Data on behavioural or biomedical variables were extracted into standard data extraction forms. Behavioural variables included communication (such as nonverbal communication, vocalisations, question asking), stereotypy, play and challenging behaviour (such as pica, self-injury, aggression and property destruction). Biomedical variables included levels of urinary peptides, relevant enzymes and antibodies.
Results were coded as positive where all participants made improvements and there were statistically significant differences between groups, negative where none of the participants made improvements or if there were no statistically significant differences between groups and mixed where some participants improved and others did not.
Percentage of non-overlapping data (PND) was calculated for single-subject designs. Repeated measures effect sizes were calculated for group designs. Repeated measures effect sizes were corrected for bias according to the technique developed by Hedges and Olkin.
One reviewer extracted data, which were checked by a second reviewer. Discrepancies were resolved through discussion.