What the study found
The study found major variability in how ALS clinical trials analyzed the ALS Functional Rating Scale-Revised (ALSFRS-R), a disability rating scale for amyotrophic lateral sclerosis. The choice of statistical method could substantially change the estimated treatment effect and the reliability of trial conclusions.
Why the authors say this matters
The authors conclude that this variability can make trial results harder to interpret and compare, and can influence clinical decision-making and drug development. They suggest that statistical consensus recommendations could improve how disability scales are used in trials and help progress toward effective therapies for neurodegenerative diseases.
What the researchers tested
The researchers systematically reviewed randomized, placebo-controlled clinical trials that used ALSFRS-R as the primary endpoint, included at least 20 randomized patients, and had at least 12 weeks of follow-up. They extracted information on statistical analysis methods and missing-data handling, then used simulation based on the Ceftriaxone trial data set to assess false-positive risk and statistical power.
What worked and what didn't
They included 45 randomized clinical trials with 7,338 patients and found 39 distinct statistical methods, combining longitudinal and cross-sectional approaches. Most trials, 55.6%, did not use all available longitudinal ALSFRS-R measurements. Applying different methods to the same data produced treatment-effect estimates ranging from negative 1.33 to positive 2.33 standard-deviation differences; 38.9% of methods were at risk of increasing false-positive rates, and power among valid strategies ranged from 17.9% to 78.2%.
What to keep in mind
The simulation used the Ceftriaxone trial data set as a realistic trial scenario, so the estimates come from that modeled context. The abstract does not describe other limitations beyond the scope of the reviewed trials and the methods assessed.
Key points
- 45 randomized ALS trials were analyzed, with a total of 7,338 patients.
- The review found 39 different statistical methods for analyzing ALSFRS-R results.
- More than half of trials did not use all available longitudinal ALSFRS-R measurements.
- Different methods applied to the same data produced wide differences in estimated treatment effects.
- About 38.9% of methods were at risk of increasing false-positive rates.
- Statistical power among valid strategies ranged from 17.9% to 78.2%.
Disclosure
- Research title:
- ALS trial analyses varied widely and changed treatment estimates
- Authors:
- Daphne N. Weemering, Jordi W. J. van Unnik, Angela Genge, Leonard H. van den Berg, Ruben P. A. van Eijk
- Institutions:
- Montreal Neurological Institute and Hospital, University Medical Center Utrecht, University Medical Center Utrecht, University Medical Center Utrecht, University Medical Center Utrecht
- Publication date:
- 2026-04-21
- OpenAlex record:
- View
Get the weekly research newsletter
Stay current with peer-reviewed research without reading academic papers — one filtered digest, every Friday.