User competency is still a major factor affecting analytical performance of glucose meters in patient service

This study investigated the analytical performance of glucose meters while in use for 1.5 years in a hospital setting. The effect of user competency, strip stability and strip lot number variation on the performance of blood glucose meters was investigated.

The studies for linearity, accuracy, imprecision, and method comparison were conducted according to Clinical and Laboratory Standards Institute (CLSI) guidelines Point of Care Testing (POCT) 12-A3. Proficiency testing results were used for meter-meter comparison. Quality control (QC) data, strip lot and users' identification were extracted from the data management system 1.5 years after the new systems were implemented. The frequency of users QC testing was used as the indicator of user competency.

The average total imprecision based on 20 glucose meters with QC materials targeting 2.5 mmol/L and 17.05 mmol/L increased by 2.4 fold and 1.83 fold, respectively, over that from initial evaluation studies. When the glucose levels were <5.6 mmol/L, the absolute bias ranged from -0.75 to +0.55 mmol/L relative to the core laboratory analyzer and the accuracy goal was met by 96.3% of samples. When the glucose levels were ≥5.6 mmol/L, the percentage bias ranged from -17.8% to +9.4% and 79.3% of samples met the accuracy goal. The users who performed QC checks less frequently (<10 times/year) showed significantly different QC mean and greater variation than users who performed QC checks >20 times/year, especially for QC targeting 2.5 mmol/L.

Poorer analytical imprecision of POCT blood glucose meters following introduction to patient service is mainly due to the low level of user competency.