Prm28 - Performance of the Early Warning System Vital to Predict Unanticipated Higher-Level of Care Admission and in-Hospital Death of Ward Patients

Abstract

apolipoprotein ε4 (APOE4) status and years of education were added as intercept and/or slope effects in the mixed model as guided by exploratory analyses of the two outcomes. A time-dependent APCC covariate accounted for the hypothesized dependence of time to diagnosis on APCC decline. The model was fitted with the {JM} package from R v3.4.3 on 2047 subjects with at least two APCC measurements and who were cognitively unimpaired and less than 90 years old at inclusion. An internal validation was performed via predictive checks. RESULTS: The model captured the link between risk factors, early cognitive decline and time to diagnosis. The model-estimated longitudinal cognitive decline (population average and individual trajectories) and overall risk of developing MCI/AD symptoms provided a good fit to the data. As expected, the model estimated a higher risk of developing MCI/AD symptoms for older subjects and for APOE4 carriers. A significant association was observed between decline in APCC and the risk of developing MCI/AD symptoms. The joint model also adequately captured the individual dynamic risk of developing MCI/AD symptoms using personal longitudinal APCC data until a given observation time. CONCLUSIONS: This novel method provides a tool to dynamically estimate the individual risk of developing MCI/AD for cognitively unimpaired subjects who have longitudinal measures from sensitive neuropsychological tests.OBJECTIVES: Decision-makers need to understand long-term clinical outcomes to assess the cost-effectiveness of new treatments from a lifetime perspective. Given limited follow-up of clinical trials, our aim was to demonstrate that estimates from experts regarding long-term survival can be combined with empirical data from multiple clinical trials to provide more credible extrapolated survival curves. METHODS: A case study was performed using clinical trial (ELIANA) data with limited follow-up (1.5 years), which was combined with expert elicited survival rates from 2-5 years for patients with relapsed or refractory B-cell pediatric acute lymphoblastic leukemia treated with CTL019, a chimeric antigen receptor T-cell therapy. Additionally, a phase I CTL019 with no expert information was included. Elicited survival proportions (and uncertainty) from experts were transformed into discrete hazards for each interval. We incorporated expert information in a Bayesian hierarchical model with branches for experts and trials. Parameters defining survival for clinical trial data or expert opinion were estimated and combined to produce estimates of the 'true' parameters of the survival function. RESULTS: Standard survival models fit solely to the clinical trial data produced extrapolations that were extremely variable, despite minor differences in model fit statistics with 5-year estimates ranging from 0.0% to 52.4% (95% CrI: 0.0%, 84.2%). Incorporating expert information using a hierarchical model produced more consistent estimates with 5-year survival estimates of 53.2% (2.8%, 78.9%). Adding another clinical study wit...

Algorithms for Prediction of Clinical Deterioration on the General Wards: A Scoping Review

“…Nineteen of the publications were full papers and two were conference abstracts. Most of the studies (n = 18) were from the United States; there was one study from South Korea, 30 one study from Portugal, 31 and one study from the United Kingdom 32 . In 15 of the studies, there was a strict focus on general or specific wards; 6 studies also included the ICU and/or emergency departments.…”

“…Earlier detection of deterioration can potentially allow coordinated alarm follow-up and lead to more efficient use of resources. 20,21,31,43,46,47 Greater numbers of variables do not always improve the quality of monitoring. For example, in one study, an algorithm combining only heart rate, respiration rate, and age outperformed an EWS that tracked six vital sign measures.…”

“…The assessment of predictive algorithm protocols must include their impact on clinical workflow, workload, and resource utilization. Earlier detection of deterioration can potentially allow coordinated alarm follow‐up and lead to more efficient use of resources 20,21,31,43,46,47 …”

“…Furthermore, they expected workflow to improve and workload to decrease. 21,23,27,31,33,35,38,41 Despite the capacity of modern EMRs to store large amounts of patient data, some investigators felt improvements to real-time access, data quality and validity, and data density are needed to ensure valid associated predictions. 21,22,24,32,37…”