Abstract 13863: Dichotomizing Systolic Blood Pressure During Out-of-Hospital Emergency Care: A Population-Based Outcomes Study
Objectives: Hypotension is common during emergency care, and may lead to organ failure or death. Although treatment by emergency medical services (EMS) for hypotensive patients may improve outcomes, the optimal opportunities are poorly defined. We sought to develop a rigorous definition of high-risk hypotension by determining the outcome-based thresholds for dichotomizing systolic blood pressure (SBP) during out-of-hospital care.
Methods: We performed a population-based cohort study of adult, non-traumatic patients transported by EMS personnel from 2002-2006. Eligible records were linked to hospital discharge data and death certificates. We grouped patients with SBP measurements by EMS into a development cohort (N=132,624) and separate validation cohort (N=22,020). We evaluated dichotomizing thresholds across SBP (40mmHg-140mmHg) for their association with death within 24 hrs, ICU admission, and 30-day mortality. Thresholds were compared using the 0/1 distance and Youden index (ROC curve-based statistical tests to compare thresholds), and adjusted Z statistics from multivariable logistic regression models.
Results: In the development cohort, 1,594 (1.2%) of patients died within 24 hrs, 7,404 (6%) were admitted to the ICU, and 6,888 (5%) died within 30 days. The area under the receiver operating characteristic curve for SBP was 0.61. Both the 0/1 distance and the Youden index suggested optimal thresholds for SBP between 110-120mmHg. The maximum Z statistic in adjusted models occurred between 110-115mmHg. In the validation cohort, a SBP threshold of 110mmHg corresponded to a sensitivity of 30% (95%CI:26.8,33.2%), specificity of 83.3% (95%CI:82.8,83.8%), and positive likelihood ratio of 1.8 (95%CI: 1.6,2.0) for 30-day mortality. Using a cutoff of 90mmHg, specificity and positive likelihood ratio increased to 95.7% (95%CI:95,96%) and 3.1 (95%CI:2.6,3.7), respectively, but sensitivity decreased to 21% (95%CI:18,24%).
Conclusions: Although out-of-hospital SBP, alone, is a poor discriminator of patient outcomes, the best threshold for identifying high-risk patients is between 110 and 115mmHg. Future application of these SBP thresholds may require additional predictors to identify those at highest risk for intensive care or death.
- © 2011 by American Heart Association, Inc.