Multicenter study of whole-blood creatinine, total carbon dioxide content, and chemistry profiling for laboratory and point-of-care testing in critical care in the United States

OBJECTIVESTo introduce a creatinine biosensor and a total carbon dioxide content (TCO2) method for whole-blood measurements, to evaluate the clinical performance of a new transportable analyzer that simultaneously performs these two and six other tests (Na, K, Cl, glucose, urea nitrogen, and hematocrit), and to assess the potential of the new analyzer for point-of-care testing in critical care by comparing results obtained by nonlaboratory personnel and by medical technologists. DESIGNMulticenter sites compared whole-blood measurements with the transportable analyzer to plasma measurements from the same specimens with local reference instruments. One site compared whole-blood results produced by nonlaboratory personnel vs. medical technologists and evaluated day-to-day and within-day precision at the point of care. SETTINGS AND PATIENTSFour medical centers in the United States. Venous and arterial specimens from 710 critically ill patients with a variety of diagnoses. Point-of-care testing in the emergency room and operating room. RESULTSThe linear regression analyses at the four medical centers showed the followingcreatinine (a) slope, 0.91 to 1.22, (b) y intercept, −0.07 to 0.15 mg/dL, and (c) r, 0.77 to 1.00; and TCO2(a) slope, 0.64 to 1.00, (b) y intercept, 1.36 to 9.6 mmol/L, and (c) r, 0.52 to 0.72 (yi, whole-blood analyses; xi, plasma reference measurements). Bland-Altman plots also were used to assess multicenter creatinine and TCO2 results. Of the other analytes, K, glucose, and urea nitrogen had the highest r-values. For the eight chemistry profile tests performed at the point of care (yi, nonlaboratory personnel results; xi, medical technologist results), the average value of r was 0.96 (sd 0.08) in the operating room and 0.96 (sd 0.06) in the emergency room, and mean paired differences (yi − xi) were not statistically or clinically significant. Precision was acceptable. CONCLUSIONSThe performance of the creatinine biosensor and the TCO2 method was acceptable for whole-blood samples. Comparisons of whole-blood results from the transportable analyzer and plasma results from the local reference instruments revealed analyte biases that may be attributed to differences between direct whole-blood analyses and indirect-diluted plasma measurements and other factors. Performance of nonlaboratory personnel and medical technologists was equivalent for point-of-care testing in critical care settings. The whole-blood analyzer should be useful when patient c