Upsala J Med Sci 98: 443-444, 1993 

9. Conclusions 
Carl-Henric de Verdier,' Torgny Groth2 and Per Hyltoft Petersen3 

'Department of Clinical Chemistry, and 2Unitfor Biomedical Systems Analysis, 
University of Uppsala, Uppsala, Sweden and 'Department of Clinical Chemistry, 

Odense University Hospital, SO00 Odense, Denmark 

In all clinical quality work the first steps ought to be to define the clinical goals and the 

quality specifications. In order to be meaningful they ought to be expressed as total 

allowable errors (TEA ) in quantitative terms. 

This project deals mainly with anaZytical goals and specifications. Three types of 

approaches have been used: 

1) The state of the art approach 

2) The biological approach 

3) The clinical usefulness approach. 

The state of the art approach aimes primarily at comparison with other laboratories. 

The biological approach gives easily available and more general goals/specifications. 

Clinical investigations are used in clinical work and it should thus be most meaningful to 

try to use the third alternative. Guidelines for this approach are, however, not generally 

available and we have in this report tried to indicate different ways to obtain 

goals/specifications following this approach. 

Statistical, graphical and computer methodologies have been discussed and applied. 

Terminology is important for understanding and for transfer of information. This 

NORDKEM project report provides suggestions for terms and abbreviations and stresses 

that a universally accepted terminology within this area is highly needed. 

Twentytwo associated Nordic projects are accounted for in the project report. They have 

brought up important questions for discussion. 

Two chapters deal with special topics. One, originating from the Nordic Protein project, 

describes the quality necessary for sharing common reference intervals for 9 specific 

proteins in the nordic countries. The other, originating from a "transferability project" 

17-935253 443 

, 



describes how 'long term bias' in the different laboratories can be assessed and serve as 

the prerequisite for making corrections. 

Analytical quality specifications (AQSpecs) should be provided by the clinical laboratories 

and the Report gives many examples of how they can be estimated. 

Correspondence: 

Carl-Henric de Verdier, M.D., professor 

Department of Clinical Chemistry 

University Hospital 

S-751 85 Uppsala, Sweden 

444