Brief Report: Autism Spectrum Disorder: A Comprehensive Survey of Randomized Controlled Trials.

Tromans et al. (2018) counted every autism RCT they could find. They ended up with 529 trials. They looked at how big each study was and how often teams worked together.

The median trial had only 36 participants. Most were single-site studies instead of multi-center projects.

Tiny trials still rule autism research. A study with 36 people can’t answer big clinical questions. The authors urge clinics and universities to pool patients and run fewer, larger trials.

Provenzani et al. (2020) dug into the same RCT pile and found another problem: 327 different outcome measures. When every lab uses its own ruler, you can’t line up the results.

Lord et al. (2005) warned about weak methods years earlier. Samuel’s numbers show the field still hasn’t fixed the sample-size issue.

LaPoint et al. (2025) now push mandatory trial registration. Their plan attacks the same quality gaps Samuel exposed, turning complaint into policy.

If you sit on a grant panel or IRB, insist on multi-site design and a core outcome set. You’ll save money and produce answers clinicians can actually use.