Diagnostic Accuracy of Arm Conicity to Detect Inaccurate Non-Invasive Blood Pressure Measurement in Obese Patients

Non-invasive blood pressure (NIBP) measurement can be challenging in patients with obesity. We aimed to determine if arm conicity, mid arm circumference or body mass index (BMI) could be used to predict pre-operative and intra-operative NIBP measurement error. Eligible patients had elective surgery and BMI ≥ 35 kg/m². Mid-arm circumference and arm slant angle (a measure of arm conicity) were measured according to standardised methods. Systolic (SBP) and diastolic (DBP) blood pressure were obtained using invasive (INV) and non-invasive methods at pre-induction (T0) and 20 min post-induction. NIBP accuracy was defined as ≤ 10 mmHg difference from invasive measurement. Post hoc analyses evaluated under and over-estimation. Receiver operating characteristic (ROC) curves were calculated. The 141 participants had a mean (SD) BMI of 46.9 (8.9) kg/m². There were 531 measurements at T0 and T20 with INV and NIBP available. Of 265 for SBP, 150 (56.7%) were inaccurate and of 266 for DBP, 113 (42.5%) were inaccurate. All ROC curves for right arm slant angle, right mid-arm circumference and BMI for SBP and DBP at T0 and T20 approached a 45° line. Area under ROC curves (AUROCs) ranged from 0.47 (0.36, 0.57) to 0.64 (0.54, 0.74), suggesting poor diagnostic performance. The best performer was right arm slant angle, with AUROCs 0.63 (0.51, 0.75) for DBP and 0.69 (0.59, 0.78) for SBP for underestimation at T0. Inaccurate measurements were common. Right arm conicity, mid-arm circumference and BMI were poor classifiers of NIBP measurement error at the pre-induction and intraoperative time points. EDITORIAL COMMENT: Accuracy of NIBP measurements in obese patients is uncertain. This clinical study compared routine oscillometric clinical blood pressure methods to a reference invasive blood pressure measure in an obese cohort. Significant disagreement between these is described, demonstrating reliability problems for some of the common arm oscillometric cuff implementations.