Performance validation of the PRADA digital care system in a cohort-controlled study demonstrates improved end-of-life registration associated with reduced adverse outcomes.

AbstractRising urgent care use, avoidable hospital deaths, and missed end-of-life (EOL) registration reflect poor care at this critical phase of life. These are known to be avoidable by earlier identification and EOL registration for effective care planning. Current identification tools are generally non-digital. PRADA, a digital system, uses prediction methodologies for early identification of individuals, escalated to clinicians in active decision support, to promote clinical assessment, the completion of EOL process, EOL registration and to improve outcomes. Our objective was to assess its effectiveness. In 12 participating general practices, baseline EOL metrics were assessed retrospectively over 2 years (n=64,887, alive 63,344 (97.6%), died 1,543 (2.4%)). Following implementation of the PRADA system outcomes were determined prospectively over 1 -year (n= 66,321 (alive 65,496 (98.7%), died 831 (1.3%)). EOL registration in those that died, versus non-registration, was associated with lower A/E attendances, emergency admissions, hospital bed day occupancy (all p<0.001) and markedly reduced deaths in hospital (18.5% vs 51.0%, p<0.001). Implementation of PRADA improved registration rates (41.8% vs 25.8%, p<0.001). This shift, for the final year cohort (n=831), identified an additional 133 deaths, avoided 24 deaths-in-hospital, saved 386 bed days, with an estimated direct cost avoidance of {pound}192,789. EOL registration increased in participating practices by 17.6% {+/-}13.7% (range -1.4 to +52.6). However, A/E attendance (74.4%) and urgent admission rates (70.6%) remained high in those registered, each indicating scope for further improvements and efficiency gains over time. EOL registration rates were low at baseline, missed registration was associated with adverse outcomes, the use of PRADA improved EOL registration by 62% (41.8% vs 25.8%). PRADA use at large scale, in real time direct care demonstrates strong utility, benefit, and performance validity. Author SummaryExisting screening tools for identifying patients with potential palliative care needs are generally manual and paper based, developed in small or disease specific cohorts, or in hospital settings. Systematic reviews show only moderate performance in accurate mortality prediction, with weak evidence for effective deployment in real time care and little validity testing. In this study, end-of-life registration in those that died was associated with reduced urgent care events, hospital bed days occupied and deaths in hospital. Large-scale deployment of PRADA, a real-time end-of-life digital health care system, over 1-year general practitioners was associated with a 62% increase in registration rates and reductions of adverse events. The study highlights between practice variation, proposing mechanisms for its mitigation with newly defined benchmarks. PRADA thus has high utility, strong clinical benefit and robust performance validity. The UK NHS 10-year plan has called for better care at the end-of-life and calls for digital solutions. Current tools do not have digital capability, nor the evidence base for use at the scale required. We demonstrate the performance, utility and validity of PRADA, as potential solution to address the inequalities in outcomes at the end of life.