Abstract: SA-PO387
Predictors of Urgent Dialysis and Hospitalization Following Ambulance Transport to the Emergency Department
Session Information
- Hemodialysis and Frequent Dialysis: CV and Risk Prediction
November 05, 2022 | Location: Exhibit Hall, Orange County Convention Center‚ West Building
Abstract Time: 10:00 AM - 12:00 PM
Category: Dialysis
- 701 Dialysis: Hemodialysis and Frequent Dialysis
Authors
- Thanamayooran, Aran, Nova Scotia Health Authority, Halifax, Nova Scotia, Canada
- Nallbani, Megi, Nova Scotia Health Authority, Halifax, Nova Scotia, Canada
- Vinson, Amanda Jean, Nova Scotia Health Authority, Halifax, Nova Scotia, Canada
- Clark, David, Nova Scotia Health Authority, Halifax, Nova Scotia, Canada
- Fok, Patrick Terrence, Nova Scotia Health Authority, Halifax, Nova Scotia, Canada
- Goldstein, Judah, Nova Scotia Health Authority, Halifax, Nova Scotia, Canada
- More, Keigan, Nova Scotia Health Authority, Halifax, Nova Scotia, Canada
- Swain, Janel M., Nova Scotia Health Authority, Halifax, Nova Scotia, Canada
- Wiemer, Hana, Nova Scotia Health Authority, Halifax, Nova Scotia, Canada
- Tennankore, Karthik K., Nova Scotia Health Authority, Halifax, Nova Scotia, Canada
Background
Dialysis patients may require timely, monitored dialysis (urgent dialysis) or hospitalization after ambulance transport to the emergency department (ambulance-ED). We developed and internally validated risk prediction models for urgent dialysis and hospitalization for ambulance-ED in a cohort of chronic dialysis patients.
Methods
We included all ambulance-ED transports for hemodialysis patients affiliated with a large regional program from 2014-2018. "Urgent dialysis" was defined as dialysis within 24 hours of ED arrival in a monitored setting or with the first ED patient blood potassium level >6.5mmol/L. Predictors included categorized vital signs prior to ambulance transport (taken by paramedics) and time from last dialysis. Logistic regression models were used to predict urgent dialysis and hospitalization and internally validated using bootstrapping. Model discrimination was evaluated using the C-statistic and calibration using the Hosmer-Lemeshow test.
Results
A total of 271 dialysis patients experienced 878 ambulance-ED transports. 63 transports (7.2%) required urgent dialysis and 299 (34.0%) resulted in hospitalization. Hypoxemia (odds ratio; OR: 4.04, 95% CI:1.75-9.33) and a time from last dialysis of 24-48 hours (OR: 3.43, 95% CI: 1.05-11.9) and >48 hours (OR: 9.22, 95% CI: 3.37-25.23) were associated with urgent dialysis. A risk prediction model for urgent dialysis had good discrimination (C-statistic: 0.83) and calibration (Hosmer-Lemeshow: 0.89). The prediction model for hospitalization had no individually significant predictors of hospitalization and moderate discrimination (C-statistic: 0.67) but good calibration (Hosmer-Lemeshow: 0.74).
Conclusion
This study highlights the possibility of predicting certain short-term outcomes in dialysis patients using information available to paramedics during ambulance-ED transport.