Upsala J Med Sci 98: 417-427, 1993 

7. Approaches towards a Combined Data- and 
Knowledge-base for Analytical Quality 

Specifications in Clinical Chemical 
Laboratories 

Carl Henric de Verdier' 
'Department of Clinical Chemistry, University of Uppsala, Uppsala, Sweden 

INTRODUCTION 

It is not realistic to believe that the management of every clinical chemical laboratory is 

willing to go through literature and reports and try to collect all clinical and technical 

background material for estimating numerical values of 'clinical goals' and for listing the 

'quality specifications' for all kinds of measurements of their laboratory. As the clinical and 

especially the technical background material has a rapid turnover time it seems to be most 

suitable to collect the material in continuously updated computerized database systems. 

A complete database ought to cover several hundred different types of components and 

the measurement of each component may be applied in many different clinical situation. 

It is thus not possible - in this report - to build up an extensive database. I have, however, 
in this preliminary list tried to present a number of different and illustrative examples, 

which can be used as models for an extended list and finally a real database. 

A database for this purpose will always contain two types of data: 

1. those describing the (analytical) measurement procedures and 

2. those describing the clinical (and the normal) situations in which it is appropriate to 

apply the measurement procedure. 

Quality assurance specifications and quality control procedures will come out as conse- 

quences of these two types of data. We have found it most appropriate in this presentation 

to use System--Measwand (Analyte) as the primary key in our data base. In many clinical 

settings one kind of measurement (analysis) is not giving the diagnosis or the advice for 

treatement. I n  such circumstances one or more clinical situtations must be defined. 

41 7 



Furthermore, in one clinical situation, it may be necessary to use the measurand in 

combination with a set of other measurands - from the clinic, the diagnostic radiology, 
department and the clinical laboratories. This may turn up to be a very complex 

assessment. Examination of a few key-examples will, however, soon learn us to draw 

conclusions, valid for a number of analogous examples. 

I am confident that we in the future will see several groups working with combined data- 

and knowledge-bases within this area and we feel sure that computerized such bases is an 

excellent medium for rapid exchange of information and knowledge. 

EXAMPLES OF COMBINED DATA- AND 

KNOWLEDGE-BASES FOR AQSpecs 

The following list have been written using the dataprogram Microsoft Excel@ 3.0. This 

program does not allow writing characters as subscripts, instead they have been written 

within [ 1. The abbreviation c* has been used for the 'conventional true value'. The 
measurands are listed in alphabetic order. In each segment clinical goals calculated in 

different ways are listed first, followed by recommendations from proficiency testing 

bodies. A few examples have been given: S--Creatinine, B--Haemoglobin and S--Urate, 

in which the clinical goals have been expressed as fractions of easily understandable 

physiological functions or by logical reasoning obtained pathological processes. 

REFERENCES 

1. 

2. 

3. 

4. 

418 

Bundeartzekammer. Qualitetatssicherung der quantitativen Bestimmungen im 

Laboratorium. Neue Richtlinien der Bundesiirtzekammer. Dt h z t e b l  1988;85 

(1 l):A-697-A-712. 

Costongs GMPJ, Janson PCV, Bas BM, et al. Short-term and long-term 

intra-individual variations and critical differences of clinical chemical laboratory 

parameters. J Clin Chem Clin Biochem 1985;23:7-16. 

Costongs GMPJ, Janson PCW, Bas BM, Hermans J, van Wersch JWJ, Brombacher 

PJ. Short-term and long-term intraindividualvariation and critical differences of 

heamatological laboratory factors.. J Clin Chem Clin Biochem 1985;23:69-80. 

de Verdier C-H. Assessing analytical quality specifications. Scand J Clin Lab Invest 

1994; (In Press). 



5. 

6. 

7. 

8. 

9. 

10. 

11. 

12. 

13. 

14. 

15. 

16. 

17. 

de Verdier C-H, Groth T, Hyltoft Petersen P. Medical need for quality specifications 

- a NORDKEM project for selecting the appropriate quality in clinical laboratories. 
Scand J Clin Lab Invest 1993;.53, Suppl 215:29-37. 

Fraser CG. The application of theoretical goals based on biological variation data 

in clinical chemistry. Arch Pathol Lab Med 1988;112:404-415. 

Fraser CG. Biological Variation in Clinical Chemistry. An Update: Collated Data, 

1988-1991. Arch Pathol Lab Med 1992;116:916-923. 

Fraser CG, Hyltoft Petersen P, Lytken Larsen M. Setting Analytical Goals for 

Random Analytical Error i n  Specific Clinical Monitoring Situations. Clin Chem 

1990;36: 1625-1628. 

Fraser CG, Hyltoft Petersen P, Ricos C, Haeckel R. Proposed Quality Specifications 

for Imprecision and Inaccuracy of Analytical Systems for Clinical Chemistry. Eur 

J Clin Chem Clin Biochem 1992;30:311-317. 

Geilenkeuser W-J, Kruse R, Rijhle G. Ringversuche der Deutschen Gesellshaft fuer 

Klinische Chemie -Interpretation der Auswertung. D G  Klinische Chemie 

1992;23(2):57-81. 

Groth T, Ljunghall S, de Verdier C-H. Optimal screening for patients with 

hyperparathyroidism with use of serum calcium observations: A decision-theoretical 

analysis. Scand J Clin Lab Invest 1983;43:699-707. 

Hyltoft Petersen P, Blaabjerg 0, Irjala K, IcCn A, Bjoro K. Upsala J Med Sci 

1993;98 (3): In Press. 

Linnet K. Analytical goal5 for P--Bilirubins. Upsala J Med Sciences 1993; 98(3): In 

Press. 

Linnet K. Analytical goals for accuracy and precision of plasma creatinine 

determinations evaluated by reference method measurements.. Upsala J Med 

Sciences 1993;98 (3): In Press. 

Lytken Larsen M, Hyltoft Petersen P, Fraser CG. Quality Specifications for Haemo- 

globin A l c  Assays i n  the Monitoring of Diabetes. Upsala J Med Sciences 19Y3;98 

(3): In Press. 

National Cholesterol Educationprogram Laboratory Standard Panel. Current Status 

of Blood Cholesterol Measurement in Clinical laboratories in the United States. 

Clin Chem 1988;34: 193-201. 

Skendzel LP, Barnett RN, Platt R. Medical useful criteria for analytical performance 

of laboratory tests. Am J Clin Path01 1985;83:200-205. 

419 



18. 

19. 

20. 

21. 

22. 

Thue G, Sandberg S, Fugelli P. Clinical assessment of haemoglobin values by 

general practitioners related to analytical and biological variation. Scand J CIin Lab 

Invest 1991;51:4.53-459. 

van Waeg G, Groth T. Allopurinol kinetics in humans as a means to assess liver 

function: design of a loading test. A m  J Physiol 1989;257:R237-R24.5. 

Westgard JO. Charts of Operational Process Specifications ("OPSpecs Charts") for 

Assessing the Pecision, Accuracy, and Quality Control Needed to Satisfy Proficiency 

Testing Performance Criteria. Clin Chem 1992;38:1226-1233. 

Westgard JO, Hyltoft Petersen P, Wiebe A. Laboratory Process Specifications for 

Assuring Quality in the U.S. National Cholesterol Education Program (NCEP). Clin 

Chern 1991;37:656-661. 

Wiggers P, Dalhoj J, Hyltoft Petersen P, Blaabjerg 0, H ~ r d e r  M. Screening for 

haemochromatosis: influence of analytical imprecision, diagnostic limit and preva- 

lence on test validity. Scand J Clin Lab Invest 1991;51:143-148. 

Correspondence: 

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

Department of Clinical Chemistry 

University Hospital 

S-7.51 85 Uppsala, Sweden 

420 





53 IAnal. Qual. Sp. IEurop. work gr. 1 CV[A]<0.027 

422 

55 IAnal. Qual. Sp. lGerman Rinpers. ICV[A]<0.06 



I151 I I I 

- .  
18 
19 
20 
21 
22 -. 

Numerical values (system. errors) Note Rejerences 

Insensitive to changes in bias 

Insensitive to changes in bias 

19 

19 

26 
27 
28 
29 
30 
31 
32 
33 I I 
34 I 
1s I I I 

Numerical values (System. errors) Note References 

Deviation<O.lO c' 13 

Deviation< 0.098 c* 9 
Deviation<O.ZO C* or <6.8J umoVL Imprecision included 20 
Deviation < 0 21 C* 

Bias unchanged CV[A]<%CV[Bw] 2 

1 

36 
37 
38 
19 

Numerical values (?stein. errors) Note References 
Bias unchanged CV[A]<%CV[Bw] 7 
Bias < 0.02 c* 11 

-_ 
40 
41 
42 
d1 

Bias changed <0.007 (0.018) 9 
Deviation <0.25 m m o E  Imprecision included 20 
Deviation< 0.10 c* 1 

I :; IBias < 0.03 14 controls/ser;2 pat samples I 21 
I I 

44 
45 
46 Numerical values (System errors) r o t e  Rejerences 

42 3 

47 
48 
49 
50 

Bias unchanged CV[A]<!ACV@w] 7 
Deviation <0.38 c* CV[A]<1/4CVp] 7 
Bias=O 2 controlsfseries 21 
Bias < 0.03 2 controlsfseries 21 

L a ,  I 

53 Deviation <O.OJI c* 
S4 Deviation <O. 10 c* 
5.5 Deviation < 0,18 c* 
56 Deviation< 0.03 c* 

1 Imprecision included 
9 

20 
1 

16 



424 



57 
E I F G 

I 

59 
60 
61 
62 
63 
6 1  
65 
66 
61 

iVunterica1 values (systenr. errors) .Vote References 
Bias unchanged CV[A]<%CV[B\v] 7 
Deviation< 0.046 CV[A]<l/4CV[B] 7 

Bias unchanged ICV[Bw]=O.O-I; Prob. 95% 5 
Bias unchanged j 2 pat samples 5 

Bias unchanged Detect +O. 17; Prob. 99% 8, 17 
Dev. <0,16c* when CV[A] 0.036 Detect+O.l6:Prob. 95% 14 

I 

. _  
79 
80 
81 
82 
91 

Nuinericol values (system. errors) h b t e  Refirences 
Bias unchanged 

Bias unchanged 

The two analyses combined. 
Prob.99%. Predictive value 22 
of a positive result 0.29. 

I 

425 

"I 

8 1  
85 
86 
87 
88 
89 
90 
91 

Nunterical volues (system errors) #ore References 
Bias unchanged CV[Aj<%CVpw] 8 
Bias unchanged CV[Bw]=0.027; Prob.95% 18 
Bias unchanged CV[Bw]=O.O2J; Prob.95% 8 
Bias unchanged CV[Bw]=0.027; Prob.95% 4 

92 
93 
94 
95 
96 
a7 

Deviation<O.OlZ c* 9 
Deviation <0.07 c* Imprecision included 20 

,l 

98 
99 
100 

I 
iVunierica1 values ($,;sent. errors) /!Vote References 
Bias unchanged /Detect+/-l%HbAlc; Prob.O.80 15 
Bias unchanged lDetect+/-2% HbAlc; Prob.0.99 15 



I A B C D 
1101 I 

426 

111 
112 
113 
114 
115 
116 
117 
118 

I S--POTASSIUM 
Nuiiierical values (randoiir errors) 

Clinical goals Biol. approach within subj variatlon CV[A]<O 006 

Anal. Qual. Sp. Europ. work. gr. CV[A]<O 021 
Anal. Qual. Sp. 
Anal. Qual. Sp. German Ringvers. 

US prof. test. 

158 
159 
1 sn 

Nuiiierical values (randoin errors) 
Clinical goals IBiol. approach n i t h n  subj variahon CV[A]<O 052 

I 
161 
162 
163 

Anal. Qual. Sp. Europ. work. gr. CV[A]<0.063 
Anal. Qual. Sp. 
Anal. Qual. Sp. German Rinpers. 

US prof. test. 



112 
113 
114 
115 
116 
117 
118 
1 1 0  

Numerical values (vsteiir. errors) h'ote References 
Bias unchanged , CV[A]<XCV[Bw] 7 

DeviaIion<0.016 (0.048) c* 9 
Deviation <0.5 m m o l L  Imprecision included I 20 
Deviation<0.08 c* Imprecision included 10 

I 

121 
122 
123 

Note References iVutirerica1 values (iysterir. errors) 
Bias unchanged CV[A]<%CV[Bw] 7 

124 
125 
126 
127 

I 
Bias changed <0.002 (0.006) 9 
Deviation <4 m o V L  Imprecision included 20 
Deviation <0.06 c* 1 

I 

152 Deviation <0.04(0.084) 9 
153 Deviation <O.I7c* Imprecision included 20 
154 CV[A]<0.18 c* Imprecision included 1 

I"" 

161 
162 
163 

Bias change < 0.053 9 
Deviation<0.09 c* or <0.71 mmoVL Imprecision included 20 
Deviation<O.24 c* Imprecision included 10 

16 - 935253 427