Noninvasive monitoring of blood pressure in the critically ill: reliability according to the cuff site (arm, thigh, or ankle).

Objective: In the critically ill, blood pressure measurements mostly rely on automated oscillometric devices pending the intra-arterial catheter insertion or after its removal. If the arms are inaccessible, the cuff is placed at the ankle or the thigh, but this common practice has never been assessed. We evaluated the reliability of noninvasive blood pressure readings at these anatomic sites.

Design: Prospective observational study.

Setting: Medical-surgical intensive care unit.

Patients: Patients carrying an arterial line with no severe occlusive arterial disease.

Intervention: Each patient underwent a set of three pairs of noninvasive and intra-arterial measurements at each site (arm, ankle, thigh [if Ramsay sedation scale >4]) and, in case of circulatory failure, a second set of measurements after a cardiovascular intervention (volume expansion, change in catecholamine dosage).

Measurements And Main Results: In 150 patients, whatever the cuff site, the agreement between invasive and noninvasive readings was markedly higher for mean arterial pressure than for systolic or diastolic pressure. For mean arterial pressure measurement, arm noninvasive blood pressure was reliable (mean bias of 3.4 ± 5.0 mm Hg, lower/upper limit of agreement of -6.3/13.1 mm Hg) contrary to ankle or thigh noninvasive blood pressure (mean bias of 3.1 ± 7.7 mm Hg and 5.7 ± 6.8 mm Hg and lower/upper limits of agreement of -12.1/18.3 mm Hg and -7.7/19.2 mm Hg, respectively). During acute circulatory failure (n = 83), arm noninvasive blood pressure but also ankle and thigh noninvasive blood pressure allowed a reliable detection of 1) invasive mean arterial pressure <65 mm Hg (area under the receiver operating characteristic curve of 0.98 [0.92-1], 0.93 [0.85-0.97], and 0.93 [0.85-0.98] for arm, ankle, and thigh noninvasive blood pressure, respectively); and 2) a significant (>10%) increase in invasive mean arterial pressure after a cardiovascular intervention (area under the receiver operating characteristic curve of 0.99 [0.92-1], 0.90 [0.80-0.97], and 0.96 [0.87-0.99], respectively).

Conclusion: In our population, arm noninvasive mean arterial pressure readings were accurate. Either the ankle or the thigh may be reliable alternatives, only to detect hypotensive and therapy-responding patients.