Download Full Text (111 KB)

Faculty Advisor(s)

Harold Merriman PT, PhD, CLT; Kurt Jackson, PT, PhD


Purpose/Hypothesis: The Repetitive Step Test (RST) assesses rapid stepping capability in anterior lateral (AL) and posterior lateral (PL) diagonal directions. This study’s goal was to determine the predictive validity of the RST to identify fallers with Parkinson’s Disease (PD). A secondary objective was to determine concurrent validity of the RST to other established tests. It was hypothesized that those who completed fewer steps with the RST would be more likely to fall and that the RST would predict falls at least as well as the other balance tests.

Materials and Methods: 13 subjects (8 F, 5 M; mean age 70.2 years) with mild to moderate PD were tested. Foot dominance was recorded (dominant [Dom] or nondominant [NonD]). The number of steps made by a single leg at 50% and 75% of leg length within 15 seconds were recorded. Data was collected in a single session: RST at 50% and 75%, Timed Up and Go (TUG), Four Square Step Test (FSST), Five Time Sit to Stand Test (5TSTS), and 10 Meter Walk Test (10MWT). Falls were then recorded for 6 months.

Results: ROC curve analysis for the RST, TUG, FSST, 5TSTS and 10MWT was performed to determine the area under the curve (AUC) and cut-off scores for sensitivity (Sn) and specificity (Sp) in order to predict future falls. The AUCs (0.702-0.762) for the RST were non-significant as were the AUCs for 5TSTS and 10MWT (p>0.05). AUC for the TUG was significant (0.929, p=0.01) with a cut-off value of 7.175 secs (Sn=0.857, Sp=0.833). AUC for the FSST was also significant (0.881, p=0.022) with a cut-off value of 8.79 secs (Sn=0.857, Sp=0.667).

Spearman’s rho correlation was used to compare the tests. RST 50% and RST 75% were analyzed separately in order to assess if one version showed stronger correlations. For the RST 50%, only Dom PL was strongly correlated to 5TSTS (r=-0.75, p=0.01). For RST 75%, NonD AL, Dom PL, and NonD PL were all strongly correlated to TUG (r≥-0.80, p=0.01), all forms of the 75% RST were strongly correlated to the FSST (r≥-0.82, p=0.01). NonD AL and Dom PL were strongly correlated to 5TSTS (r≥-0.77, p=0.01).

Conclusion: The AUCs for the RST were found to be non-significant, demonstrating poor ability of the RST to predict future falls in patients with PD. However, the TUG and FSST were good to excellent for fall prediction. All RST 75% directions were strongly correlated with at least one pre-validated balance test, with NonD AL and Dom PL strongly correlating with three balance tests. This generally supports our hypothesis that the RST especially at 75% leg length would show good concurrent validity with the selected pre-validated tests.

Clinical Relevance: Balance and gait measures are important for clinicians treating individuals with PD because they quantify the functional deficits seen in this population. Some tests require a considerable amount of clinic space and/or have a ceiling effect. Our findings support the use of the TUG and FSST for fall prediction in persons with PD but not the RST. However, the RST 75% could be a useful functional tool for identifying specific stepping impairments. The small number of subjects in this study may have impacted our findings.

Publication Date



Physical Therapy | Rehabilitation and Therapy

Determining the Concurrent and Predictive Validity of the Repetitive Step Test in Individuals with Parkinson's Disease