Reliability and validity of the T-shirt test for the assessment of unsupported sitting in manual wheelchair users with spinal cord injury

This study aimed to evaluate the validity and reliability of the T-shirt test (TST) in assessing sitting stability under three thigh support conditions and with timed outcomes derived in six ways among individuals with a spinal cord injury (SCI). The TST was performed five times under three thigh support conditions (85%, 55% and 25% of total thigh length) in two evaluations spaced between 7-14 days. For each thigh condition, six different outcomes were derived (average or best time from 2, 3, and 5 trial). All outcomes derivation showed excellent reliability on test day (intraclass correlation coefficient; ICC >= 0.997) and excellent test-retest reliability (ICC >= 0.874) for each thigh support condition. The TST showed high inverse correlations with the Spinal Cord Independence Measure III (SCIM)-mobility score for all outcomes and support conditions (rho >=-0.706), except for Best-5; moderate inverse correlations with total SCIM-total scores for most outcome derivations and support conditions (rho >=-0.636); and a moderate inverse correlation with confidence and capacity domains of Wheelchair Skills Test-Questionnaire for most outcome derivation and support conditions (rho >=-0.504). The TST could discriminate cervical from high and low thoracic levels of injuries under minimal thigh support condition. Overall, all the TST-derived outcomes and support conditions showed adequate validity and test-retest reliability, but Best-5 had inconsistency. Under the minimal thigh support condition, all outcome derivations except Best-3 could discriminate cervical from other injury-level groups. Although all outcome derivations and thigh support conditions provided reliable results, we recommend using the average of 3 trials under the maximal thigh support condition.