Psychometric validation of the Manual Ability Measure-36 (MAM-36) in patients with neurologic and musculoskeletal disorders

Abstract

Objectives: To evaluate the psychometric properties of the Manual Ability Measure-36 (MAM-36), a new hand function outcome measure, and to examine differences in manual abilities and item parameters in patients with neurologic and musculoskeletal conditions.

Design: Convenience sample from 2 time periods, cross-sectional.

Setting: Outpatient rehabilitation units and private hand clinics.

Participants: Patients (N=337; mean age, 50.3+/-14.9y) with a variety of neurologic and musculoskeletal (orthopedic) diagnoses. Most of these individuals were community dwelling, and all had residual functional limitations in the hand(s).

Interventions: Not applicable.

Main outcome measures: Rasch analysis was performed on MAM-36 data to evaluate both scale structure and psychometric properties, which include rating distribution, step measures, item fit, separation, and dimensionality. A t test was performed to examine the differences in manual abilities in patients with the 2 conditions. Uniform differential item functioning (DIF) between neurologic and musculoskeletal groups was examined. (DIF occurs when subgroup members within the sample with the same level of the underlying trait being measured respond differently to an individual item.) Manual ability estimates were recalibrated with step and common item anchoring; they were compared with those derived from the original analysis.

Results: The 36 items measured a single construct with no misfitting items. The scale was used as intended. The items can reliably separate the participants into 5 ability strata. Neurologic patients had a significantly lower mean manual ability than musculoskeletal patients. Fourteen items exhibited DIF. However, DIF had no effect on either scale quality or calibration of manual ability. We decided that a single rating scale is appropriate for both groups.

Conclusions: This study showed that the MAM-36 has more than adequate psychometric properties and can be used as a generic outcome measure for patients with a wide variety of clinical diagnoses.