Performance-Based Assessment of Instrumental Activities of Daily Living: Validation of the Sydney Test of Activities of Daily Living in Memory Disorders (STAM)

Objectives: The distinction between dementia and mild cognitive impairment (MCI) relies upon the evaluation of independence in instrumental activities of daily living (IADL). Self- and informant reports are prone to bias. Clinician-based performance tests are limited by long administration times, restricted access, or inadequate validation. To close this gap, we developed and validated a performance-based measure of IADL, the Sydney Test of Activities of Daily Living in Memory Disorders (STAM).

Design: Prospective cohort study (Sydney Memory and Ageing Study).

Setting: Eastern Suburbs, Sydney, Australia.

Participants: 554 community-dwelling individuals (54% female) aged 76 and older with normal cognition, MCI, or dementia.

Measurements: Activities of daily living were assessed with the STAM, administered by trained psychologists, and the informant-based Bayer-Activities of Daily Living Scale (B-ADL). Depressive symptoms were measured with the Geriatric Depression Scale (15-item version). Cognitive function was assessed with a comprehensive neuropsychological test battery. Consensus diagnoses of MCI and dementia were made independently of STAM scores.

Results: The STAM showed high interrater reliability (r = 0.854) and test-retest reliability (r = 0.832). It discriminated significantly between the diagnostic groups of normal cognition, MCI, and dementia with areas under the curves ranging from 0.723 to 0.948. A score of 26.5 discriminated between dementia and nondementia with a sensitivity of 0.831 and a specificity of 0.864. Correlations were low with education (r = 0.230) and depressive symptoms (r = -0.179), moderate with the B-ADL (r = -0.332), and high with cognition (ranging from r = 0.511 to r = 0.594). The mean time to complete the STAM was 16 minutes.

Conclusions: The STAM has good psychometric properties. It can be used to differentiate between normal cognition, MCI, and dementia and can be a helpful tool for diagnostic classification both in clinical practice and research.