2001 OPEN FORUM Abstracts
AGREEMENT AND REPRODUCIBILITY OFA POINT-OF-CARE SYSTEM FOR BLOOD GASES ANALYSIS (i-STAT)
Rodrigo Morales, M.D.; MichaelKirkpatrick, M.D.; Pat Hoffman RPFT, RRT. Division of Pulmonary, Universityof South Alabama, Mobile, AL.
Background: Theintroduction of a Point-of-care (POC) system for blood gases analysis (i-STAT,i-STAT Corporation) to our clinical practice prompted the questions of agreementwith our previous system (IL Synthesis, Instrumentation Laboratory) and thereproducibility of the new system. Both systems use direct potentiometry forthe calculation of PCO2 and pH and the pO2 is measured amperometrically.
Method: Two groups of arterialblood samples were obtained and analyzed for pH, pCO2 and PO2. The first group(n=75) was analyzed by both systems. The second group (n=50) was analyzed bythe POC system twice. The bias, or mean difference and the limits of agreementfor the two groups were calculated by the statistical method described by Blandand Altman for comparing methods of measurement. The 95% Confidence Intervals(CI) of the limits of agreement of the groups were assessed to determinate theagreement between both systems and the reproducibility of the POC system inthe clinical practice.
Results:
Group 1: Agreementbetween I-stat and IL Synthesis | |||||
| Bias | LLA | 95%CI | ULA | 95%CI | |
| pH | -0.01 | -0.05 | -0.045,-0.055 | 0.03 | 0.025,0.035 |
| pCO2 (kPa) | -0.09 | -0.69 | -0.59,-0.79 | 0.5 | 0.4,0.6 |
| pO2 (kPa) | 0.29 | -1.74 | -1.41,-2.08 | 2.33 | 2.00,2.67 |
| Group2: Reproducibility of I-stat | |||||
| Bias | LLA | 95%CI | ULA | 95%CI | |
| pH | 0.001 | -0.02 | -0.017,-0.023 | 0.02 | 0.017,0.023 |
| pCO2 (kPa) | 0.03 | -0.41 | -0.32,-0.49 | 0.46 | 0.38,0.55 |
| pO2 (kPa) | 0.09 | -1.1 | -0.86,-1.37 | 1.27 | 1.03,1.51 |
LLA: Lower Limitof Agreement ULA: Upper Limit of Agreement
Experience:The POC system has been in service at our institution for 26 months.
Conclusion:In our clinical practice both systems can be used interchangeable. The biasfound between both systems can be partially explained by the reproducibilityof the i-STAT. We recommend the interchangeable use of the i-STAT and the ILSyntheis when the clinician determines that the limits of agreement of the biaswill not significantly interfere with the medical care of the patient.