The effect of the interval between blood pressure determinations on the delay in the detection of changes: a computer simulation.

Abstract

The frequency of automated noninvasive blood pressure (NIBP) measurements during routine anesthesia is a balance between potentially deleterious effects of frequent cycling and a delay in detecting changes caused by a long cycle time. A computer model generated systolic blood pressures that changed to a new, random value after a period of stability. We sampled these data at intervals between 1 and 10 min to simulate NIBP measurements. A separate algorithm, based on Trigg's Tracking Variable, indicated when a change had been detected. For each set of variables, the simulation was repeated 1000 times, and the average time to detect a change was recorded. The mean time to detect a change was 8.0 min with a 1-min cycle, 8.9 min with a 2-min cycle, 10.8 min with a 5-min cycle, and 13.0 min with a 10-min cycle. As the cycle time increased, the delay in detecting changes increased but only by approximately half the increase in the cycle time. The optimum variables for the trend detection algorithm also changed as the NIBP interval increased. Provided that abrupt changes in blood pressure are not anticipated, a 1- or 2-min cycle time for NIBP offers little advantage over a longer period. We used a computer model to study the effect of increasing noninvasive blood pressure (NIBP) sampling interval on the detection of blood pressure changes. The detection time increased only 50% of the increase in the sampling interval. This information may help optimize NIBP intervals in different circumstances.