Are Performance-Based Measures Sufficiently Reliable for Use in Multicenter Trials?

Background. The literature contains few reports of the test-retest reliability of performance-based measures. The purpose of this study was to determine the test-retest reliability of a battery of seven timed, performance-based mea­ sures used to assess the functional limitations of frail, older adults. Methods. One hundred and five frail, elderly subjects were twice administered a battery of timed tests approxi­ mately 2 weeks apart: 8-foot walk, get-up-and-go test, stair climb, single and repetitive standing from a chair, and sin­ gle and repetitive lO-pound lifts with the upper limbs. Agreement between the mean times recorded for accomplishing each task at the two administrations was assessed. Results. Intraclass correlation coefficients ranged from .25 for the single chair stand to .79 for the 8-foot walk. Only the time taken for the single lO-pound lift was significantly greater at the first administration as compared with the second. Conclusions. Timed performance-based measures have a wide range of test-retest reliability. Performance-based protocols that reflect familiar tasks with discrete starting and ending points may achieve higher reliability than tasks that are unfamiliar to subjects or may have ambiguous elements in them.