Deprescribing in Older People Approaching End of Life: A Randomized Controlled Trial Using STOPPFrail Criteria

OBJECTIVES: Older people approaching end of life are commonly prescribed multiple medications, many of which may be inappropriate or futile. Our objective was to examine the effect of applying the STOPPFrail, a recently developed deprescribing tool, to the medication regimens of older patients with advanced frailty. DESIGN: Randomized controlled trial. SETTING: Two acute hospitals in Ireland. PARTICIPANTS: Adults 75 years or older (n = 130) with advanced frailty and polypharmacy (five or more drugs), transferring to long-term nursing home care. INTERVENTION: A STOPPFrail-guided deprescribing plan was presented to attending physicians who judged whether or not to implement recommended medication changes. MEASUREMENTS: The primary outcome was the change in the number of regular medications at 3 months. Secondary outcomes included unscheduled hospital presentations, falls, quality of life, monthly medication costs, and mortality. RESULTS: Intervention (n = 65) and control group (n = 65) participants were prescribed a mean (plus or minus standard deviation [SD]) of 11.5 (+/-3.0) and 10.9 (+/-3.5) medications, respectively, at baseline. The mean (SD) change in the number of medications at 3 months was -2.6 (+/-2.73) in the intervention group and -.36 (+/-2.60) in the control group (mean difference = 2.25 +/- .54; 95% confidence interval [CI] = 1.18-3.32; P < .001). The mean change in monthly medication cost was -$74.97 (+/-$148.32) in the intervention group and -$13.22 (+/-$110.40) in the control group (mean difference $61.74 +/- $26.60; 95% CI = 8.95-114.53; P = .02). No significant differences were found between groups for any of the other secondary outcomes. CONCLUSION: STOPPFrail-guided deprescribing significantly reduced polypharmacy and medication costs in frail older people. No significant differences between groups were observed with regard to falls, hospital presentations, quality of life, and mortality, although the trial was likely underpowered to detect differences in these outcomes.