Behavioral predictors of improved speech output in minimally verbal children with autism.

Only 38 children were included, and they came from different phases of pilot, proof-of-concept, and ongoing trial work. The paper says the second study was a randomized clinical trial, but it does not describe the allocation method. Baseline ADOS and MSEL scores were missing for the earliest participants, so the analyses depended on complete-case and imputation methods. The complete-case sample is described inconsistently: one passage reports 17 participants with complete data, another reports 24. Predictors were tested and then reduced to significant ones, so the final model may overfit the small sample. AMMT combined intoned singing and rhythmic drum tapping, and the study did not test the components separately. Improvement was measured on practiced and unpracticed words, not on spontaneous communication. Double scoring of 10% of probes yielded 68.0% agreement on syllables approximately correct, 70.1% on consonants correct, and 54.7% on vowels correct. All participants scored below the 1st percentile on the Mullen Expressive Language subtest, so raw scores were used instead of T-scores.

No conflicts of interest or funding declaration appears in the supplied text. The publication types list NIH extramural and non-U.S. government research support.

Do not conclude that singing and drum tapping alone caused the speech gains, or that phonetic inventory can be used to choose therapy for individual patients. The study included 38 children from mixed pilot and trial phases, did not separate treatment components, and predictor results changed when missing data were handled differently.