Professional Documents
Culture Documents
Current Databases On Biological Variation
Current Databases On Biological Variation
within-subject (CVw) and the between-subject ence change value (RCV, difference between
(CVb) variation. two consecutive measurements of one ana-
Many works have estimated the biological lyte in a person and representing a signi®cant
variation components and four compilations of change in health status), number of speci-
results of biological variation have been pre- mens needed to establish the homeostatic set
sented: Ross [1], Fraser [2, 3] and SebastiaÂn- point, and index of heterogeneity (ratio
Gambaro et al. [4]. In 1992, the European between the observed CV of a set of
Group for the Evaluation of Reagents and individual variances to the theoretical CV)
Analytical Systems in Laboratory Medicine [7];
(EGELAB) published quality speci®cations for 3. Descriptive information such as mean, stan-
imprecision and bias (at that time termed dard deviation and units of measurement
``inaccuracy'') for 34 quantities [5]. The Spanish obtained for each population studied in each
Scand J Clin Lab Invest Downloaded from informahealthcare.com by University of Hong Kong on 06/06/13
Society of Clinical Biochemistry and Molecular article, total number of subjects included and
Pathology (SEQC) has recommended quality strati®ed by sex when available, period of
speci®cations for imprecision, bias, bias in time covered (also expressed in days),
alternative sites and total error for 50 serum number of samples obtained for each subject
and urine quantities [6]. The quality speci®ca- studied, model used by each for calculating
tions in these works were derived from data the analytical coef®cient of variation, values
on biological variation; however, they were of between-run and within-run analytical
obtained by averaging the published data precision, numerator of the individuality
without assessing the reliability of this available index used by the author, type of population
information.
studied, health status, year of publication,
The aim of the present work was to prepare a
®rst author and journal of publication;
database with reliable information that could be
4. Relevant observations which could affect the
used to derive de®nitive quality speci®cations
For personal use only.
For each analyte, all articles having If w2 therefore, the estimations of components from
and/or classi®ed in MM group 1 were initially these articles were considered unsuitable for
considered to be suitable for exclusion in the deriving quality speci®cations.
®nal evaluation. Articles with other scores were Another example in which separate trends
included. were observed was in results from serum
Information from all eligible articles was creatinine. There were clear differences between
subsequently evaluated on the basis of their data from healthy subjects and data from
CVw values. These ®gures were arranged in patients suffering different pathologies. In
ascending order and inspected for evident these cases estimates of biological variation
tendencies that would indicate relationships were obtained from the group of healthy people
with other data ®elds (e.g. study period, sex to derive quality speci®cations.
of participants, health status, fasting condition). Table II shows the desirable quality speci®ca-
Scand J Clin Lab Invest Downloaded from informahealthcare.com by University of Hong Kong on 06/06/13
When such relationships were found, these tions for precision, bias and total error for all of
articles were separated from the general evalua- the quantities accepted from the bibliography
tion and their CVw and CVb were calculated studied.
individually.
When there were no trends requiring separa-
tion of articles into groups, the medians of the
within- and between-subject CV from all the D ISCU SSION
articles referring to the speci®c analyte were
calculated to determine the quality speci®ca- This work focused on establishing de®nitive
tions. analytical quality speci®cations for clinical
Desirable quality speci®cations for precision laboratories, but the database can be used for
(I), bias (B) and total error (TE) were calculated other purposes, such as establishment of delta
using the following formulae: checks, promoting common reference change
For personal use only.
TABLE I. S-cholesterol.
494
CVw CVb CVa II Nspec IF Mean Units n n (sex) Time Days S/S CVa types MM Year State
II~index of individuality; Nspec~number of specimens; If~index of ®duciability; S/S~samples/subjects; MM~mathematical model; M~male; f~female; h~hours;
d~days; w~weeks; m~months; y~years; WR~within-run; BR~between-run; dup~duplicate; trip~triplicate; quat~quadruplicate; H~health; P~pathology.
Current databases on biological variation
495
496 C. RicoÂs et al.
TABLE II.
Desirable speci®cations
Desirable speci®cations
Desirable speci®cations
Desirable speci®cations
Desirable speci®cations
S Tissue polypeptide antigen (TPA) 28.7 40.4 14.4 12.4 36.1 45.8
U Total catecholamines, concentration, 24 h 24.0 32.0 12.0 10.0 29.8 38.0
S Transferrin 3.0 4.3 1.5 1.3 3.8 4.8
S Triglyceride 21.0 37.2 10.5 10.7 28.0 35.1
S Triiodothyronine (T3) 8.7 14.4 4.4 4.2 11.4 14.3
S Urate 8.6 17.2 4.3 4.8 11.9 14.8
U Urate concentration, 24 h 24.7 22.1 12.4 8.3 28.7 37.1
Scand J Clin Lab Invest Downloaded from informahealthcare.com by University of Hong Kong on 06/06/13
information, those relevant only locally and 2 Fraser CG. The application of theoretical goals
those with little objective discussion. based on biological variation data in pro®ciency
testing. Arch Pathol Lab Med 1988; 112: 404 ± 15.
Several quantities (particularly hormones) 3 Fraser CG. Biological variation in clinical chem-
have been studied in very few articles and the istry: an update. Collated data, 1988 ± 1991. Arch
results found are highly discrepant. Profes- Pathol Lab Med 1992; 116: 916 ± 23.
sionals in laboratory medicine should be 4 SebastiaÂn-Gambaro MA, LiroÂn HernaÂndez PJ,
strongly encouraged to study the quantities fuentes-Arderiu X. Intra- and inter-individual
biological variability data bank. Eur J Clin Chem
for which results are discrepant, the 90
Clin Biochem (1997); 35: 845 ± 52 (also available at
described in only one paper and the numerous www.westgard.co).
quantities that have not been the subject of 5 Fraser CG, Hyltoft Petersen P, RicoÂs C, Haeckel R.
study. Proposed quality speci®cations for the imprecision
and inaccuracy of analytical systems for clinical
chemistry. Eur J Clin Chem Clin Biochem 1992; 30:
311 ± 7.
ACKNOWLEDGEMENTS 6 RicoÂs C, Alvarez V, JimeÂnez CV, HernaÂndez A,
Minchinela J, Perich C, SimoÂn M. Transferability
of results produced in the clinical laboratory. Quim
We express our thanks to Paco Campos and Clin (1996); 15: 442 ± 4; 1997; 16: 218.
Carlos GonzaÂlez-Oller for their time and 7 Fraser CG, Harris EK. Generation and application
dedication. We also thank Callum Fraser, Per of data on biological variation in clinical chemistry.
Hyltof Petersen and Jean Claude Libeer for Crit Rev Lab Sci 1989; 27: 409 ± 37.
their valuable help in the planning of this work. 8 Fraser CG, Browning MCK. The ``Index of
®duciability'' proposed for use in evaluation and
comparison of methods. Clin Chem 1988; 34:
1356 ± 7.
REFERENCES