Scand J Clin Lab Invest 2004; 64: 175 184

The reference change value: a proposal to interpret laboratory reports in serial testing based on biological variation
C. RIC O S, * * F . CA V A , * { J . V. GARCI A- LA RI O , * { A . HE RN A NDEZ, * N. I G LE SIA S, * C. V . J IM E NEZ, * } J . M IN C H I NE L A , * } C. P ERIC H , * , M. SIMO N , * * * M . V . D OM EN EC H & V . A L V A R E Z * { { *Laboratoris Clnics Hospital Vall dHebron, Barcelona; {Area de Laboratorio, Fundacion Hospital Alcorcon, Madrid; {Laboratorio de Analisis Clnicos, Hospital de Motril, Granada; Laboratori Clnic de LHospitalet, Barcelona; }Laboratori Clnic del Barcelones Nord, ` Barcelona; ,Laboratori Clnic Bon Pastor, Barcelona; **Laboratori Clnic Intercomarcal Vilafranca, Barcelona; {{Laboratori Clnic de Cornella de Llobregat, Barcelona, Spain; Laboratori Clnic Manso, Barcelona Ricos C, Cava F, Garca-Lario JV, Hernandez A, Iglesias N, Jimenez CV, Minchinela J, Perich C, Simon M, Domenech MV, Alvarez V. The reference change value: a proposal to interpret laboratory reports in serial testing based on biological variation. Scand J Clin Lab Invest 2004; 64: 175184. Background: A proposal to calculate and use the reference change value (RCV) as an objective guide for interpreting the numerical results obtained in clinical laboratory serial testing is introduced in this study. Methods: A database showing the results of a compilation of 191 publications on biological variation and including information on a number of analytes provided the standardized criterion based on biology for calculating the RCVs. Results: For each of the 261 analytes included in the study, the RCV was determined using Harriss formula, replacing analytical imprecision with the desirable specication of analytical quality based on half the within-subject biological variation at 95% probability levels. The result is a guide for a common criterion to identify clinically signicant changes in serial results. Conclusions: The RCV concept is an approach that can be offered by laboratories to assess changes in serial results. The RCV data in this study are presented as a point of departure for a widely applicable objective guide to interpret changes in serial results. Key words: Biological variation; quality assurance; reference change value Carmen Ricos, Unitat de la Qualitat, Laboratoris Clncis Hospital Universitari Vall dHebron, Passeig Vall dHebron 119, ES-08036, Barcelona, Spain. E-mail. cricos@hg.vhebron.es IN T R O D U C T I O N Data on biological variation (BV) are used in the medical laboratory for various purposes
*Members of the Committee for Quality Assurance and Laboratory Accreditation of the Spanish Society for Clinical Biochemistry and Molecular Pathology (SEQC).

such as dening quality specications, designing quality control rules, checking for errors (delta checks) using consecutive results [1, 2], and to provide information regarding patient health status. This study deals with the last application. Laboratory tests provide essential information for the medical decision process, but they are subject to intrinsic variability, which affects 175

DOI 10.1080/00365510410004885

176

C. Ricos et al. be minimized by providing standard instructions to patients before taking samples and by using written protocols for sample collection, transport, handling, centrifugation, etc. Analytical variation can be reduced by carefully maintaining equipment, by properly training personnel and by using written, standardized procedures. There are several general approaches to interpretation of laboratory results, e.g. comparison with cut-off values, comparison with population reference values, or comparison with previous results from the same individual. The conventional reference interval approach is the most commonly used aid to interpret laboratory results, but it has limitations. Marked individuality of the analyte explains some of the difculties in laboratory test interpretation with this method. The variations in the levels of an analyte in most individuals over time are smaller than the dispersion of the reference interval, and intra-individual BV is generally less than interindividual BV. The ratio of intra-individual to inter-individual BV is called the index of individuality (II). The lower the II, the greater is the inherent individuality of the analyte tested. The test results of a person can be outside of his/her healthy range but still lie within the population-based reference interval. In the case of two consecutive results, both may be within the reference interval but the difference between them may be clinically signicant for that particular person. In general terms, the usefulness of population reference intervals for monitoring patients is limited when the II is less than 0.6 and acceptable when the II is greater than 1.4 [3, 9 11]. When several consecutive results for one person are available (serial testing), there is another interesting approach for interpretation of laboratory results. In serial testing, results for a specic analyte are evaluated over time in a single individual. To condently report a signicant change in serial results, the difference in these results must exceed the difference explained by the inherent variation. Thus, assuming that some CVP is essentially included in the CVI and assuming that it can be minimized through adherence to good written standard operating procedures, good training and good laboratory practices, the total

the nal results generated. In general terms, proper interpretation of laboratory results requires the following: . Knowledge of the type and magnitude of variation that is included in the result. . Use of an approach that allows comparison of the result with criteria that dene a state of health. Several components of variation affect laboratory results. Some relate to laboratory activity (preanalytical and analytical variation) and others relate to the normal human biological variation in the levels of analytic constituents. Biological variation has two components: withinsubject and between-subject variation. Withinsubject or intra-individual BV is the random uctuation of a human body constituent around the homeostatic setting point. Additional sources of intra-individual variation include the inuences of the natural ageing process, gender, weight, diet, exercise, daily seasonal rhythms and, of course, the alterations caused by pathological processes. The individual homeostatic setting points also vary, and this variation is known as between-subject or interindividual variation [3, 4]. Biological varia- tion is generally expressed in terms of coefcients of variation (CVI and CVG for intra-individual and inter-individual variation, respectively). A great deal of effort has been dedicated to estimating the biological variation of the constituents analysed in clinical laboratories and a recent compilation of this data is now available for a large number of quantities [5 8]. Analytical and biological variations are random, can be considered Gaussian and are cumulative. The total variance associated with a laboratory result (s2T) is expressed by the following formula: S2T ~S2P zS2A zS2I with s2P being the preanalytical, s2A the analytical and s2I the intra-individual variation. Expressed in terms of coefcient of variation, the formula is as follows: 1=2 CVT ~ CV2P zCV2A zCV2I The lower the total variation in a test result, the easier it is to distinguish small but clinically important changes that guide medical decisionmaking. Preanalytical sources of variation can

The RCV proposal variation for each result in a single sample analysed once is: 1=2 CVT ~ CV2A zCV2I Because the variation involved is random and symmetrically distributed, we can with a certain probability derive the interval within which the values of variation will lie by using standard normal deviates (Z-scores) [4, 11]. To conclude that the difference between two serial results is signicant and could be biologically relevant (change in health status), the difference must be greater than the sum of the variation of each result. This difference is called the critical difference or the reference change value. This latter term emphasizes that the interpretation is based on the difference in a result from a single individual with respect to his/her previous results, instead of on population-based values. The reference change value (RCV) is expressed by the following formula [3, 4, 11, 12]: 1=2 RCV~ZP CV2A zCV2I z CV2A zCV2 I 1=2 RCV~21=2 ZP CV2A zCV2I in which RCV is the reference change value, ZP is the standard deviation for the appropriate probability of error, and CVA and CVI are the analytical and intra-individual biological coefcients of variation. In this paper we introduce a proposal to calculate and use the RCV as an objective, exible guide for identication of signicant differences between serial results that could have clinical relevance. Included in the formula are the following: . Intra-individual variation. Values are taken from an extensive compilation of data on components of biological variation. . Analytical variation. Desirable specications (goals) proposed for analytical variation are based on biological variation. . Z-scores at two levels of probability (95% and 99%) for bidirectional situations. MATERIALS AND METHODS To perform this study we used a database [5 8] showing the results of a compilation of 191 publications on biological variation and including information for 316 body uid constituents. Estimates of intra-individual and inter-individual

177

components of variation are given for 267 analytes in the healthy state and several pathologic situations. Additionally, other reported information, e.g. number of subjects, certain conditions (age, gender, fasting, state of health), study period, sampling time, number of samples per subject, and analysis results (mean values, analytical coefcient of variation) are provided. The present work only deals with the data from healthy subjects. For each analyte studied, the CVI values compiled were listed in ascending order to clearly identify extreme values. These values were segregated and carefully studied to determine whether there was a reasonable basis for eliminating them from the calculations. Examples of excluded data were values corresponding to non-fasting subjects for calculating CVI and CVG for glucose and triglycerides. The median of the remaining values was then calculated. For each analyte included, the reference change value was calculated using the abovementioned formula [1, 3], replacing the analytical imprecision with the desirable specication of analytical quality based on biological variation (0.5 CVI) [10 13].  1=2 RCV~ZP 21=2 CV2 z0:5 CVi 2 I When the range of probability of change is bidirectional and xed at 95%, ZP is equal to 1.96. RESULTS Data for the 261 analytes comprising the database are presented in Table I. This table includes CVI data, desirable specications for CVA, the calculated RCV values for these specications and 95% probability levels, using bidirectional Z-scores. D I S C US S I O N Laboratory test results are often used for monitoring patients, which involves reviewing these results over time. When analytes are strongly regulated, the intra-individual variation is narrower than the inter-individual variation (the ratio CVI/CVG is small) and comparison of serial results with the population-based reference range does not provide useful information

178

C. Ricos et al.

TABLE I. Proposal for reference change values (RCVs) according to biological variation, analytical quality specications and probability level using bidirectional Z-scores. Biological variation Analyte SSSUSSSSUPSSUSUSSSSSPSSSSUSUSSSUSSPSSSSSSBSSSSSSSUSSSSS11-Deoxycortisol 17-Hydroxiprogesterone 5Nucleotidase 5-HIAA concentration, 24 h a1-Acid glycoprotein a1-Antichymotripsine a1-Antitrypsin a1-Globulins a1-Microglobulin concentration, overnight a2-Antiplasmin a2-Globulins a2-Macroglobulin a2-Microglobulin output, overnight a-Amylase a-Amylase concentration, random a-Amylase, pancreatic a-Carotene Acid phosp, tartrate-resistant (TR-ACP) Acid phosphatase (ACP) Acid phosphatase activity, prostatic (PAP) Activate partial thromboplastine, time Adenosine deaminase (ADA) Alanine aminopeptidase Alanine aminotransferase Albumin Albumin concentration, rst morning Aldosterone Aldosterone concentration, 24 h Alkaline phosphatase Alkaline phosphatase, bone isoform Alkaline phosphatase, placental Ammonia output, 24 h Androstendione Angiotensin converting enzyme (ACE) Antithrombin III Apolipoprotein A1 Apolipoprotein B Ascorbic acid Aspartate aminotransferase a-Tocopherol b2-Microglobulin Basophils, count b-Carotene b-Cryptoxanthin b-Globulins Bilirubin, conjugated Bilirubin, total C Peptide C Telopeptide type I collagen C Telopeptide type I collagen/Creat C3 Complement C4 Complement CA 125 antigen CA 15.3 antigen CA 19.9 antigen CVI 21.3 19.6 23.2 20.3 11.3 13.5 5.9 11.4 33.0 6.2 10.3 3.4 29.0 9.5 94.0 11.7 35.8 8.0 8.9 33.8 2.7 11.7 4.1 24.3 3.1 36.0 29.4 32.6 6.4 6.2 19.1 24.7 11.5 12.5 5.2 6.5 6.9 25.8 11.9 13.8 5.9 28.0 36.0 36.7 10.1 36.8 25.6 9.3 8.0 35.1 5.2 8.9 29.2 5.7 24.5 CVA 10.7 9.8 11.6 10.2 5.7 6.8 3.0 5.7 16.5 3.1 5.2 1.7 14.5 4.8 47.0 5.9 17.9 4.0 4.5 16.9 1.4 5.9 2.1 12.2 1.6 18.0 14.7 16.3 3.2 3.1 9.6 12.4 5.8 6.3 2.6 3.3 3.5 12.9 6.0 6.9 3.0 14.0 18.0 18.4 5.1 18.4 12.8 4.7 4.0 17.6 2.6 4.5 14.6 2.9 12.3 RCV Desirable specication (CVA~0.5 CVI)
desirable

RCV95% 66.0 60.7 71.9 62.9 35.0 41.8 18.3 35.3 102.3 19.2 31.9 10.5 89.9 29.4 291.3 36.3 110.9 24.8 27.6 104.7 8.4 36.3 12.7 75.3 9.6 111.6 91.1 101.0 19.8 19.2 59.2 76.5 35.6 38.7 16.1 20.1 21.4 80.0 36.9 42.8 18.3 86.8 111.6 113.7 31.3 114.0 79.3 28.8 24.8 108.8 16.1 27.6 90.5 17.7 75.9

The RCV proposal

TABLE I. (Continued ). Biological variation Analyte SSUSUSSSSSSSSSSSSPSSSSSPatientUUUPSUUPPB(B)PlatPBUSPPPPSP(B)ErythrSSUSUSSSSSCA 549 antigen Calcium Calcium concentration, 24 h Calcium ionized Calcium output, 24 h Carbohydrate decient transferrin Carcinoembryonic antigen (CEA) Ceruloplasmin Chloride Cholesterol Cholinesterase Cholinesterase, immunoreactive Cholinesterase, activity CK MB% CK MB, activity CK MB, mass Copper Copper Cortisol C-Propeptide type 1 procollagen C-Reactive protein Creatine kinase (CK) Creatinine Creatinine clearance Creatinine concentration, 24 h Creatinine output, 24 h C-Telopeptide type I collagen/creat., 2nd morning Cysteine Dehydroepiandrosterone sulphate Deoxipyridinoline/Creatinine, 24 h Deoxipyridinoline/Creatinine, morning spot Dipeptidyl-peptidase IV Elastase Eosinophils, count Epinephrine Epinephrine Erythrocytes, count Oestradiol Oestradiol Factor V coagulation Factor VII coagulation Factor VIII coagulation Factor X coagulation Ferritin Fibrinogen Folate Folate Follicle stimulating hormone (males) Free oestradiol Free testosterone Free testosterone Free thyroxine (FT4) Free triiodothyronine (FT3) Fructosamine Galactosy hydroxylisine c-Globulins CVI 9.1 1.9 27.5 1.7 26.2 7.1 12.7 5.7 1.2 6.0 7.0 6.4 5.4 6.9 19.7 18.4 4.9 8.0 20.9 8.2 52.6 22.8 4.3 13.6 24.0 11.0 23.5 5.9 11.6 15.4 26.5 8.2 13.6 21.0 25.3 48.3 3.2 30.4 18.1 3.6 6.8 4.8 5.9 14.9 10.7 12.0 24.0 8.7 38.6 9.3 51.7 7.6 7.9 3.4 11.8 14.6 CVA 4.6 1.0 13.8 0.9 13.1 3.6 6.4 2.9 0.6 3.0 3.5 3.2 2.7 3.5 9.9 9.2 2.5 4.0 10.5 4.1 26.3 11.4 2.2 6.8 12.0 5.5 11.8 3.0 5.8 7.7 13.3 4.1 6.8 10.5 12.7 24.2 1.6 15.2 9.1 1.8 3.4 2.4 3.0 7.5 5.4 6.0 12.0 4.4 19.3 4.7 25.9 3.8 4.0 1.7 5.9 7.3 RCV Desirable specication (CVA~0.5 CVI)
desirable

179

RCV95% 28.2 5.9 85.2 5.3 81.2 22.0 39.4 17.7 3.7 18.6 21.7 19.8 16.7 21.4 61.1 57.0 15.2 24.8 64.8 25.4 163.0 70.7 13.3 42.1 74.4 34.1 72.8 18.3 35.9 47.7 82.1 25.4 42.1 65.1 78.4 149.7 9.9 94.2 56.1 11.2 21.1 14.9 18.3 46.2 33.2 37.2 74.4 27.0 119.6 28.8 160.2 23.6 24.5 10.5 36.6 45.2

180

C. Ricos et al.

TABLE I. (Continued ). Biological variation Analyte SSS(B)ErythrBSSP,SSSSSBBSPUUSSSSSSS(B)LeucSSSBSSSSSSPSSBBSSSSSB(B)Erythr(B)LeucSUSU(B)Erythr(B)Erythr (B)Erythrc-Glutamyltransferase Globulins, total Glucose Glucose 6 phosphate dehydrogenase (G6PDH) Glutathion peroxidase Glycated albumin Glycated total protein Haptoglobin HDL cholesterol HDL1 cholesterol HDL2 cholesterol HDL3 cholesterol Hematocrit Haemoglobin Haemoglobin A1 C Homocysteine Hydroxiproline/creatinine Hydroxiproline/minute, night urine Hydroxybutyrate dehydrogenase Immunoglobulin A Immunoglobulin G Immunoglobulin M Immunoglobulins k chains Immunoglobulins l chains Insulin Interferon receptor Interleukin-1b Interleukin-8 Iron Lactate Lactate dehydrogenase (LDH) Lactate dehydrogenase 1 isoform (LD1) Lactate dehydrogenase 2 isoform (LD2) Lactate dehydrogenase 3 isoform (LD3) Lactate dehydrogenase 4 isoform (LD4) Lactate dehydrogenase 5 isoform (LD5) Lactoferrin LDL cholesterol LDL cholesterol direct LDL receptor mRNA Leucocytes, count Lipase Lipoprotein (a) Lutein Luteinizing hormone Lycopenen Lymphocytes, count Magnesium Magnesium Magnesium Magnesium concentration, 24 h Magnesium ionized Magnesium output, 24 h Mean corpuscular haemoglobin (HCM) Mean corpuscular haemoglobin conc (MCHC) Mean corpuscular volume (MCV) CVI 13.8 5.5 4.9 32.8 7.2 5.2 0.9 20.4 7.1 5.5 15.7 7.0 2.8 2.8 2.1 9.0 25.9 36.1 8.8 5.4 4.5 5.9 4.8 4.8 21.1 14.0 30.0 24.0 26.5 27.2 8.6 6.3 4.9 4.8 9.4 12.4 11.8 8.3 6.5 21.5 10.9 23.1 8.5 23.7 14.5 43.1 10.4 5.6 18.3 3.6 45.4 1.9 38.3 1.6 1.7 1.3 CVA 6.9 2.8 2.5 16.4 3.6 2.6 0.5 10.2 3.6 2.8 7.9 3.5 1.4 1.4 1.1 4.5 13.0 18.1 4.4 2.7 2.3 3.0 2.4 2.4 10.6 7.0 15.0 12.0 13.3 13.6 4.3 3.2 2.5 2.4 4.7 6.2 5.9 4.2 3.3 10.8 5.5 11.6 4.3 11.9 7.3 21.6 5.2 2.8 9.2 1.8 22.7 1.0 19.2 0.8 0.9 0.7 RCV Desirable specication (CVA~0.5 CVI)
desirable

RCV95% 42.8 17.0 15.2 101.6 22.3 16.1 2.8 63.2 22.0 17.0 48.7 21.7 8.7 8.7 6.5 27.9 80.3 111.9 27.3 16.7 13.9 18.3 14.9 14.9 65.4 43.4 93.0 74.4 82.1 84.3 26.7 19.5 15.2 14.9 29.1 38.4 36.6 25.7 20.1 66.6 33.8 71.6 26.3 73.4 44.9 133.6 32.2 17.4 56.7 11.2 140.7 5.9 118.7 5.0 5.3 4.0

The RCV proposal

TABLE I. (Continued ). Biological variation Analyte (B)PlatBSSUBU(B)PlatPUSSUUBBBSUUPatientSPBBBS(B)LeucUUSSSSPSPUUPSPUBBSSSSSSSSBPSMean platelet volume (MPV) Monocytes, count Mucinous carcinoma-associated antigen (MCA) Myoglobin N-acetyl glucosaminidase concentration,overnight Neutrophils, count Nitrogen, output Norepinephrine Norepinephrine N-telopeptide type I collagen/creatinine Osmolality Osteocalcin (z1 trab) Oxalate concentration, 24 h Oxalate output, 24 h pCO2 pH (pH units) pH [Hz] Phosphate Phosphate concentration, 24 h Phosphate output, 24 h Phosphate tubular reabsorption Phospholipids Plasminogen Platelet distribution wide (PDW) Plateletcrit Platelets Potassium Potassium Potassium concentration, 24 h Potassium output, 24 h Prealbumin Procollagen type I C-terminal Procollagen type I N-terminal Prolactin (men) Prolyl endopeptidase Prostatic specic antigen (PSA) Protein C Protein concentration, 24 h Protein output, 24 h Protein S Protein, total Prothrombin, time Pyridinoline/creatinine, morning spot Pyruvate Red cell distribution wide (RDW) Reticulocyte highly uorescent, count Reticulocyte low uorescent, count Reticulocyte medium uorescent, count Reticulocyte, count Retinol Rheumatoid factor Riboavin SCC antigen Selenium Selenium Sex hormone binding globulin (SHBG) CVI 4.3 17.8 10.1 13.9 48.6 16.1 13.9 9.5 19.5 20.5 1.3 7.2 44.0 42.5 4.8 0.2 3.5 8.5 26.4 18.0 2.7 6.5 7.7 2.8 11.9 9.1 4.8 13.6 27.1 24.4 10.9 7.8 6.8 6.9 16.8 14.0 5.8 39.6 35.5 5.8 2.7 4.0 8.7 15.2 3.5 10.0 1.6 13.0 11.0 14.8 8.5 5.2 39.4 12.0 12.0 12.1 CVA 2.2 8.9 5.1 7.0 24.3 8.1 7.0 4.8 9.8 10.3 0.7 3.6 22.0 21.3 2.4 0.1 1.8 4.3 13.2 9.0 1.4 3.3 3.9 1.4 6.0 4.6 2.4 6.8 13.6 12.2 5.5 3.9 3.4 3.5 8.4 7.0 2.9 19.8 17.8 2.9 1.4 2.0 4.4 7.6 1.8 5.0 0.8 6.5 5.5 7.4 4.3 2.6 19.7 6.0 6.0 6.1 RCV Desirable specication (CVA~0.5 CVI)
desirable

181

RCV95% 13.3 55.2 31.3 43.1 150.6 49.9 43.1 29.4 60.4 63.5 4.0 22.3 136.4 131.7 14.9 0.6 10.8 26.3 81.8 55.8 8.4 20.1 23.9 8.7 36.9 28.2 14.9 42.1 84.0 75.6 33.8 24.2 21.1 21.4 52.1 43.4 18.0 122.7 110.0 18.0 8.4 12.4 27.0 47.1 10.8 31.0 5.0 40.3 34.1 45.9 26.3 16.1 122.1 37.2 37.2 37.5

182

C. Ricos et al.

TABLE I. (Continued ). Biological variation Analyte S(B)Erythr(B)LeucUUS(B)ErythrSSalivaUSSSSSSSUSSSSSUUSUUU(B)Erythr (B)Erythr SPSSSPSodium Sodium Sodium Sodium concentration, 24 h Sodium output, 24 h Superoxide dismutase Superoxide dismutase T3-uptake Testosterone Testosterone Testosterone Thyroglobulin Thyroid stimulating hormone (TSH) Thyroxin binding globulin (TBG) Thyroxine (T4) Tissue polipeptide specic antigen (TPS) Tissue polypeptide antigen (TPA) Total catecholamines, concentration, 24 h Transferrin Triglyceride Triiodothyronine (T3) Tumor necrosis factor-a Urate Urate concentration, 24 h Urate output, 24 h Urea Urea concentration, 24 h Urea output, 24 h Vanilmandelic acid concentration, 24 h Vitamin B12 Vitamin B6 VLDL cholesterol Von Willebrand factor Water Zeaxanthine Zinc Zinc CVI 0.7 1.8 51.0 24.0 28.7 17.1 12.3 4.5 17.3 25.0 9.3 13.0 19.7 6.0 6.0 36.1 28.7 24.0 3.0 20.9 8.7 43.0 8.6 24.7 18.5 12.3 22.7 17.4 22.2 5.2 1.4 27.6 0.001 3.1 34.7 9.3 11.0 CVA 0.4 0.9 25.5 12.0 14.4 8.6 6.2 2.3 8.7 12.5 4.7 6.5 9.9 3.0 3.0 18.1 14.4 12.0 1.5 10.5 4.4 21.5 4.3 12.4 9.3 6.2 11.4 8.7 11.1 2.6 0.7 13.8 0.0 1.6 17.4 4.7 5.5 RCV Desirable specication (CVA~0.5 CVI)
desirable

RCV95% 2.2 5.6 158.1 74.4 88.9 53.0 38.1 13.9 53.6 77.5 28.8 40.3 61.1 18.6 18.6 111.9 88.9 74.4 9.3 64.8 27.0 133.3 26.7 76.5 57.3 38.1 70.3 53.9 68.8 16.1 4.3 85.5 0.0 9.6 107.5 28.8 34.1

on the patients health status [1, 2]. This is true for many common analytes [5 8]; thus other criteria are needed to assess the patients status during serial testing. One approach that can be usesd to interpret changes in serial results is based on medical criteria. This information is usually obtained through a system of inquiries directed to clinicians to determine at what magnitude a difference in serial results indicates a clinical change in patient status. This system has shown important deciencies and has been recently revised by Sandberg & Thue [14] with a more restricted focus related to specic clinical situations. Sto ckl

et al. [15] realized that doctors opinions are highly inuenced by the analytic state of the art and that the real variations caused by biological variation are often ignored. It would be of help to the clinician if the laboratory could provide an objective means for interpreting the information contained in analytical reports. One approach towards achieving this is through the RCV concept. We introduce a proposal to calculate and use the RCV according to the standardized criterion based on biology. The general expression for the RCV is: 1=2 RCV~21=2 ZP CV2 zCV2 A I

The RCV proposal All the components of this formula should be closely examined to justify the proposal as a standardized criterion. Intra-individual variation The CVI should not be interpreted as an estimate of a true value, but is actually a mean of obviously different intra-individual values. The values for intra-individual variation in this proposal are also the mean of all published values in healthy subjects for each analyte, compiled in our database. Nevertheless, there are some theoretical disadvantages to the use of BV data from healthy subjects for estimating RCVs. Intra-individual variation in acute pathologic states may be larger than that seen in healthy individuals. For this reason the use of RCV data derived from healthy individuals may lead to false-positive detection of changes in patients. Nevertheless, it is commonly considered better to register a change and then decide that it is unimportant clinically, than to miss a change in a patients condition. Analytical variation The value assigned to this component of the formula may be different for each laboratory since it is established by internal quality control protocols. Consequently, among laboratories acting independently, the RCV obtained by applying the formula will also vary. Nevertheless, to promote harmony in laboratory medicine, which is the aim of our scientic society, we support the use of unied criteria for decision-making. When laboratories maintain their analytical CV within the limits derived from biological variation, there will be a signicant advance along this line. Then the reference change concept will only depend on intra-individual CVs and the resultant values will be common to all laboratories for each analyte tested. In this proposal, we use the quality specications based on biology for the CVA in the formula. Desirable performance is dened by CVA~0.5 CVI. This means that intrinsic test result variability due to biology has been increased by 11.8%. For analytes in which desirable standards are difcult to meet, a minimum goal (CVA~0.75 CVI) has been established, and for analytes in which desirable

183

standards are easy to meet, an optimum goal (CVA~0.25 CVI) has been established [16, 17]. When these goals are applied to RCV data, we obtain a guide for desirable, minimum and optimum RCV values. When a laboratorys variation (precision) is lower than the desirable or optimum goal, the laboratory is able to establish that small numerical changes in consecutive results are signicant, thereby providing higher clinical sensitivity. However, when a laboratorys variation does not meet the desirable or even the minimum goal, the numerical difference in consecutive results will have to be larger for the laboratory to mark the difference as signicant, and it will be difcult to detect changes in consecutive results that are biologically relevant. At present, most clinical laboratories make use of the Laboratory Information Systems (LIS). One of the tasks of the LIS is to help in the interpretation of laboratory results through comparison with reference values, cut-off points, etc. Since many laboratory results are used for monitoring, it would be useful for all LIS to have the capability to introduce RCV data and provide ags or information in laboratory reports on the signicance of changes in an individuals serial results. Each laboratory would adjust the RCV values according to its own CVA and the interest in specic probability ranges. We present the RCV data in this work as a point of departure, an objective guide to interpret changes in serial results that is widely applicable. This guide is also intended to make laboratory professionals more aware of the repercussions of their performance on medical decisions.

A C K NO W L E D G E M E NT S We thank Celine language advice. Cavallo for English

R E F E RE N C E S
1 Ricos C, Andreu A, Schwartz S. Metodos de deteccion del error extraanaltico. Rev Diag Biol 1988; 37: 269 71. 2 Fraser CG, Stevenson HP, Kennedy IMG. Biological variation data are necessary prerequisites for objective autoverication of clinical laboratory data. Accreditation Qual Assur 2002; 7: 445 60.

184

C. Ricos et al.
principles to practice. Washington: AACC Press; 2001. p. 91 116. Harris EK, Boyd JC. Statistical bases of reference values in laboratory medicine. New York: Marcel Dekker Inc; 1995. Harris EK, Yasaka T. On the calculation of a reference change for comparing two consecutive measurements. Clin Chem 1983; 29: 25 30. Hyltoft Petersen P, Fraser CG, Kallner A, Kenny D. Strategies to set global analytical quality specications in laboratory medicine. Scand J Clin Lab Invest 1999; 59: 475 585. Sandberg S, Thue G. Quality specications derived from objective analyses based upon clinical needs. Scand J Clin Lab Invest 1999; 59: 531 4. Stockl D, Baadenhuijsen H, Fraser CG, Libeer JC, Hyltoft Petersen P, Ricos C. Desirable routine analytical goals for quantities assayed in serum. Discussion paper from the members of the External Quality Assessment (EQA) Working Group on analytical goals in laboratory medicine. Eur J Clin Chem Clin Bichem 1995; 33: 157 69. Fraser CG, Hyltoft Petersen P, Libeer JC, Ricos C. Proposals for setting generally applicable quality goals solely based on biology. Ann Clin Biochem 1997; 34: 8 12. Fraser CG. Quality specications. In: Biological variation: from principles to practice. Washington: AACC Press; 2001. p. 29 66.

3 Fraser CG, Harris EK. Generation and application of data on biological variation in clinical chemistry. Crit Rev Clin Lab Sci 1989; 27: 409 37. 4 Fraser CG. Change in serial results. In: Biological variation: from principles to practice. Washington: AACC Press; 2001. p. 67 90. 5 Ricos C, Alvarez V, Cava F, Garca-Lario JV, Hernandez A, Jimenez CV, Minchinela J, Perich M, Simon M. Current databases on biological variation: pros, cons and progress. Scand J Clin Lab Invest 1999; 59: 491 500. 6 Sociedad Espanola de Bioqumica Clnica y Patologa Molecular. Comite de Garanta de la Calidad y Acreditacion de Laboratorios . Especi caciones de la calidad analtica en laboratorios con distintos niveles de recursos. Quim Clin 2000; 19: 219 36. 7 Ricos C, Alvarez V, Cava F, Garca-Lario JV, Hernandez A, Jimenez CV, Minchinela J, Perich M, Simon M. Biological variation database. http:// www.westgard.com/guest21/htm 8 Sociedad Espanola de Bioqumica Clnica y Patologa Molecular. Comite de Garanta de la Calidad y Acreditacion de Laboratorios. Base de datos de variacion biologica. http://www.seqc.es/ bd/vb.htm 9 Harris EK. Effects of intra and interindividual variation on the appropriate use of normal ranges. Clin Chem 1974; 20: 1535 42. 10 Fraser CG. The utility of population based reference values. In: Biological variation: from

11 12 13

14 15

16

17

Received: 5 July 2003 Accepted: 14 January 2004