In this study, we compared the responsiveness and validity of the Box and Block Test (BBT), the Nine-Hole Peg Test (NHPT), and the Action Research Arm Test (ARAT). We randomized 59 patients with stroke into one of three rehabilitation treatments for 3 weeks. We administered six outcome measures (BBT, NHPT, ARAT, Fugl-Meyer Assessment [FMA], Motor Activity Log [MAL], and Stroke Impact Scale [SIS] hand function domain) pretreatment and posttreatment. We used the standardized response mean (SRM) to examine responsiveness and the Spearman rank correlation coefficient (rho) to examine concurrent validity. The BBT, NHPT, and ARAT were moderately responsive to change and not significantly different (SRM = 0.64-0.79). The correlations within the BBT, NHPT, and ARAT were moderate to good at pretreatment (rho = -0.55 to -0.80) and posttreatment (rho = -0.57 to -0.71). The BBT and ARAT showed fair to moderate correlations with the FMA, MAL, and SIS hand function domain at pretreatment and posttreatment (rho = 0.31-0.59), whereas the NHPT demonstrated low to fair correlations with the FMA and MAL (rho = -0.16 to -0.33) and moderate correlations with the SIS hand function domain (rho = -0.58 to -0.66). Our results indicate that the BBT, NHPT, and ARAT are suitable to detect changes over time. While simultaneously considering the responsiveness and validity attributes, the BBT and ARAT can be considered more appropriate for evaluating dexterous function than the NHPT. Further studies with larger samples are needed to validate these findings.
Trial registration: ClinicalTrials.gov NCT00778453.