Clinimetric properties of the performance-oriented mobility assessment

The Performance-Oriented Mobility Assessment (POMA) is a widely used instrument that provides an evaluation of balance and gait. It is used clinically to determine the mobility status of older adults or to evaluate changes over time. To support the use of the POMA for these purposes, the clinimetric properties (in particular, responsiveness) were determined.

Participants (78% female; mean age=84.9 years) were living in either self-care or nursing-care residences. Concurrent and discriminant validity were assessed with the total group (N=245), whereas reliability and responsiveness were determined with a subsample (n=30). Fall-related predictive validity was assessed with a subsample of 72 participants.

In addition to the POMA, several reference performance tests were administered. The POMA was assessed on 2 consecutive days by 2 raters (observers). The analyses included the calculation of Spearman rank correlation coefficients (R), limits of agreement (LOA) with Bland-Altman plots, minimal detectable changes at the 95% confidence level (MDC(95)), and sensitivity and specificity with regard to predicting falls. When possible, findings for the total scale (POMA-T) were complemented by findings for its balance subscale (POMA-B) and its gait subscale (POMA-G).

The interrater and test-retest reliability for the POMA-T and the POMA-B were good (R=.74-.93), whereas for the POMA-G, the reliability values, although high as well, were systematically slightly lower (R=.72-.89). The Spearman correlations with the reference performance tests (R=|.64|- |.68|) indicated satisfactory concurrent validity for the POMA-T and the POMA-B, but the corresponding findings for the POMA-G (R=|.52|- |.56|) were less convincing. The discriminant validity values of the 3 scales were about the same. The LOA for the POMA-T were on the order of -4.0 to 4.0 for test-retest agreement and -3.0 to 3.0 for interrater agreement. On the basis of the MDC(95) values, it was concluded that changes in POMA-T scores at the individual level should be at least 5 points and that those at the group level (n=30) should be at least 0.8 point to be considered reliable. Even when optimal cutoff points were used, sensitivity and specificity values (varying between 62.5% and 66.1%) for the POMA-T as well as for its 2 subscales indicated poor accuracy in predicting falls.

The POMA-T and its subscale POMA-B have adequate reliability and validity for assessing mobility in older adults. The POMA-T is useful for demonstrating intervention effects at the group level. Changes within subjects, however, should be at least 5 points before being interpreted as reliable changes. The accuracy of the POMA-T in predicting falls is poor.