Predictive Value of Upper Extremity Outcome Measures After Stroke-A Systematic Review and Metaregression Analysis

Abstract

A better understanding of motor recovery after stroke requires large-scale, longitudinal trials applying suitable assessments. Currently, there is an abundance of upper limb assessments used to quantify recovery. How well various assessments can describe upper limb function change over 1 year remains uncertain. A uniform and feasible standard would be beneficial to increase future studies' comparability on stroke recovery. This review describes which assessments are common in large-scale, longitudinal stroke trials and how these quantify the change in upper limb function from stroke onset up to 1 year. A systematic search for well-powered stroke studies identified upper limb assessments classifying motor recovery during the initial year after a stroke. A metaregression investigated the association between assessments and motor recovery within 1 year after stroke. Scores from nine common assessments and 4,433 patients were combined and transformed into a standardized recovery score. A mixed-effects model on recovery scores over time confirmed significant differences between assessments (P < 0.001), with improvement following the weeks after stroke present when measuring recovery using the Action Research Arm Test (β = 0.013), Box and Block test (β = 0.011), Fugl-Meyer Assessment (β = 0.007), or grip force test (β = 0.023). A last-observation-carried-forward analysis also highlighted the peg test (β = 0.017) and Rivermead Assessment (β = 0.011) as additional, valuable long-term outcome measures. Recovery patterns and, thus, trial outcomes are dependent on the assessment implemented. Future research should include multiple common assessments and continue data collection for a full year after stroke to facilitate the consensus process on assessments measuring upper limb recovery.

Keywords: metaregression; motor assessments; motor function; motor recovery; motor rehabilitation; stroke; upper limb.

Figure 1

Prisma flow diagram displaying the literature search and eligibility checking process.

Figure 2

Last observation carried forward (LOCF) prediction of recovery per assessment over the time course of 1 year, cross-fading the underlying raw values. The solid black line represents the predicted recovery pattern, based on log-transformed difference in means (MNLN) within the confidence interval of the prediction (dashed lines). The green horizontal line resembles a healthy score, respective to 100% recovery. The gray dots represent the underlying data points, with the size depicting the sample size of the respective data point. How these underlying data points are related is presented in Supplementary Figure 1. ARAT, Action Research Arm Test; BBT, Box and Block test; F-M, Fugl–Meyer Assessment (upper extremity); GRIP, grip force; MI, Motricity Index; PEG, peg test; RMA, Rivermead Motor Assessment; SIS, Stroke Impact Scale; WMFT, Wolf Motor Function Test.