Therapeutic Drug Monitoring of Lithium: A Study of the Accuracy and Analytical Variation Between Laboratories in Denmark

BACKGROUND: Serum lithium is monitored to ensure levels within the narrow therapeutic window. This study examines the interlaboratory variation and inaccuracy of lithium monitoring in Denmark.

METHODS: In 16 samples consisting of (1) control materials (n = 4), (2) pooled patient serum (n = 5), and (3) serum from individual patients (n = 7), lithium was measured in 19 laboratories using 20 different instruments. The lithium concentrations were targeted by a reference laboratory. Ion-selective electrode (n = 5), reflective spectrophotometric (RSM, n = 5), and spectrophotometric (n = 10) methods were used.

RESULTS: Acceptable accuracy-interpreted as total differences from target values (bias) less than or equal to ±12%-was generally found in patient samples above 0.7 mmol/L. Below 0.7 mmol/L, 8 instruments had 2 or more patient samples exceeding a difference to the targeted reference value of >12%. Seven of these instruments had a systematic positive or negative bias and more so at lower lithium concentrations. Three poorly calibrated instruments were found in the ion-selective electrode group, 3 in the spectrophotometric group, and 2 in the RSM group. The instruments using reflectance spectrophotometry (RSM) were on average 21% positively biased when measuring control materials. However, this effect was not observed in patient samples.

CONCLUSIONS: Large interlaboratory variation was found below 0.7 mmol/L because of 7 instruments with a poor accuracy and 1 with poor precision. Methods should be recalibrated or substituted. Controls below 0.7 mmol/L are recommended.