Upsala J Med Sci 95: 257-258, 1990 

External Quality Assurance Model for Improvement of 
Specificity and Assuracy of Serum Hormone 

Determinations in the Nordic Countries 
A. Ruokonen, P. Hyltoft-Petersen and 0. Blaabjerg 

Departments o/ Clinical Clwmistry, Oulun Yltopisto, Finland and 
Odense University Hospilal, Odense, Denmark 

Previous external quality assurance projects of hormone determi- 
nations in the Nordic countries and other areas have shown the 
wide scatter of results without essential improvement during many 
years. Reasons for the wide scatter like wrong calibration, 
unspecific reactions and various factors causing imprecision have 
been mainly characterized but until now very little has been done 
to eliminate them. In this project we try to reduce their 
influence. The purpose of this project is to: 
- set up a new quality assurance model for improvement of 

specificity and accuracy of hormone determinations in the 
Nordic countries 

- to test the model for serum thyrotropin (TSH), free thyroxine 
(FT,) thyroxine (T,) , triiodothyromine (T,) and cortisol (C) 
determinations 

- to define practical goals of accuracy and specificity for 

- to maintain the quality, if acceptable accuracy and specifi- 
these hormone determinations 

city are reached by the quality assurance model. 

Control material (quality assurance samples and the calibrator) 
is frozen human serum and its T, and C concentrations have been 
determined by GC-MS. Quality assurance samples are used to 
evaluate specificity and calibration bias (at three concentration 
levels) of the analytical method of each laboratory. If needed, 
the calibrator is used for recalibration. 
There are four kinds of quality assurance samples (Fig 1). One 

of the controls is a normal serum pool ( A )  with known con- 
centrations of hormones, another is the same pool ( B )  with no or 
very low hormone concentrations, the third pool (C) is the same 

17 - 908573 257 



as pool A ,  but it also contains constituents expected to cause 
nonspecific reactions. Samples A and B are used to test accuracy 
and samples C and D to test specificity. 
In result reports quality goals have been defined and comments 

are given on reproducibility, linearity, specificity and bias of 
the methods of each laboratory. This project is carried out after 
the protein project. 

NORMAL SERUM SAMPLES 
(TSH, FT4, T4, T3. C) 

I 

Fis 1. 

LIKE SAMPLE A + CONSTITUENTS 
EXPECTED TO CAUSE UNSPECIFIC 
REACTIONS ( T 4 ,  RT3, 11-DEOXYCORTISOL, 
TBG, PREALBUMIN) 

L 

QUALITY ASSURANCE MODEL 

LIKE SAMPLE A BUT WITH NO 
OR VERY LOU HORMONE 
CONCENTRATIONS AND NO 
CONSTITUENTS EXPECTED TO 
CAUSE UNSPECIFIC REACTIONS 

(TEST OF ZERO OR LOW) 

SAMPLES (A AND B) TO TEST ACCURACY SAMPLES (C AND D) TO TEST SPECIFICITY 

LIKE SAMPLE B + CONSTITUENTS 
EXPECTED TO CAUSE UNSPECIFIC 
REACTIONS (T4, RT3, 11-DEOXYCORTISOL, 
TBG, PREALBUMIN) 

1 

B l  D I 

Correspondence: 
Dr. Aim0 Ruokonen, 
Department of Clinical Chemistry, 
Oulun Yliopisto, 
90220 OULU, Finland 

258