropriate comparator than those approved in 2011 to 2012 (p = 0.01). Their estimated odds of single-arm testing were 7.5 times higher (p = 0.04). Inappropriate comparators came with 2 times higher odds, but the p-value of 0.13 does not meet the study’s threshold.

Differences across therapeutic areas

Respiratory and sensory organ indications all had an appropriate comparator in at least one population (8 of 8 and 5 of 5). Anti-infectives, immunomodulators and genito-urinary drugs followed, while dermatological drugs were lowest at 29% (2 of 7). Cancer drugs, the largest group with 123 indications, reached 52%.

Source: Glaus et al., eClinicalMedicine (November 2026), Results section and Table 1 (indication counts per therapeutic area in parentheses).

Differences also appeared within areas. All eight prostate cancer drugs had an appropriate comparator in at least one population, against seven of 20 non-small cell lung cancer drugs and one of 14 lymphoma drugs. The results text gives no figure for alimentary and metabolism drugs (28 indications).

Standard and conditional pathways

Standard approvals were more likely to have an appropriate comparator (152 of 246, 62%) than conditional marketing authorisations (11 of 44, 25%), a difference of 37 percentage points (95% CI 22 to 51). Single-arm designs made up 7% of standard approvals (16 of 246) and 68% of conditional approvals (30 of 44). Accelerated assessment reached 78% (21 of 27) for appropriate comparators and 7% (2 of 27) for single-arm designs. Standard approvals sat 16 percentage points below accelerated assessments on appropriate comparators, with a 95% confidence interval from 34 points below to 2 points above. That interval includes zero. A related Syenza News review of EMA critical comments on patient-reported outcomes found that design concerns often cited open-label or single-arm structures as weakening causal attribution.

Why the trend may be happening

The G-BA names the comparator during benefit assessment, which usually happens years after a trial starts. The authors say it is possible that the comparator the G-BA later defined was not yet an established therapy when the trial began. They say a separate analysis is needed to test this, and that the reasons manufacturers choose comparators still need evaluating. Some HTA bodies rely on real-world evidence, but the authors note those sources do not always provide sufficient data in quantity and quality.

Source: Glaus CEG, Herrle C, Hediger S, Vokinger KN. The use of appropriate, inappropriate and no comparators in efficacy trials for European Medicines Agency-approved drugs 2011-2024: a retrospective cohort study. eClinicalMedicine, November 2026, volume 101, article 104319. Full text: doi:10.1016/j.eclinm.2026.104319.