A new device for measuring esophageal variceal pressure

Background: Esophageal varices are a frequent source of bleeding in patients with cirrhosis. Elevated intravariceal pressure is associated with variceal bleeding. There is no simple, easy-to-use device for noninvasive measurement of intravariceal pressure. The purposes of this study were to develop a noninvasive method for measuring intravariceal pressure, and to develop a model of esophageal varices that can be used to test this pressure measurement device. Methods: A variceal pressure measurement device was constructed by placing a 20 MHz US transducer in a latex balloon catheter sheath and attaching the catheter to a pressure transducer. The pressure measurement device was passed though the accessory channel of a large-channel endoscope and tested in blinded fashion by using tip deflection to compress each of 4 variceal models with the device. The pressure within each model was measured 10 times by 2 separate investigators blinded to the actual pressures. The mean (SD) pressure was calculated for each model. The variceal models were made of nitrocellulose dialysis tubing filled with water. Each "varix" had the same diameter but a different intraluminal pressure (5.5,10,15, 21.5 mm Hg). Observations: The correlation coefficient between the actual and measured "varix" pressures for the first investigator (L.S.M.) was r = 0.96: 99% Cl [0.93, 0.98]. For the varix models with pressures of 21.5, 15, 10, and 5.5 the percent errors were, respectively, 9.5, 3.9, 5.1, and 0.7. The correlation coefficient between the actual and measured varix pressures for the second investigator (Q.D.) was r= 0.97: 99% Cl [0.94,0.98]. For the varix models with pressures of 21.5, 15, 10, and 5.5 the percent errors were, respectively, 10.3, 3.4, 9.8, and 1.1. The correlation coefficient between the 2 investigators (L.S.M., O.D.) for the varix model pressures was r= 0.97: 99% Cl [0.95, 0.99]. Conclusion: The variceal pressure measuring device developed for this study measured intravariceal pressure in a model varix with a low percent error and high correlation to the actual pressures. Intraobserver and interobserver variability was low.