Random variation and systematic error caused by various preanalytical variables, estimated by linear mixed-effects models.

BACKGROUND

We wanted to determine whether specific, preanalytical sample handling increases preanalytical variation and bias test results compared with optimal handling.

METHODS

Blood was collected into 4 serum-separation tubes from each arm of 60 outpatients. In 30 of the patients, half of the tubes were transported in the pneumatic tube system, while the other half were manually delivered. In the remaining patients, the blood samples were collected using 21-gauge straight needles (green needles) and 23-gauge butterfly needles. Half of the tubes were mixed by inverting 5-6 times, and the other half by one inversion. Linear mixed-effects models were used as statistical method.

RESULTS

Transporting samples in the pneumatic tube system caused a significant bias to the results for LD (4.5 U/L, p<0.001) and magnesium (0.0021 mmol/L, p=0.003). For CK and glucose, the preanalytical variation was significantly higher for samples transported in the pneumatic tube system vs manual delivery. Using butterfly needles resulted in lower values (p<0.05) for calcium (-0.0072 mmol/L), CK (-0.75 U/L) and LD (-1.6 U/L) compared with 21-gauge needles. The preanalytical variation for ALP was significantly higher with butterfly needles.

CONCLUSIONS

The specific sample handling had significant but small random and systematic effects on results for some analytes.