Audio file
S. Sardesai1, S. Durairaj1, A. Arumugam2, V. Guddattu3, J. Solomon1, S. Mohapatra1, S. Kumaran1
1Manipal Academy of Higher Education, Department of Physiotherapy, Udupi, Karnataka, India, 2University of Sharjah, Department of Physiotherapy, Sharjah, United Arab Emirates, 3Manipal Academy of Higher Education, Department of Data science, Udupi, Karnataka, India

Background: Models predicting post-stroke upper extremity motor recovery, derived from subjective and/or objective outcome measures have been formulated, however, there is no systematic review to corroborate evidence on performance of motor recovery prediction models using instrument-based measures. Therefore, with this systematic review we aim to provide evidence on post-stroke upper extremity motor recovery prediction models or predictors encompassing instrument-based outcome measures. 

Purpose: This systematic review would help us in corroborating evidence regarding instrument- based outcomes thus making post-stroke assessment more objective and measurable in its course. 

Methods: The protocol has been registered in PROSPERO (CRD42019133644). The protocol for this systematic review has been published in “Physical Therapy Reviews” journal (Taylor and Francis). Six online databases (PubMed, Web of Science, Scopus, OvidSP, Proquest, CINAHL) have been searched from inception to 25th April 2020 using appropriate search terms. Two reviewers have independently screened relevant articles for inclusion and data extraction from included studies using a modified CHecklist for critical Appraisal and data extraction for systematic Reviews of prediction Modelling Studies (CHARMS) is completed. Risk of bias of included studies was assessed using the Prediction model Risk Of Bias ASsessment Tool (PROBAST). Overall level of evidence has been determined using the Grading of Recommendations Assessment Development and Evaluation (GRADE) criteria. 

Results: A total of 6378 articles were derived from the six online databases. After de-duplication we had 4784 articles for title screening. After exclusion, 267 abstracts were reviewed and 109 full text articles have been selected for inclusion. Data extraction was done using a customized data extraction form derived from CHARMS checklist and individual studies' risk of bias was assessed using PROBAST. 

Conclusion(s): A majority of the prediction models have included candidate predictors such as motor evoked potentials derived from transcranial magnetic stimulation, corticospinal integrity from diffusion tensor imaging, electroencephalography and functional magnetic resonance imaging. However, a very small number of longitudinal cohort studies were found that have included predictors pertaining to movement quality such as kinematic analysis, pointing towards a need for initiating good quality primary studies in this avenue.

Implications: Most systematic reviews have analogized the available data on prediction models based on patient-reported or therapist-rated outcomes. On the contrary, other reviews have looked into outcomes that are instrument-based but there have been methodological differences in the studies. Thus, a systematic review that would substantiate the evidence for available prediction models based on neuroimaging, neurophysiological and biomechanical outcome measures for predicting upper extremity recovery in the first 12 months would make a valuable addition to the existing literature. 

Funding, acknowledgements: N/A

Keywords: Cerebrovascular accident, Prediction, Regression models

Topic: Neurology: stroke

Did this work require ethics approval? No
Institution: N/A
Committee: N/A
Reason: This is a systematic review and thereby does not require ethics approval as it does not involve patient related data

All authors, affiliations and abstracts have been published as submitted.

Back to the listing