Reference intervals based on hospitalized 'healthy’ patients and medical students in relation to analytical bias for serum potassium

Abstract

The reliability of reference intervals for measurements of serum (S)-potassium in Danish hospital laboratories was investigated (i) by estimation of reference interval based on two different, healthy subpopulations and (ii) by comparison of reference intervals for S-potassium with analytical bias in each of 52 Danish laboratories.

(i) The reference values from 227 hospitalized ‘healthy’ patients were obtained during the period 1979 to 1987 from the first-drawn serum specimen from the hospitalized patients, who were later discharged from the hospital without a diagnosis. The estimated 0.95 reference interval was 3.34 to 4.52 mmol l^-1. The other reference sample group consisted of 314 medical students from whom blood was collected in the period from 1983 to 1987. Here the estimated reference interval was from 3.44 to 4.53 mmol l^-1. The concentration values from both reference sample groups were corrected for analytical bias (+0–05 mmol l^-1).

(ii) The 52 Danish laboratories revealed a considerable variability in reference intervals which, regarding the lower reference limit, ranged from 3.2 to 3.7 mmol l^-1 in strong contrast to the analytical bias (ranging from-0.08 to +0.15 mmol l^-1) in 50 laboratories (two outliers). There was no relationship between lower reference limit and analytical bias in the individual laboratories. It is concluded that analytical performance allows for more uniform (even common) reference intervals throughout the Danish and perhaps Nordic hospital laboratories.

Key Words:

• confidence intervals
• probit transformation
• quality assurance
• REFVAL