Screening for autism spectrum disorders: what is the evidence?

Jo and the team hunted for any autism screening test that works in the general population. They combed every paper they could find. No age cut-off, no country limit—just real evidence that a screener picks up kids who truly have autism.

They also looked for proof that catching autism early through screening leads to better later outcomes. Without that second piece, a test is just labels on paper.

The review came up empty. No tool had enough backing to justify screening every child. The authors also saw too little solid intervention research. Their bottom line: universal autism screening is not ready for prime time.

In short, we still lack a gold-standard screener and the follow-up science that shows screening helps kids in the long run.

Fernell et al. (2014) fired back eight years later. They argue we do not need a twenty-year wait for perfect trials. Shorter, rigorous studies can justify early screening now. This sets up an apparent contradiction: Jo et al. say “hold off,” Elisabeth et al. say “move ahead carefully.” The gap is mostly about risk tolerance, not facts.

Later big-picture reviews swallow the 2006 warning whole. Trembath et al. (2023) umbrella review and Tromans et al. (2018) survey of 529 trials both include the screening evidence gap as part of the field’s larger growing pains. Provenzani et al. (2020) show why progress is messy: 406 autism trials used 327 different outcome measures. With that much variety, validating any single screener is like hitting a moving target.

Lord et al. (2005) set the stage one year earlier by begging for stronger RCTs across autism research. Jo et al. simply applied that same demand to screening tools.

If you do intake assessments, do not treat any one screener as gospel. Use tools like M-CHAT as a flag, not a diagnosis, and pair them with your clinical judgment. Push your team to collect clear, uniform outcome data after referrals. Better measurement now is what will finally give us the evidence Jo et al. said was missing.