Pilot evaluation of the psychometric properties of a self-medication Risk Assessment Tool among elderly patients in a community setting

BACKGROUND: Although community pharmacists in the United Kingdom are expected to assess elderly patients' needs for additional support in managing their medicines, there is limited data on potentially useful assessment tools. We sought to evaluate a 13-item assessment instrument among community dwelling elderly patients, 65 years and above. The instrument is composed of a cognitive risk sub-scale of 6 items and a physical risk sub-scale of 7 items. FINDINGS: The instrument was administered to elderly patients in a survey performed in a community to the west of Glasgow, Scotland. The survey recruited 37 participants, 31 from 4 community pharmacies and 6 patients whose medication management tasks were managed by the West Glasgow Community Health and Care Partnership (managed patients). Community pharmacists independently rated 29 of the 37 participants' comprehension of, and dexterity in handling their medicines. We assessed scale reliability, convergent validity and criterion validity. In sub-analyses, we assessed differences in scores between the managed patients and those recruited from the community pharmacies, and between multi-compartment compliance aid users and non-users. The instrument showed satisfactory internal consistency (Cronbach's alpha of 0.792 for 13-item scale). There was significant strong negative correlation between the cognitive risk sub-scores and community pharmacists' assessment of comprehension (ρ = -0.546, p = 0.0038); and physical risk sub-scores and community pharmacists' assessment of dexterity (ρ = -0.491, p = 0.0093). The Area Under the Receiver Operator Characteristic Curve (AUC ± SE; 95%CI) showed that the instrument had good discriminatory capacity (0.86 ± 0.07; 0.68, 0.96). The best cut-off (sensitivity, specificity) was ≥4 (65%, 100%). In the sub-analyses, managed patients had significantly higher cognitive risk sub-scores (6.5 versus 4.0, p = 0.0461) compared to non-managed patients. There was a significant difference in total risk score (4 versus 2, p = 0.0135) and cognitive risk sub-score (4 versus 1.5, p = 0.0029) between users and non-users of multi-compartment compliance aids. CONCLUSIONS: This instrument shows potential for use in identifying elderly patients who may have problems managing their own medicines in the community setting. However, more robust validity and reliability assessments are needed prior to introduction of the tool into routine practice.