Complex speech-language therapy interventions for stroke-related aphasia: the RELEASE study incorporating a systematic review and individual participant data network meta-analysis

Background People with language problems following stroke (aphasia) benefit from speech and therapy. Optimising therapy for aphasia recovery is a research priority. Objectives The objectives were to explore patterns predictors of communication recovery, optimum intervention provision, whether or not effectiveness varies by participant subgroup domain. Design This comprised systematic review, meta-analysis network individual data. Setting Participant data collected in clinical settings. Interventions under investigation was after stroke. Main outcome measures main absolute changes scores baseline on overall ability, auditory comprehension, spoken language, reading writing functional communication. Data sources participants Electronic databases systematically searched, including MEDLINE, EMBASE, Cumulative Index Nursing Allied Health Literature, Linguistic Language Behavior Abstracts SpeechBITE (searched inception 2015). results screened eligibility, published unpublished sets (randomised controlled trials, non-randomised cohort studies, case series, registries) at least 10 reporting duration severity identified. Existing collaborators primary researchers named identified records invited contribute electronic sets. Individual the public domain extracted. Review methods demographics, interventions, outcomes quality criteria independently extracted two reviewers, available as Meta-analysis used generate hypotheses. Results We retrieved 5928 174 across 28 countries, comprising 75 (3940 data), 47 randomised trial (1778 data) 91 (2746 median age 63 years (interquartile range 53–72 years). 53 unavailable, but potentially eligible, trials (46 these appeared include therapy). Relevant filtered into each analysis. Statistically significant included (functional communication, data: 532, n = 14 trials) sex (overall 482, 11 trials; trials). Older being longer time since onset predicted poorer recovery. A negative relationship between score change ( p < 0.0001) may reflect reduced improvement possible high scores. frequency, duration, intensity dosage variously associated naming There insufficient examine spontaneous greatest gains ability [14.95 points (95% confidence interval 8.7 21.2 points) Western Aphasia Battery-Aphasia Quotient] [0.78 0.48 1.1 Aachen Test-Spontaneous Communication] receiving 4 5 days weekly; comprehension [5.86 1.6 10.0 Test-Token Test], 3 weekly. [15.9 8.0 23.6 [0.77 0.36 1.2 participation 2 (and more than 9) hours weekly, whereas highest [7.3 4.1 10.5 Test] excess 9 weekly (with similar notes weekly). While clinically made alongside different intensities, [18.37 10.58 26.16 [5.23 1.51 8.95 20–50 Network meta-analyses interventions unstable. Relative variance acceptable (< 30%). Subgroups specific interventions. Limitations graded low risk bias predominantly based highly selected participants, assessments thereby limiting generalisability. Conclusions Frequency, baseline, varied subgroup. Future work These exploratory findings require confirmatory study designs test hypotheses generated develop tailored Study registration registered PROSPERO CRD42018110947. Funding project funded National Institute Care Research (NIHR) Social Delivery programme will be full ; Vol. 10, No. 28. See NIHR Journals Library website further information. also provided Tavistock Trust Aphasia.