Multisite evaluation of a continuous intraarterial blood gas monitoring system

Abstract

Background: We compared the performance of a new, continuous intraarterial blood gas (CIABG) monitor with arterial values obtained periodically and analyzed by conventional equipment. Methods: A CIABG monitoring system consisting of a sterile, disposable, fiberoptic sensor and a microprocessor- controlled monitor with a self-contained calibration unit and detachable display panel was used. The sensor was inserted through a 20-G radial artery cannula. Light was transmitted from the monitor to the sensor tip where it reacted with fluorescent dyes sensitive to oxygen or hydrogen ions (analytes). The change in the intensity of the photoluminescent radiation caused by the analytes was measured every 20 s and derived blood gas values were displayed. Twenty-nine sensors were evaluated in 29 surgical or intensive care unit patients at one of three institutions (Stanford University Hospital, Evanston Memorial Hospital, and the Palo Alto Veterans Administration Hospital). The duration of study averaged 6 h (5-8 h) in the operating room, and 46 h (7-121 h) in the intensive care unit. A total of 552 values were compared with those obtained at regular intervals and analyzed in the hospital blood gas laboratory. Average bias (mean difference between lab value and CIABG), precision (SD of difference), and drift (change in the bias with time were determined. Results: At arterial oxygen tension (P(O2)) values of 32-528 mmHg, the average bias was -1% meaning that the average CIABG monitor values were 1% lower than those obtained by conventional equipment. The precision was 15%. At arterial P(O2) values of 32-99 mmHg, average bias and precision were -0.3 ± 8.9 mmHg. At arterial carbon dioxide tension (P(CO2)) values of 24-54 mmHg, average bias and precision were 1.3 ± 3.3 mmHg, and at pHa values of 7.23-7.57, average bias and precision were 0.01 ± 0.04. Observed drift per day was -1.2% for arterial P(O2), 0.3 mmHg for arterial P(CO2), and 0.01 for pH. Bias and precision for samples compared in two pairs of like-model in vitro blood gas analyzers were 0.4 ± 4.6% for arterial P(O2) over the full range, and 0.4 ± 3.7 mmHg for values less than 100 mmHg, -0.5 ± 1.8 mmHg for arterial P(CO2), and 0.01 ± 0.01 for pHa. Although the occasional marked discrepancies between one or more CIABG and in vitro values could sometimes be corrected by flushing the arterial catheter or repositioning the sensor, usually we could not determine the cause of the discrepancy or which values were the more accurate. Conclusions: Over the range of values and under the clinical conditions studied, CIABG monitoring provides immediate blood gas results and trend information with sufficient agreement with in vitro results to be reliable for decision making in most clinical circumstances. Generally, the differences in the values between the two methods of analysis were the result of the combination of the inherent errors of each method. Additional studies need to be undertaken to evaluate the performance of the CIABG monitor across wider ranges of blood gas values, especially for arterial P(O2) values less than 60 mmHg and arterial P(CO2) values greater than 50 mmHg.