Estimation of the Minimal Preanalytical Uncertainty for 15 Clinical Chemistry Serum Analytes

8/13/2019 Estimation of the Minimal Preanalytical Uncertainty for 15 Clinical Chemistry Serum Analytes

1/7

Estimation of the Minimal Preanalytical Uncertainty for 15Clinical Chemistry Serum Analytes

Marit Sverresdotter Sylte,1* Tore Wentzel-Larsen,2 and Bjrn J. Bolann1,3

BACKGROUND: We sought a model to estimate preana-

lytical uncertainty of blood samples collected and pro-cessed by using optimal procedures.

METHODS: Optimal preanalytical handling of bloodsamples included use of a loosely fastened tourni-

quet, wide bore needles, recommended clottingtime and centrifugation speed, and minimal storage

before analysis. Blood was collected from each arm

of 20 volunteers into 2 rapid-serum tubes and 2serum-separation tubes. Linear mixed-effects mod-els were used to estimate the between-venipuncture

SD, the preanalytical SD (excluding venipuncture),the measurement repeatability SD, and system-

atic differences between the tubes and betweenvenipunctures.

RESULTS: No significant systematic differences werefound between successive venipunctures. However,

statistically significant mean differences were seenbetween serum-separation tubes and rapid-serum

tubes for 7 of the 15 analytes. The preanalytical SD(excluding venipuncture) for lactate dehydrogenase

(3.2 U/L, 95% CI 2.83.7) was significantly higherthan the SD for measurement repeatability (1.9 U/L,

95% CI 1.72.1). For potassium both the preanalyti-cal SD (excluding venipuncture) (0.092 mmol/L,

95% CI 0.0800.11) and the between-venipunctureSD (0.075 mmol/L, 95% CI 0.048 0.12) were signif-

icantly higher than the measurement-repeatabilitySD (0.031 mmol/L, 95% CI 0.0280.035). For glu-

cose the between-venipuncture SD (0.20 mmol/L,95% CI 0.140.27) was significantly higher than

the preanalytical SD (excluding venipuncture) (0.07mmol/L, 95% CI 0.06 0.08), and the measure-

ment repeatability SD (0.057 mmol/L, 95% CI0.0510.064).

CONCLUSIONS: By applying linear mixed-effects models

we have estimated the minimal preanalytical uncer-tainty that will influence all patient results.

2010 American Association for Clinical Chemistry

Preanalytical uncertaintyis attributable to variations inblood sample collection and sample handling that oc-

cur before the blood sample is analyzed (1 ). Inthepre-

analytical phase of clinical chemistry analyses manysources may contribute to erroneous results (2 ). Pre-

analytical errors, such as missing identification of thepatient, use of the wrong blood tubes, and ordering of

the wrong tests, should not be confused with the quan-tifiable preanalytical uncertainty that is due to varia-

tion in current laboratory practice such as the use ofdifferent blood tubes, variation in clotting time (pre-

centrifugation phase), and variation in centrifugalforce and storage time(3 ).

The use of variable preanalytical routines in cur-rent practice may increase uncertainty and lead to a

bias in the results(3 )and possible alterations in inter-pretation of the test. In some cases the preanalytical

uncertainty may be the dominant variability.The Guide to Expression of Uncertainty in Measure-

ment(GUM)4 establishes general rules for expressinguncertainty in measurement (4 ), including those for

converting the uncertainty estimates into standardform, combining them, and calculating the combined

uncertainty.In some preanalytical studies, ANOVA and Bland-

Altman plots have been used to compare an ideal pre-analytical situation to an alternative situation (58). In

other studies(911 )the preanalytical uncertainty hasbeen estimatedaccording to a specific practice,without

addressing the effect expected for an optimal preana-lytical practice.

1 Laboratory of Clinical Biochemistry and 2 Centre for Clinical Research, Hauke-land University Hospital, Helse Bergen HF, Bergen, Norway; 3 Institute ofMedicine, University of Bergen, Bergen, Norway.

* Address correspondence to this author at: Laboratory of Clinical Biochemistry,Haukeland University Hospital, Helse Bergen HF, Postbox 1, N-5021 Bergen,Norway. Fax 47-55973115; e-mail [email protected].

Received February 24, 2010; accepted May 16, 2010.

Previously published online at DOI: 10.1373/clinchem.2010.1460504 Nonstandard abbreviations: GUM, Guide to Expression of Uncertainty in Mea-

surement; SST, serum-separation tube; RST, rapid-serum tube; ALP, alkalinephosphatase; ALT, alanine aminotransferase; CK, creatine kinase; GGT, -glutamyltransferase; HDL-C, HDL-cholesterol; LDH, lactate dehydrogenase; TC,total cholesterol; TG, triglycerides; H-index, hemoglobin index.

Clinical Chemistry56:813291335 (2010)

General Clinical Chemistry

1329


2/7

We have previously presented a model for con-structing a preanalytical uncertainty budget for clinicalchemistry analyses (3 ). In this model, the effects of var-ious alternative preanalytical practices were estimated,compared to an optimal practice for handling the

blood samples. However, the model did not determinethe uncertainty in the optimal practice itself.

The aim of our current study was to estimate thepreanalytical uncertainty when the phlebotomy andthe sample handling are performed optimally accord-ing to existing standards (12, 13). This minimal pre-analytical uncertainty will influence all patient results,and may be used as a standard of reference for evalua-tion of preanalytical uncertainty in current practice.

To estimatethe uncertainty caused by the phlebot-omy procedure itself, we performed venipuncture onboth arms of each participant. Collection of blood

samples into 2 different types of blood tubes enabled usto detect any differences in preanalytical variation. Lin-ear mixed-effects models (14)were used to estimatethe between-venipuncture variation, and the preana-lytical variation (except for between-venipuncturevariation), together with any fixed effects, and themeasurement repeatability for 15 clinical chemistryanalytes.

Materials and Methods

SAMPLE HANDLING AND ANALYSIS

After we obtained approval for the study from the local

ethics committee and informed and written confirma-tion from study participants in accordance with theHelsinki Declaration, we performed phlebotomy on 20nonfasting volunteers from our laboratory in accor-dance with a standardized procedure(13 ).

To minimize trauma to erythrocytes, we usedwide-bore (21-gauge, Becton Dickson) needles for ve-nipunctures. All phlebotomywas performed between 9and 10 AM by the same medical technician, with eachparticipant in a sitting position forapproximately 10 minat ambient temperature. The tourniquet was loosely fas-tened and released after 1 min, immediately whenblood appeared. Repeated clenching and unclenching of

the fist wasnot allowed. The blood samples were collectedinto plastic serum-separation Vacutainer (Becton Dick-son) gel tubes: serum-separation tubes (SST)-II advancetubes 3.5 mL containing silica clot activator and rapid-serum tubes (RST) 4.0 mL containing a thrombin-basedmedical clotting agent.

Two tubes of each type were collected in randomorder from each arm, i.e., 4 tubes from each arm, for atotal of 8 tubes from each person. The tubes were com-pletely filled, mixed gently by 5 inversions, and put in avertical position. The mean venipuncture duration forboth arms was 3 min (range 27 min). The right arm

was always phlebotomized first, and therefore the clot-ting time wasalways about 12min longer for the tubescollected from the right arm. The clotting time wasstandardized according to the manufacturers recom-mendation; for the RST tubes it varied from 5 to 8 min,

and for the SSTs from 30 to 33 min. As recommendedby the manufacturer, the tubes were centrifuged for 10min at 1600gin a swing-out centrifuge (Kubota 5930,Kubota Corporation) at 20 C.

Immediately after centrifugation, the serum fromeach gel tube was separated into 2 secondary tubes.Thus the tubes from each person were analyzed in du-plicate, randomly under repeatability (within-run)conditions, on Roche Modular Analytics SWA (serumwork area) instruments by photometric methods(Roche Diagnostics).

Albumin was measured with the bromcresol green

method; alkaline phosphatase (ALP) liquid accordingto IFCC; alanine aminotransferase (ALT) according toIFCC with pyridoxal phosphate activation; calciumwitho-cresolphthalein; creatine kinase (CK) liquid ac-cording to IFCC; creatinine with creatinine plus;-glutamyltransferase (GGT) liquid according toIFCC; glucose with Gluco-quant Glucose/HK; HDL-cholesterol (HDL-C) with HDLC3 (HDL-C plus 3rdgeneration), no pretreatment; lactate dehydrogenaseliquid (LDH) according to IFCC; magnesium withxylidyl blue; potassium and sodium with ISE(ion-selective electrode) indirect; total cholesterol(TC) with CHOD-PAP (cholesterol oxidase phenol

4-aminophenazone peroxidase); and triglycerides(TG) with GPO-PAP (glycerol phosphate oxidase4-aminophenazone). A photometric method, the he-moglobin index (H-index), wasused for measuring he-molysis; 100 H-index units correspond to approxi-mately 0.06 mmol/L (0.1 g/dL) of hemoglobin. Thesamples from the RSTs were analyzed on average 33min (range 21 62 min) after phlebotomy and from theSSTs on average 61 min (range 4695 min) afterphlebotomy.

STATISTICAL ANALYSIS

Data were analyzed by useof linear mixed-effects mod-

els(14). Mixed-effects models allow clustered multi-level data and also allow separate estimates of fixed ef-fects and random effects. Fixed effects representsystematic differences due to covariates e.g., betweenarms, andSSTvs RSTtubes, similarlyto regression mod-elsfor unclustereddata. Random effectswere expressedasSDs for random variation between groups at each level,e.g., between persons, arm from which sample was col-lected, tubes from each arm, and, finally, duplicates fromeach tube, representing measurement repeatability. Weused the package Linear and Nonlinear Mixed EffectsModels in R (The R Foundation for Statistical Comput-

1330 Clinical Chemistry56:8 (2010)


3/7


4/7

CK, creatinine, HDL-C, TC, and TG the between-venipuncture, preanalytical (excluding venipuncture)and measurement-repeatability SDs were of similar

magnitude, and generally demonstrated overlappingCIs.

The between-venipuncture SD for sodium re-sulted in a wide 95% CI; owing to this statisticalinstability, and as recommended elsewhere (14), themodel was simplified by combining the between-venipuncture and preanalytical (excluding venipunc-ture) random effects to a common estimate SD 0.26(95% CI 0.14 0.47) mmol/L (Table 3). The CI for pre-analytical SD (excluding venipuncture) showed insta-bility for GGT measured in SST tubes, magnesiummeasured in RST tubes, and sodium measured in bothRST and in SST tubes (Table 2).

Sensitivity analysis was done by excluding possibleoutliers, 1 outlier each for ALT, magnesium, and TC.Because the outliers had no influence on the estimatedSDs(Table 3) or on thedifferencesdisplayed in Table 1,they were not excluded in the main analysis.

Discussion

To reduce random preanalytical variation it is vital toidentify the contributions of the various uncertaintysources, from venipuncture to analysis. Linear mixed-effects models (14) constitute a well-developed and

suitable method to identify different sources of varia-tion, and fixed effects in a data set when both effectsmay be present.

The procedures applied for optimal collectionand handling of the blood samples before analysisare based on guidelines(12, 13 ). Complete clottingnormally occurs within 30 60 min at room temper-ature, but it could be accelerated to 5 min with theuse of blood tubes containing thrombin (12 ), suchas the RST tubes used in this study. For centrifugetime and relative centrifugal force the tube manufac-turers recommendations should be consulted(12 ).Serum should be separated from the erythrocytes assoon as possible, or within a maximum limit of 2 hafter the phlebotomy(12 ). Temperature and time ofstorage may influence the stability of biochemical

components in serum(12 ).The uncertainty chain starts with the choice of

blood-collection tube. We found a fixed effect repre-senting a significant, systematic difference in the resultsfrom the SST vs the RST tubes for some analytes (Table1). This difference may be due to the SSTs havinglonger clotting time, the RST tubes being analyzedabout 30 min sooner than the SSTs, and the fact thatthe tubes have different clotting agents. The estimatedmean difference between the SST and the RST tubes forglucose (0.16 mmol/L, 95% CI 0.18 to 0.13) wassignificant, meaning that the choice of correct tubes is

Table 2. Preanalytical variation (excluding between-venipuncture) estimated for use

of the RST tubes and SST tubes.

Preanalytical variation, excluding venipuncture, SD (95% CI)

RST tubes SST tubes

Albumin, g/L 0.32 (0.220.45) 0.39 (0.290.53)

ALP, U/L 0.52 (0.370.72) 0.74 (0.570.97)

ALT, U/L 0.47 (0.161.41) 0.35 (0.101.55)

Calcium, mmol/L 0.012 (0.0100.018) 0.009 (0.0060.015)

CK, U/L 0.93 (0.671.28) 1.33 (1.001.78)

Creatinine, mol/L 0.51 (0.340.75) 0.70 (0.490.99)

GGT, U/La 0.33 (0.170.64)

Glucose, mmol/L 0.06 (0.040.08) 0.07 (0.060.10)

HDL-C, mmol/L 0.012 (0.0080.020) 0.013 (0.0080.021)

LDH, U/L 2.8 (2.13.6) 2.7 (2.03.6)

Magnesium, mmol/La 0.0024 (0.00040.013)

Potassium, mmol/L 0.096 (0.0760.12) 0.089 (0.0710.11)

TC, mmol/L 0.038 (0.0230.063) 0.057 (0.0410.078)

TG, mmol/L 0.005 (0.0020.014) 0.012 (0.0080.020)

a Due to a wide 95% CI the preanalytical SDs (excluding between-venipuncture) for GGT measured in SST tubes, magnesium measured in RST tubes, and for sodium

measured in both RST and SST tubes are not shown.



5/7

important to minimize the preanalytical uncertainty,even when following an optimal protocol. Our firststudy (3 ) showed that measured glucose concentra-tions were significantly decreased by prolonged clot-ting time. Because of glycolysis the concentration ofglucose decreases with prolonged contact with the clot(1 ).

The H-index was significantly higher in the SSTtubes, and this result may explain the significant differ-ence between the tubes for potassium, LDH, and mag-nesium, with the SSTs showing higher concentration.

It might be expected that the preanalytical varia-tion (excluding venipuncture) may be different for thedifferent tube types because of different clotting times.

Table 3. Sample analyte data.

Concentration,mean (range)

Between-

venipuncturevariation

Preanalyticalvariation

(excluding

between-venipuncture)

Measurementrepeatability

Preanalyticalincludingbetween-

venipunctureSDa

Preanalyticaland

analytical SDcombinedb

Albumin, g/L 46.1 (41.650.4) 0.47 (0.320.69)c 0.38 (0.310.45) 0.33 (0.300.37) 0.60 0.69

1.0d 0.8 0.7

ALP, U/L 71.1 (46107) 0.87 (0.591.27) 0.64 (0.540.76) 0.49 (0.440.55) 1.08 1.18

1.2 0.9 0.7

ALT, U/L 27.8 (1067) 0.47 (0.260.86) 0.32 (0.0941.076) 1 .16 (1.041.29) 0.57 1.29

1.7 1.1 4.2

Calcium, mmol/L 2.40 (2.202.53) 0.012 (0.0100.019) 0.012 (0.0100.016) 0.015 (0.0140.017) 0.017 0.023

0.5 0.5 0.6

CK, U/L 118.3 (52191) 1.51 (1.042.19) 1.02 (0.851.23) 0.97 (0.871.08) 1.83 2.07

1.3 0.9 0.8Creatinine, mol/L 70.4 (5792) 0.90 (0.621.31) 0.58 (0.460.72) 0.67 (0.600.75) 1.07 1.26

1.3 0.8 1.0

GGT, U/L 26.2 (8160) 0.42 (0.280.65) 0.24 (0.130.46) 0.61 (0.550.68) 0.49 0.78

1.6 0.9 2.3

Glucose, mmol/L 5.2 (4.048.50) 0.20 (0.140.27) 0.07 (0.060.08) 0.057 (0.0510.064) 0.21 0.22

3.8 1.3 1.1

HDL-C, mmol/L 1.82 (1.033.07) 0.022 (0.0150.032) 0.014 (0.0100.018) 0.019 (0.0170.021) 0.026 0.032

1.2 0.8 1.0

LDH, U/L 177.3 (97223) 2.4 (1.53.9) 3.2 (2.83.7) 1.9 (1.72.1) 4.0 4.4

1.4 1.8 1.1

Magnesium, mmol/L 0.85 (0.750.94) 0.004 (0.0020.008) 0.004 (0.0020.008) 0.011 (0.0100.012) 0.0056 0.0123

0.5 0.4 1.3

Potassium, mmol/L 4.38 (3.754.93) 0.075 (0.0480.12) 0.092 (0.0800.11) 0.031 (0.0280.035) 0.119 0.122

1.7 2.1 0.7

Sodium, mmol/L 140.5 (135.5143.8) 0.26 (0.140.47)e 0.75 (0.690.82) 0.79

0.5

TC, mmol/L 5.23 (3.577.65) 0.071 (0.0490.104) 0.048 (0.0390.060) 0.054 (0.0490.061) 0.086 0.102

1.4 0.9 1.0

TG, mmol/L 0.91 (0.492.18) 0.016 (0.0110.024) 0.011 (0.0100.014) 0.013 (0.0120.015) 0.020 0.024

1.8 1.2 1.5

a The combined SDs in column 6 include the between-venipuncture and the preanalytical (excluding between-venipuncture) SDs.b The combined SDs in column 7 include the between-venipuncture, the preanalytical (excluding between-venipuncture) and the measurement repeatability SDs.

c Line 1, SD (95% CI).d Line 2, CV%.e Because of a wide 95% CI for the between-venipuncture SD, the venipuncture and preanalytical SD (excluding between-venipuncture) was combined to a common

estimate.

Minimal Preanalytical Uncertainty in Clinical Chemistry

Clinical Chemistry56:8 (2010) 1333


6/7

However, the preanalytical SDs (excluding venipunc-ture) estimated for each tube type were not found to besignificantly different (Table 2). Therefore the datafrom both tubes were used in the main analyses (Tables1 and 3, and Fig. 1).

Because samples were collected by phlebotomyperformed in both arms of the sample donors, we in-cluded a fixed effect representing the arm used (left orright), and a random effect between the venipunctures.We found no significant bias between arms (Table 1),

but a considerable between-venipuncture variation(Table 3). A reported study on the effect of specimencollection on coagulation assays (7 )also revealed nosignificant difference between the results from phle-botomy in the left vs right arm.

The between-venipuncture SD may be a result ofvariation in blood flow, muscle strength, blood pres-sure, and other biochemical or physiological effects.Small differences in the position of the arm, thestrength of the tourniquet, and the cut, depth, and du-ration of the venipunctures may all contribute to theuncertainty of the venipuncture. In sum, the between-venipuncture variation turned out to be somewhat

higher than the preanalytical variation (excluding ve-nipuncture) for several analytes,and for glucose in par-ticular it was the dominant source of variation. ForALP and potassium the between-venipuncture SDwas significantly higher than the measurementrepeatability.

Results of a study on the influence of needle-boresize showed that using small needles could lead tohigher variability of potassium results (8 ). A study ofthe influence of tourniquet application on venousblood sampling for serum chemistry showed that se-rum electrolytes are not affected by stasis up to 180 s

(5 ). However, Lippi et al. concluded that albumin, cal-cium, and potassium showed clinically significant dif-ferences after as little as 1 min of stasis, and severalother analytes after 3 min of stasis(16 ).

The preanalytical SD (excluding venipuncture)is a result of variations in variables that include fill-ing volume, sample mixing, clotting time, extent ofhemolysis,centrifugal force, room temperature andhumidity, and amount of time before analysis. Evenwith optimal routine, some variation is inevitable.

For LDH and potassium the preanalytical SD(excluding venipuncture) was significantly higherthan the measurement-repeatability SD. It is wellknown that LDH and potassium are particularly sen-sitive to hemolysis(1 ), and there may be some leak-age out of the cell even without visible hemolysis.These analytes are therefore particularly importantto standardize. Hemolysis has been suggested to be asuitable indicator for preanalytical quality(17 )andis a leading cause of unsuitable specimens in clinicallaboratories(18 ). In the study by Fuentes-Arderiu etal.(9 )the preanalytical uncertainty (including veni-puncture) for glucose was estimated to be 3.2%, and

for potassium to be 3.1%. Kouri et al. (19 )estimatedthe preanalytical uncertainty to be 6.6% for potas-sium, 0.7% for TC, and 4.7% for albumin, by addinguncertainties from, e.g., specimen collection and de-lay in pretreatment and transportation.

The estimated measurement repeatability SDs(Table 3) were in agreement with the measurement re-peatability SDs calculated from control material usedin our clinical routine. Although the measurement re-peatability SD is the smallest achievable measurementvariation, it was the dominant source of variation forALT, GGT, and magnesium.

LDH CK

0

5

10

15

20

ALT Creatinine ALP GGT Albumin

0.0

0.5

1.0

1.5

Glucose Potassium TC0.00

0.02

0.04

HDL-C TG Calcium Magnesum0

0.0005

0.0010

Variance

Variance

Variance

Variance

Fig. 1. The contribution of variance from each source, for which the height of the combined bar is the total variance.

The bars consist of variances calculated from the between-venipuncture (light grey), the preanalytical excluding venipuncture

(dark grey), and the measurement repeatability (black) SDs. The respective variance contributions are plotted in actual variance

units (squared SDs). Analytes with comparable scales are grouped together.



7/7

Linear mixed-effects models are powerful tools forquantifying variance from a variety of sources. Withappropriate experimental designs, this approach can beused for the evaluation of interindividual, preanalyti-cal, and analytical sources. Both random and fixed pre-analytical effects can be determined and combinedwith measurement variation to give the total uncer-tainty of laboratory results. Empirical Bayesian modelsalso could be used as an alternative statistical methodto linear mixed-effects models(14 ).

By use of samples collected under optimal condi-tions, we have estimated the minimal preanalytical un-certainty. This minimal uncertainty will influence allpatient results, and may be compared with preanalyti-cal uncertainty in current practice. For many of theanalytes tested, the contribution of venipuncture tothe total uncertainty was considerable. The method we

used is practical and can easily be adopted by routinelaboratories.

Author Contributions:All authors confirmed they have contributed tothe intellectual content of this paper and have met the following 3 re-

quirements: (a) significant contributions to the conception and design,acquisition of data, or analysis and interpretation of data; (b) draftingor revising the article for intellectual content; and (c) final approval ofthe published article.

Authors Disclosures of Potential Conflicts of Interest:No authorsdeclared any potential conflicts of interest.

Role of Sponsor:The funding organizations played no role in the

designof study, choiceof enrolledpatients,review and interpretationof data, or preparation or approval of manuscript.

Acknowledgments: We acknowledge the biomedical laboratory sci-

entists at thesection forautomaticanalysisat theHaukelandUniver-sity Hospital, Laboratory of Clinical Biochemistry.

References

1. Guder WG, Narayanan S, Wisser H, Zawta B.

Diagnostic samples: from the patient to the

laboratory: the impact of preanalytical vari-

ables on the quality of laboratory results. 4th,

updated ed. Weinheim: Wiley-VCH; 2003.

114 p.

2. Alsina MJ, Alvarez V, Barba N, Bullich S, Cortes

M, Escoda I, Martinez-Bru C. Preanalytical quality

control program: an overview of results (2001

2005 summary). Clin Chem Lab Med 2008;46:

84954.

3. Rynning M, Wentzel-Larsen T, Bolann BJ. A model

for an uncertainty budget for preanalytical vari-

ables in clinical chemistry analyses. Clin Chem2007;53:13438.

4. Joint Committee for Guides in Metrology. Evalu-

ation of measurement data guide to the ex-

pression of uncertainty in measurement. 1st ed.,

2008; corrected version, 2010. http://www.bipm.

org/utils/common/documents/jcgm/JCGM_100_

2008_E.pdf (Accessed June 2010). JCGM 100:

2008 (GUM 1995 with minor corrections).

5. Cengiz M, Ulker P, Meiselman HJ, Baskurt OK.

Influence of tourniquet application on venous

blood sampling for serum chemistry, hematolog-

ical parameters, leukocyte activation and eryth-

rocyte mechanical properties. Clin Chem Lab Med

2009;47:76976.

6. Haverstick DM, Brill LB, Scott MG, Bruns DE.

Preanalytical variables in measurement of free

(ionized) calcium in lithium heparin-containing

blood collection tubes. Clin Chim Acta 2009;403:

1024.

7. Lippi G, Guidi GC. Effect of specimen collection

on routine coagulation assays and D-dimer mea-

surement. Clin Chem 2004;50:21502.

8. Lippi G, Salvagno GL, Montagnana M, Brocco G,

Guidi GC. Influence of the needle bore size used

for collecting venous blood samples on routine

clinical chemistry testing. Clin Chem Lab Med

2006;44:100914.

9. Fuentes-Arderiu X, Acebes-Frieyro G, Gavaso-Navarro L, Castineiras-Lacambra MJ. Pre-

metrological (pre-analytical) variation of some

biochemical quantities. Clin Chem Lab Med 1999;

37:9879.

10. Garcia-Panyella M, Padro-Miquel A, Dot-Bach D,

Fuentes-Arderiu X. Pre-analytical variation of

some haematological quantities. Clin Chem Lab

Med 2008;46:116870.

11. Padro-Miquel A, Garcia-Panyella M, Garcia-

Gimenez MJ, Fuentes-Arderiu X. Pre-metrological

variation of immunological quantities. Scand

J Clin Lab Invest 2008;68:4312.

12. CLSI. Procedures for the handling and processing

of blood specimens for common laboratory tests;

approved guidelinefourth edition. Document nr

H18-A4. Wayne (PA): CLSI; 2010. 57 p.

13. CLSI. Procedures for the collection of diagnostic

blood specimens by venipuncture; approved stan-

dardsixth edition. Document nr H3-A6. Wayne

(PA): CLSI; 2007. 56 p.

14. Verbeke G, Molenberghs G. Linear mixed models

for longitudinal data. New York: Springer-Verlag;

2000. 608 p.

15. Pinheiro JC, Bates DM. Mixed-effects models in S

and S-PLUS. New York: Springer-Verlag; 2001.

528 p.

16. Lippi G, Salvagno GL, Montagnana M, Brocco G,

Guidi GC. Influence of short-term venous stasison clinical chemistry testing. Clin Chem Lab Med

2005;43:86975.

17. Soderberg J, Jonsson PA, Wallin O, Grankvist K,

Hultdin J. Haemolysis index: an estimate of pre-

analytical quality in primary health care. Clin

Chem Lab Med 2009;47:9404.

18. Lippi G, Blanckaert N, Bonini P, Green S, Kitchen S,

Palicka V, et al. Haemolysis: an overview of the

leading cause of unsuitable specimens in clinical

laboratories. Clin Chem Lab Med 2008;46:76472.

19. Kouri T, Siloaho M, Pohjavaara S, Koskinen P,

Malminiemi O, Pohja-Nylander P, Puukka R. Pre-

analytical factors and measurement uncertainty.

Scand J Clin Lab Invest 2005;65:46375.

Minimal Preanalytical Uncertainty in Clinical Chemistry

Clinical Chemistry56:8 (2010) 1335

Estimation of the Minimal Preanalytical Uncertainty for 15 Clinical Chemistry Serum Analytes

Documents

Transcript of Estimation of the Minimal Preanalytical Uncertainty for 15 Clinical Chemistry Serum Analytes