Table 2. Multiple linear regression models of clinical and environmental factors, which are independently associated with bias in blood glucose measurement with point-of-care glucometers.
Glucometer | Sampling site | Clinical and environmental factors | Measurement error (95%CI) § | P-value |
---|---|---|---|---|
Precision PCx | Arterial | Mean blood glucose (mg/dl) † | 0.12 (0.08 to 0.16) | <0.001 |
Hematocrit (%) | -0.89 (-1.38 to -0.40) | <0.001 | ||
Accu-chekAdvantage II | Arterial | Mean blood glucose (mg/dl) † | 0.10 (0.06 to 0.15) | <0.001 |
Hematocrit (%) | -0.74 (-1.29 to -0.19) | 0.009 | ||
Arterial pH | -34.8 (-62.8 to -6.7) | 0.020 | ||
Accu-chekAdvantage II | Fingerstick | Mean blood glucose (mg/dl) † | 0.12 (0.06 to 0.17) | <0.001 |
Use of norepinephrine | -25.8 (-40.1 to -11.5) | 0.001 | ||
Hematocrit (%) | -1.04 (-1.75 to -0.32) | 0.005 | ||
Accu-chekAdvantage II | Venous | Mean blood glucose (mg/dl) † | 0.22 (0.13 to 0.30) | <0.001 |
Arterial pH | -77.4 (-135.7 to -19.1) | 0.010 |
§ = Measurement error represents the change in the difference of blood glucose between point-of-care glucometer and arterial central lab measurement for one unit increase in the clinical or environmental factor;
† = mean blood glucose is the mean between the point-of-care and arterial central lab blood glucose values; CI = confidence interval.