Building on the past: Systematic identification, data extraction and synthesis of pre-existing individual stroke patient datasets to inform the development and design of future clinical trials

Abstract

Background The number of stroke rehabilitation trials reported is rapidly increasing. Efficient trial design contributing to advances in rehabilitation should be informed by completed trials in the field. More than 50,000 people in the UK each year acquire aphasia: a stroke related language impairment affecting the ability to speak, understand speech, read and write with significant consequences for quality of life. Existing Cochrane systematic review evidence indicates that speech and language therapy (SLT) benefits language recovery in people with aphasia, however, the specific patient and intervention factors which predict optimal recovery and rehabilitation are unclear. By using a wider dataset with individual patient data (IPD) analysis we are enhancing the evidence synthesis process with the aim of addressing these evidence gaps. RELEASE (rehabilitation and recovery of people with Aphasia after stroke) is an international collaboration of aphasia researchers which seeks to achieve this goal. Objectives Funded by the National Institute for Health Research (Health Services and Delivery Research - 14/04/22) we have systematically gathered IPD from pre-existing aphasia research datasets to examine the natural history of recovery from aphasia, the predictors of recovery and optimal interventions (by rehabilitation regimen, delivery model and the aims and content of treatment). Methods We invited contributions of primary datasets from members of the Collaboration of Aphasia Trialists (cats). We also conducted a systematic search of existing published research to identify a comprehensive set of potentially existing aphasia research datasets which met our inclusion criteria. Research datasets were required to include a minimum of 10 people, a measure of aphasia severity as a consequence of stroke and information on time since stroke. We invited researchers from these studies to contribute data and to create a unique multilingual, international, interdisciplinary resource in this clinical field. Results Following a systematic search of the literature, we screened 5276 titles (including 2346 abstracts and 1152 full texts), from which we identified 874 eligible studies. We have received 76 study datasets contributing IPD from 4597 people with aphasia (56 through the systematic search and 20 via cats). These data have been contributed from 23 countries and we have identified a further 2400 IPD in the public domain. The substantive challenge is our planned IPD metaanalysis to examine recovery, predictors of recovery and effectiveness of intervention approaches. Our statistical analysis plan states that a one-stage approach will be conducted for the primary analyses, although a two-stage approach will also be explored. Network meta-analyses and meta-regression (some of which includes subgroup analyses) are also planned. We will discuss the methodological challenges, particularly which arise when there are non-standardized data, some non-randomized data, a large number of outcome measurements and some degree of sparse data. Conclusions RELEASE is the largest systematically developed, evidence synthesis study in the field of aphasia, and is more complex than most IPD trial meta-analyses. Our research will not only provide important evidence relating to the recovery of people with aphasia, but will also be an exemplar to researchers who plan to create databases to analyse complex individual patient data.