Systematic Review of Discourse and Social Communication Interventions in Traumatic Brain Injury.

Only one RCT was included; the remaining 20 studies were nRCTs, cohort studies, or single-case designs, most without control groups and with small samples. Only the RCT was double-blinded; blinding of assessors was frequently absent across the other studies, increasing the risk of bias. Outcome measures varied widely across studies in type, number, and operational definition, making cross-study comparison difficult and precluding meta-analysis. Several studies relied on self-reported outcomes, which are susceptible to expectancy bias, particularly in the nRCTs where participants knew they were receiving an experimental treatment. Four studies did not report maintenance data, and two discourse approaches showed no carryover of skills at follow-up, so durability of effects is uncertain for a subset of the evidence. The review is a qualitative synthesis; it cannot estimate the size of treatment effects or compare approaches statistically. The field is in an early stage: treatments are not consistently manualized, and no single quality scale applied uniformly across all study designs.

The authors declared no competing financial or nonfinancial interests.

Do not read the finding that all 21 studies showed improvement as evidence that these treatments work. The evidence is mostly small, uncontrolled studies where gains could reflect practice effects, expectancy bias, or natural recovery. The one RCT found no difference between the two treatments it compared. The five building-block components were identified by pattern-matching across heterogeneous studies, not by controlled comparison of one component against another, so they are a starting point for program design rather than a proven recipe.