Feasibility and effectiveness of a digital voice assistant for improving anti-osteoporosis medication adherence, and osteoporosis knowledge and attitudes, in postmenopausal women with osteoporosis: A 12-month randomised controlled trial

Abstract Summary Digital voice assistants (DVAs) are feasible for delivering a digital health intervention designed to improve osteoporosis self-management in postmenopausal women with osteoporosis. However, the DVA intervention did not improve anti-osteoporosis drug adherence, nor did it enhance osteoporosis knowledge or attitudes in this population. Purpose To determine feasibility and effectiveness of a digital voice assistant (DVA) intervention for improving anti-osteoporosis medication adherence, and osteoporosis knowledge and attitudes, in postmenopausal women with osteoporosis. Methods This 12-month single-blinded, randomised controlled trial included 50 postmenopausal women with osteoporosis randomised to DVA (N = 25) or control (N = 25) for 6 months, followed by a 6-month follow-up period. DVA participants received an Amazon Alexa device that delivered osteoporosis education videos, medication reminders and interactive quizzes. Control participants received emails with links to osteoporosis information. Anti-osteoporosis medication possession ratio (MPR; acceptable adherence defined as ≥ 0.8) was determined using Pharmaceutical Benefits Schedule data. Osteoporosis knowledge was measured using the Osteoporosis Knowledge Assessment Tool (OKAT) and medication attitudes were measured using the Adherence Evaluation of Osteoporosis Treatment (ADEOS-12) questionnaire. Results The mean ± SD age of participants was 64.3 ± 6.1 years and 6-month DVA intervention adherence (number of DVA sessions accessed) was 79.5% (95%CI: 73.9, 84.9). The proportion of participants with acceptable 12-month MPRs was similar between groups (control: 86.4% [95%CI: 77.0, 93.6]; DVA: 95.0% [95%CI: 88.4, 100.0], P = 0.34). Mean OKAT scores improved in both groups after both 6- and 12 months, but there were no significance between groups. Changes in mean ADEOS-12 scores did not differ between baseline and 6 months in DVA compared to control (0.61 [95%CI: − 0.80, 2.03]) but worsened post-intervention from 6 to 12 months (net difference: − 1.42 [95%CI: − 2.80, − 0.06]). Conclusions This DVA-delivered intervention achieved good adherence but did not improve medication adherence, osteoporosis knowledge, or attitudes compared with control. Future studies should target populations with poor adherence to anti-osteoporosis medication.