Accuracy and Performance of Triage Blood Pressure Measurements in A Real-World Clinic Setting

Abstract

ObjectivesTo investigate the agreement and accuracy of triage blood pressure (BP) in a real-world clinic setting, compared with the reference standard. Study designPaired triage and standardized BP measurements from patients 4 through 21 years old evaluated in an obesity-related hypertension clinic were obtained via chart-review. Triage BPs were measured by a medical assistant or nurse, often by automated device. Triplicate manual BPs were obtained by the clinic physician. Bland-Altman analyses determined mean differences between paired triage and mean standardized BPs. GEE-based multivariable relative risk (RR) regression determined the RR of triage BP overestimation by ≥5mmHg. Overall agreement, sensitivity, specificity, positive predictive value, and negative predictive value of triage BP measurements identifying hypertensive BP were determined. Results130 participants with 347 clinic encounters were included. Mean age was 13.3 years (standard deviation 3.94), 76% were Black, and 58% were male. Overall mean systolic and diastolic BP difference was 8.7 mmHg (95% limits on agreement: -16.66, 34.07) and 4.1 mmHg (95% limits on agreement: -18.56, 26.68), respectively. Triage systolic BP was more likely overestimated by ≥5 mmHg when a large adult (RR=1.49; 95%CI: 1.00, 2.21) or thigh cuff (RR=1.94; 95%CI: 1.08, 3.51) was required compared with when a child/adult cuff was required. Overall agreement in identifying hypertensive BP was 57.6%. Sensitivity (52.6%), specificity (63.4%), positive predictive value (60.8%), and negative predictive value (55.3%) were low across all cuffs. ConclusionsThere was poor agreement between usual triage and standardized BP measurements, with potential for significant clinical implications.