Validity of balance measures in cerebellar ataxia: A prospective study with 12-month follow-up

Introduction Balance deficits are common in cerebellar ataxia. Determining which balance outcome measures are psychometrically strong for this population remains an unmet need. Objective To evaluate the validity and responsiveness of two clinic-based balance measures (Berg Balance Scale [BBS] and balance sub-component of the Scale for the Assessment and Rating of Ataxia [SARA-bal]) and two laboratory-based balance measures (Sensory Organization Test [SOT] and Limits of Stability [LOS]) in cerebellar ataxia. Design Prospective cohort study. Setting Institutional study assessing 40 participants with cerebellar ataxia at baseline, and 6 and 12 months. Main outcome measures Balance was assessed using the BBS, SARA-bal, SOT, and LOS; disease severity was assessed using the SARA; and Patients' Global Impression of Change (PGIC) was used to estimate responsiveness to disease progress at 6 and 12 months. Results BBS and SARA-bal (Spearman's correlation coefficient, rho S = -0.89, p < .01) demonstrated strong criterion validity. Convergent validity was moderate to high (rho S range: -0.75 to 0.92) and external validity was low (rho S range: -0.75 to 0.11). Composite SOT scores (SOT-COM; rho S = 0.29, p < .01) and maximal excursion (MXE-LOS) in the forward (F) and right (R) directions of the LOS (rho S = 0.18, p < .01) demonstrated moderate to low criterion and convergent validity. The area under the receiver-operating characteristic curve (AUC(ROC)) and its effect size (standard response mean [SRM]) for categorizing "stable" and "worsened" patients at 6 and 12 months were satisfactory for the BBS (AUC(ROC): 0.75; SRM-Stable: 1.06; SRM-Worsened: 1.16), SARA-bal (AUC(ROC): 0.76; SRM-Stable: 0.86; SRM-Worsened: 0.85), and MXE-LOS(R) (AUC(ROC): 0.29; SRM-Stable: 0.41; SRM-Worsened: 1.39). Conclusion BBS and SARA-bal have moderate to strong criterion and convergent validity and adequate responsiveness to balance changes. Both laboratory-based measures (SOT and LOS) demonstrated a high floor effect. The SOT-COM and MXE-LOS(R) demonstrated moderate to low criterion validity, with only the MXE-LOS(R) displaying adequate responsiveness to balance changes after 6 and 12 months.