Ten studies (n=121) were included in the review: nine pre-post studies (three were also described as between-groups design and five as individual design); and one described as a between-groups study only. The authors reported that all studies showed weak methodological structure. Sample sizes ranged from two to 54 patients.
Comparisons of the Snoezelen approach versus baseline behaviour (four studies), an active therapeutic intervention (two studies) or a non-active control condition (three studies) resulted in moderate to large effect sizes (0.63 to 2.63), but did not reach statistical significance. All three comparisons showed evidence of statistical heterogeneity (I2 ranged between 42% and 91%).
Comparison of baseline behaviour versus generalised behaviour after Snoezelen sessions (seven studies) indicated a significant improvement in adaptive behaviour post intervention resulted in a large effect size (0.76, p<0.001). There was no evidence of significant statistical heterogeneity.
Subgroup analyses indicated large effect sizes for all subgroups, with statistically significant results reported for pre-post design (effect size 1.99, p<0.05), assessment situation (generalised behaviour effect size 0.84, p<0.001 and non-generalised behaviour effect size 2.25, p<0.05) and maladaptive outcome (effect size 1.13, p<0.001). With the exception of generalised behaviour and maladaptive behaviour, the remaining subgroups indicated high statistical heterogeneity.