Preanalytical Variables and Off-Site BloodCollection: Influences on the Results of the

Prothrombin Time/International Normalized RatioTest and Implications for Monitoring of Oral

Anticoagulant TherapyJohanna H.H. van Geest-Daalderop,1* Andre B. Mulder,1

Leandra J.M. Boonman-de Winter,2 Martha M.C.L. Hoekstra,3 andAnton M.H.P. van den Besselaar3

Background: The quality of oral anticoagulant therapymanagement with coumarin derivatives requires reli-able results for the prothrombin time/International Nor-malized Ratio (PT/INR). We assessed the effect onPT/INR of preanalytical variables, including ones re-lated to off-site blood collection and transportation to alaboratory.Methods: Four laboratories with different combinationsof blood collection systems, thromboplastin reagents,and coagulation meters participated. The simulated pre-analytical variables included time between blood col-lection and PT/INR determinations on samples stored atroom temperature, at 46 C, and at 37 C; mechanicalagitation at room temperature, at 46 C, and at 37 C;time between centrifugation and PT/INR determination;and times and temperatures of centrifugation. For vari-ables that affected results, the effect of the variable wasclassified as moderate when 10% or as large if >25% of samples showedsuch a change.Results: During the first 6 h after blood collection, INRchanged by >10% in 10% in>25% of samples. In all laboratories, a 24-h delay at37 C or with mechanical agitation had a large effect. Weobserved no clinically or statistically relevant INR dif-ferences among studied centrifugation conditions (cen-trifugation temperature, 20 C or no temperature control;centrifugation time, 5 or 10 min).Conclusions: We recommend a maximum of 6 h be-tween blood collection and PT/INR determination. Theimpact of a 24-h delay should be investigated for eachcombination of materials and conditions. 2005 American Association for Clinical Chemistry

The purposes of effective and safe oral anticoagulationtherapy (OAT)4 with coumarin derivatives are to reducethe risk of thromboembolic events and to minimize theincidence of bleeding complications. This can be achievedby maintaining the prothrombin time (PT), transformed tothe International Normalized Ratio (INR), within thetherapeutic ranges (1 ). In The Netherlands, the manage-ment of OAT for outpatients is performed by specializedanticoagulation clinics, called Thrombosis Services. AllThrombosis Services form part of the network of theDutch Federation of Thrombosis Services. In 2003, the 62

1 Thrombosis Service, Department of Laboratory of Clinical Chemistry andHaematology, Jeroen Bosch Hospital, s-Hertogenbosch, The Netherlands.

2 Thrombosis Service and Contract Research, Departments of StichtingHuisartsen Laboratorium, Etten-Leur, The Netherlands.

3 Haemostasis and Thrombosis Research Center, Department of Haema-tology, Leiden University Medical Center, Leiden, The Netherlands.

*Address correspondence to this author at: Trombosedienst s-Hertogen-bosch, Jeroen Bosch Ziekenhuis, Deutersestraat 2, 5223 GV s-Hertogenbosch,The Netherlands. Fax 31-73-6998257; e-mail ankees.vangeest@planet.nl.

Received September 22, 2004; accepted December 28, 2004.Previously published online at DOI: 10.1373/clinchem.2004.043174

4 Nonstandard abbreviations: OAT, oral anticoagulation therapy; PT, pro-thrombin time; INR, International Normalized Ratio; and ISI, InternationalSensitivity Index.

Clinical Chemistry 51:3561568 (2005) Hemostasis and

Thrombosis

561

Thrombosis Services took care of 327 067 patients withOAT and carried out 4 629 097 PT/INR determinations(The Netherlands has a population of 16.4 million). Toreach the appropriate intensity of therapy, the INR mustbe within narrow therapeutic intervals for each indication(2.03.5 or 2.54.0, according to the recommendations ofthe Dutch Federation of Thrombosis Services) (2 ). In 2003,a mean of 74% of the INRs were within the therapeuticranges. The quality of OAT management depends on thepatient education, knowledge of the variable patient char-acteristics, prescription of the right dose of the coumarinderivatives, and the reliability of the result of the PT/INRdetermination. The latter variable depends on preanalyti-cal and analytical influences.

The INR was introduced in 1983 with the intention ofharmonizing the results of the PT (3, 4). Since that time,many studies concerning the analytical and preanalyticalvariables have been carried out to improve the harmoni-zation of the INR, including (local) International Sensitiv-ity Index (ISI) calibration and the study of various throm-boplastin reagents, plasma calibrants, blood collectiontechniques, and citrate concentrations (5, 6). In addition,external quality assessment has been developed and hasmade a contribution to the harmonization of the INR (7 ).Although use of the INR substantially reduces the differ-ences in PT results, interlaboratory variation still persists.Possibly some preanalytical variables that have beenreceived less attention may contribute to the variation inPT/INR results.

In The Netherlands, blood collection for determinationof the PT/INR in outpatients on OAT takes place primar-ily at locations remote from the laboratory, including thepatients homes. Transport of blood samples to the labo-ratory can take 16 h. Furthermore, the samples aretransported by car, in which case they experience mechan-ical agitation and widely varying temperatures. In addi-tion, samples are sometimes centrifuged on arrival at thelaboratory and then sit until analyses can be carried out.Whether all of these variables influence the results of theINR is not known.

The aim of our study was to investigate whether thefollowing preanalytical variables influence the PT/INRresults and whether any changes seen are clinically rele-vant: time between blood collection and PT/INR deter-mination on samples stored at room temperature, 46 C,and 37 C; mechanical agitation at room temperature,46 C, and 37 C; time between centrifugation and PT/INR determination; and times and temperatures of cen-trifugation.

Four laboratories with different combinations of bloodcollection systems, thromboplastin reagents, and coagu-lation meters participated.

Materials and MethodsquestionnaireTo collect data about the different procedures and circum-stances of off-site blood collection and PT/INR determi-

nation, we developed a questionnaire that was submittedto each Thrombosis Service in The Netherlands at the endof 2003. The questionnaire asked for the minimum andmaximum time between blood collection and determina-tion of the PT/INR, the positions of samples and methodof temperature regulation during transport, the time andmethods of centrifugation, and the methods used todetermine the PT/INR in plasma or whole blood.

participating centersThis comparative study was carried out in the laboratoriesof three Thrombosis Services respondents, s-Hertogen-bosch, Etten-Leur, and Veldhoven (Centers 1, 2, and 4),and in the laboratory of the Haemostasis and ThrombosisResearch Center in Leiden (Center 3). The centers usedifferent combinations of blood collection systems,thromboplastin reagents, and coagulation meters. Table 1shows the blood collection systems, sodium citrate con-centrations, thromboplastin reagents, automated coagula-tion meters, and ISI used by the four centers. For oneexperiment (different times of centrifugation), Center 1used another blood collection system, Vacutainer insteadof Monovette. Blood collection took place by venipunc-ture according to standard procedures. The PT/INRswere determined according to routine practices in thecenters.

design of the experimentsWe simulated the preanalytical variables in the fourlaboratories (Table 2). In general, 20 patients participatedin each experiment (the exact numbers of patients arenoted in the tables). From each patient we collected fouror five (experiments at 46 and 37 C) tubes instead of theusual one. After blood collection, we subjected the tubesto the several conditions.

The result of the PT/INR determination in tube 1(stored at room temperature for 0.51 h after bloodcollection) was the reference value. For the experiments at46 C and 37 C, we used one extra tube (1a). Tube 1awas incubated at the tested temperature for 24 h, and thePT/INR was determined at the same times as for tube 1.PT/INR determinations were carried out at fixed times. Inconjunction with the PT/INR determinations for tube 2(incubated for 3 h), 3 (incubated for 6 h), and 4 (incubatedfor 24 h), we also assayed tubes 1 and 1a. Because of thedifferent centrifugation procedures, as noted in the ques-tionnaire, we carried out two more experiments: one withthe centrifuge cooled to 20 C vs no temperature controland the other at different centrifugation times (5 and 10min). In the last experiment we also counted the platelets.

statistical analysisContinuous variables were compared by the t-test forpaired PT/INR determinations. A 5% significance levelwas used. All statistical analyses were performed withSPSS 10 for Windows.

562 van Geest-Daalderop et al.: Preanalytical Variables and PT/INR Determination

ResultsquestionnaireFifty-eight of the 62 Thrombosis Services responded to thequestionnaire. The questions and answers are shown inTable 3. In most cases, blood samples were reported toarrive at the central laboratory within 6 h after bloodcollection. When blood collection takes place in the labo-ratory and not at an off-site location, the PT/INR test istypically performed after 0.5 h, but only eight ThrombosisServices perform the test within 0.5 h. Most ThrombosisServices transport the tubes in a vertical position and havesome kind of temperature control in the car. The methodsof centrifugation (rpm, g force, cooled or not, and dura-tion) differ widely. There is no correlation between theused rpm, g force, and duration of centrifugation. MostThrombosis Services determine the PT/INR in plasma.

experimentsShown in Table 4 and Fig. 1 are summaries of the resultsof experiments 1a, 1b, 1c, 2a, 2b, 2c, and 3a (see Table 2).The mean (SD) INRs, the mean percentage changes {ob-tained with the formula: % change [(INRt INRt1)/INRt1] 100, where t1 is the reference value}, the rangesof the maximum positive and negative percentagechanges, the number of individual INR values that ex-ceeded the limits of 10% change, and the P values aregiven in Tables 17 of the Data Supplement that accom-panies the online version of this article at http://www.clinchem.org/content/vol51/issue3/. We considered anINR difference 10% as clinically relevant.

For daily practice, the clinically relevant differences areimportant because they can cause dosage adjustment ofthe coumarin derivatives. We defined a change in INR asmoderate when 25% of individual INRs exceeded the10% limit and as large when 25% of individual INRsexceeded the 10% limit. Shown in Fig. 1 are the percent-ages of individual INRs that exceeded the 10% limit andthe moderate and large changes. The data for experiments3b and 3c are shown in Tables 8 and 9 of the online DataSupplement.

Experiment 1a: Time between blood collection and PT/INRdeterminations in samples stored at room temperature (Table 1in the online Data Supplement). For Center 1 (n 20patients), the mean percentage changes in INR werenegative or zero; all mean changes were 10%. Thepercentage of samples with negative changes that ex-ceeded the 10% limit was moderate (25% of samples)after 24 h (tubes 1 and 4); most of these were from onepatient. The differences in INR values for tube 1 werestatistically significant after 6 and 24 h. For Center 2 (n 20 patients), the mean percentage changes were all nega-tive and exceeded 10% after 24 h. The percentages ofsamples with negative changes that exceeded the 10%limit were large (25% of samples) after 6 h (tubes 1 and3) and 24 h (tubes 1 and 4). All changes were statisticallysignificant.

Table1.Blood

collectionsystem

s,sodium

citrateconcentrations,thromboplastinreag

ents

andmanufacturers,typesof

thromboplastinreag

ents,automated

coag

ulationmetersandmanufacturers,andISIused

bythefour

participatingcenters.

Center

Blood

collectionsystem

Sodium

citrate,

mol/L

Thromboplastinandmanufacturer

Type

ofthromboplastin

Coagu

lation

meter

andmanufacturer

ISI

1Mon

ovette

(plastic;Sarsted

t)0.106

Hep

atoQuick

(Diagn

ostic

aStago

)Rab

bitbrainreag

entplus

adso

rbed

bovine

plas

ma

STA

Rac

k(Roc

heDiagn

ostic

sGmbH

)0.940

.93

Vacu

tainer

(plastic;Bec

tonDickins

onVa

cutainer

Systems)

(cen

trifu

gatio

nexpe

rimen

t)

0.109

2Ve

noject

II(plastic;Te

rumo

Europe

N.V.)

0.109

Rec

ombiplas

tin(In

strumen

tatio

nLa

boratory

B.V.)

Rec

ombina

nthu

man

reag

ent

Elec

tra1600C(M

edical

Labo

ratory

Automation,

Inc.)

0.991

.06

3Va

cutainer

(glass

;Bec

tonDickins

onVa

cutainer

Systems)

0.109

Sym

plas

tinHTF

(BioMrieux,Inc.)

Hum

antis

suefactor

from

cultu

redhu

man

cells

Thrombo

lyzer/MAX

(Beh

nkElek

tron

ik)

1.24

4Va

cutainer

(glass

;Bec

tonDickins

onVa

cutainer

Systems)

0.105

Rec

ombiplas

tin(In

strumen

tatio

nLa

boratory

B.V.)

Rec

ombina

nthu

man

reag

ent

ACL-Fu

tura

(Instrumen

tatio

nLa

boratory)

0.79

Clinical Chemistry 51, No. 3, 2005 563

Experiment 1b: Time between blood collection and PT/INRdeterminations in samples stored at 46 C (Table 2 of theonline Data Supplement). For Center 1 (n 20 patients), themean percentage changes in INR were negative or nearzero and became positive after 24 h (tubes 1, 1a, and 4); allmean changes were 10%. All maximum changes werewithin the 10% limit, except for tube 4 (incubation for24 h), for which the percentage of samples with positivechanges that exceeded the 10% limit was moderate. The

differences were statistically significant in tubes 1 and 1aafter 6 h and in tube 4 after 24 h. For Center 2 (n 20patients), the mean percentage changes in INR werenegative for all tubes; all mean changes were 10%. Thepercentage of samples with negative changes in INR thatexceeded the 10% limit was moderate (25% of samples)after 6 h and was large (25% of samples) for tube 4 after24 h. All changes were statistically significant except intube 1a after 3 h and 24 h.

Table 2. Experiments, conditions, and participating centers.Experiments Conditions Center(s)

1a. Time between blood collection and PT/INR test onspecimens stored at room temperature

Tests after 0.51, 3, 6, and 24 h; centrifugation justbefore the tests

1 and 2

1b. Time between blood collection and PT/INR test onspecimens stored at 46 C

Tests after 0.51, 3, 6, and 24 h; centrifugation justbefore the tests; storage, in refrigerator

1 and 2

1c. Time between blood collection and PT/INR test onspecimens stored at 37 C

Tests after 0.51, 3, 6, and 24 h; centrifugation justbefore the tests; storage, in waterbath (Center 1)or incubator (Center 2)

1 and 2

2a. Time between blood collection and PT/INR test withmechanical agitation and storage at room temperature

Tests after 0.51, 3, 6, and 24 h; centrifugation justbefore the tests; agitation, tubes lying horizontalon roller mixer

1 and 2

2b. Time between blood collection and PT/INR test withmechanical agitation and storage at 46 C

Tests after 0.51, 3, 6, and 25 h; centrifugation justbefore the tests; storage, in refrigerator; agitation,tubes vertical in orbital shaker

3

2c. Time between blood collection and PT/INR test withmechanical agitation and storage at 37 C

Tests after 0.51, 3, 6, and 25 h; centrifugation justbefore the tests; storage, in incubator; agitation,tubes vertical in orbital shaker

3

3a. Time between centrifugation and PT/INR test Tests after 0.51, 3, 6, and 24 h; centrifugation justafter blood collection; plasma left on the spun-down blood cells

1 and 2

3b. Centrifugation with and without control of temperature Control at 20 C, 1892g, 3000 rpm, and 10 min 43c. Centrifugation at different times and platelet counts Times, 5 and 10 min; 2500g, 3822 rpm;

temperature, 4 C1

Table 3. Questionnaire: Questions and answers from 58 Thrombosis Services.Preanalytical variables n

Maximum time between blood collection and PT test 46 h 56Other 2

Minimum time between blood collection and PT testin the central laboratory

0.5 h 500.5 h 8

Position of the tubes during transport Vertical 49Horizontal 5Both 4

Temperature: special measures during transport Yes 46No 12

CentrifugationTime Just before determination 52

Not applicable (whole blood) 6rpm and g 24204700 rpm; 10006240g Differ widelyCooled Yes 21

No 31Duration 10 min 35

10 min 810 min 6Other 3

PT measurements In plasma 52In whole blood 6

564 van Geest-Daalderop et al.: Preanalytical Variables and PT/INR Determination

Experiment 1c: Time between blood collection and PT/INRdeterminations in samples stored at 37 C (Table 3 in the onlineData Supplement). For Center 1, the mean percentagechanges were negative for all tubes except tube 4 (24-hincubation), for which the change in INR was positive; allmean changes were 10%. The samples with negativechanges that exceeded the 10% limit were mainly fromone patient. The percentage of samples with positivechanges for tube 4 (24-h incubation) that exceeded the10% limit was large. All changes were statistically signif-icant, except in tube 3 (incubation for 6 h). For Center 2,the mean percentage changes in INR for all samples werenegative except for tubes 1a and 4 after 24 h, for which thechanges in INR were positive; all mean changes were10% except for tube 1a after 24 h. The percentages ofsamples with negative changes that exceeded the 10%limit after 3 h (tubes 1a and 2) and 6 h (tubes 1a and 3)were large; the percentage of samples with positivechanges that exceeded the 10% limit was also large after24 h (tubes 1a and 4). All changes were statistically

significant, except in tube 1 after 24 h and in tube 1a after0.51 h.

Experiment 2a: Time between blood collection and PT/INRdeterminations with mechanical agitation and storage at roomtemperature (Table 4 of the online Data Supplement). ForCenter 1 (n 20 patients), the mean percentage changesin tube 1 were negative or zero and were 10%; in theother tubes they were positive after 3 h (tube 2), 6 h (tube3), and 24 h (tube 4) of mechanical agitation and were10% after 24 h. The percentage of samples with positivechanges that exceeded the 10% limit was moderate (25%of samples) after 6 h and large (25% of samples) after24 h. All changes were statistically significant, except fortube 1 after 24 h. For Center 2 [n 21 patients except fortube 1 after 3, 6, and 24 h (n 20 patients)], the meanpercentage changes in tube 1 were negative and were10%; in the other tubes, they were positive after 3 h(tube 2), 6 h (tube 3), and 24 h (tube 4) of mechanicalagitation and were 10% after 24 h. The percentage of

Table 4. Summary of the results of experiments 1a, 1b, 1c, 2a, 2b, 2c, and 3a.a

Experiment Tube Time,b h

Mean % change (P) Range of maximum negative and positive % changes

Center 1 Center 2 Center 3

Center 1 Center 2 Center 3

1a 1 0.51c

2 3 0.0 (0.832) 4.6 (0.013)d 12.5 12.3 18.6 16.63 6 2.2 (0.101) 8.2 (0.001)d 11.8 15.9 22.1 4.24 24 1.4 (0.182) 10.2 (0.001)d 12.1 5.5 22.1 0.7

1b 1 0.51c

2 3 0.2 (0.740) 3.8 (0.004)d 2.5 4.6 16.8 5.33 6 1.1 (0.180) 5.0 (0.001)d 6.5 2.5 13.9 2.84 24 3.4 (0.037)d 8.6 (0.001)d 5.2 12.7 21.3 4.2

1c 1 0.51c

2 3 3.1 (0.003)d 7.7 (0.001)d 12.9 2.9 18.0 4.93 6 1.5 (0.154) 5.2 (0.012)d 13.4 7.9 16.4 9.04 24 7.5 (0.002)d 8.1 (0.004)d 6.3 24.1 13.3 18.8

2a 1 0.51c

2 3 1.5 (0.048)d 0.6 (0.816) 3.6 5.2 12.2 10.43 6 6.0 (0.001)d 5.8 (0.032)d 1.8 13.0 14.5 21.64 24 39.0 (0.001)d 22.9 (0.001)d 52.7 59.0

2b 1 0.51c

2 3 1.5 (0.003)d 6.0 1.43 6 2.6 (0.001)d 7.9 1.04 24 6.8 (0.001)d 14.3 0.0

2c 1 0.51c

2 3 2.5 (0.001)d 4.9 0.03 6 0.1 (0.858) 4.9 3.74 24 31.5 (0.001)d 46.5

3a 1 0.51c

2 3 2.7 (0.001)d 1.0 (0.575) 9.8 2.2 14.4 17.63 6 1.8 (0.003)d 0.4 (0.750) 6.7 5.6 10.7 8.04 24 0.7 (0.281) 1.8 (0.211) 6.7 4.7 9.6 12.4

a See Table 2 for a description of the experiments.b Time between blood collection and PT/INR determination.c Reference value.d Statistically significant difference (P 0.05).

Clinical Chemistry 51, No. 3, 2005 565

samples with positive changes that exceeded the 10%limit was moderate after 6 h and large after 24 h. Allchanges were statistically significant except in tube 2 (3-hincubation).

Experiment 2b: Time between blood collection and PT/INRdeterminations with mechanical agitation and storage at46 C (Table 5 of the online Data Supplement). This exper-iment was performed only at Center 3 (n 20 patients).The mean percentage changes were negative except fortube 1a after 0.51 h and 25 h, when the changes were nearzero; all mean changes were 10%. The percentage ofpatients with negative changes that exceeded the 10%limit was moderate at 25 h of incubation (tube 4). Allchanges were statistically significant, except in tube 1 after25 h and in tube 1a after 0.51, 6, and 25 h.

Experiment 2c: Time between blood collection and PT/INRdeterminations and mechanical agitation and storage at 37 C(Table 6 of the online Data Supplement). This experiment wasperformed only at Center 3 (n 20 patients). The meanpercentage changes in INR were negative or near zero for

all tubes until 6 h and were 10%; for tubes withmechanical agitation and storage at 37 C, they becamepositive after 6 h and were 10% after 25 h. The percent-age of samples with positive changes in INR that ex-ceeded the 10% limit was large after 25 h. All changeswere statistically significant except in tube 1 after 25 h andin tubes 1a and 3 after 6 h.

Experiment 3a: Time between centrifugation and PT/INRdeterminations (Table 7 of the online Data Supplement). ForCenter 1, the mean percentage changes were negative; allmean changes were 10%. All maximum changes werealso within the 10% limit. All changes were statisticallysignificant, except in tube 4 after 24 h. For Center 2, themean percentage changes were positive or negative, andall were 10%. The percentage of negative or positivechanges that exceeded the 10% limit was moderate. Therewere no statistically significant differences.

Experiment 3b: Centrifugation with and without control oftemperature (Table 8 of the online Data Supplement). Thisexperiment was performed only at Center 4. The mean

100%

75%

50%

25%

0%

-25%

-50%

Center 1 Center 2

Experiment 1a

Center 1 Center 2

Experiment 1b

Center 1 Center 2

Experiment 1c

Center 1 Center 2

Experiment 2a

Center 3

Exp. 2b

Center 3

Exp. 2c

Center 1 Center 2

Experiment 3a

10%

change

Perc

enta

ge o

f IN

R v

alu

es w

ith>

Fig. 1. Clinically relevant changes for experiments 1a, 1b, 1c, 2a, 2b, 2c, and 3a in Centers 1, 2, and 3 (see Table 2).Shown are the percentages of individual INR values that exceed the 10% limit. Values between the dotted lines corresponding to 25% and 25% represent moderatechanges, and changes outside these lines are considered large changes. Time between blood collection and PT/INR determination: , 3 h; , 6 h; f, 24 h.

566 van Geest-Daalderop et al.: Preanalytical Variables and PT/INR Determination

percentage difference for tube 2 compared with tube 1was negative. The difference was within the 10% limit andwas not statistically significant.

Experiment 3c: Centrifugation at 10 and 5 min, and number ofplatelets (Table 9 of the online Data Supplement). This exper-iment was performed only at Center 1. The mean percent-age difference for tube 2 compared with tube 1 wasnegative. The difference was within the 10% limit and wasnot statistically significant. The mean numbers of plateletswere 68 109/L after 5 min of centrifugation and 18 109/L after 10 min.

DiscussionOur survey showed that off-site blood collection for thecontrol of OAT produces various preanalytical conditions.

A NCCLS guideline (8 ) recommends determination ofthe PT/INR within 24 h of blood collection and mainte-nance of sample tubes at 24 C or at room temperature(1824 C). Several studies (917) have generally con-firmed that such storage for 24 h does not lead to clinicallyimportant changes in the mean PT/INR results. Thesestudies, however, considered only the mean values andpossibly overlooked changes in individual samples thatcould lead to a change in the dose of coumarin deriva-tives. Other studies have taken into consideration thatINR values of individual patients can fall outside thechosen limits (9, 12, 14). Leeming et al. (12 ) concludedthat for some individual patients, storage of samples for24 h changes the PT/INR values, leading to a dosageadjustment; these authors do not support that practice.Studies of mechanical agitation that simulate the condi-tions in a car during transport are few. In one study, thetubes were subjected twice to a gentle agitation of only 30min (14 ). We found no studies on storage at 37 C in theliterature. Centrifugation at high speed compared withroutine centrifugation showed comparable results (18 ).

The statistical significance of the differences in INRcannot be used as a criterion to draw conclusions aboutthe daily practice of calculating the dose of coumarinderivatives. The mean PT/INRs and mean percentagechanges are important to give an overall picture of thecourse of the results during the experiments. We wereparticularly interested in clinically relevant changes. Forthe preanalytical variables, there are no guidelines forpercentage changes compared with reference value thatare acceptable. For calibration procedures and for testingof the local PT systems, a deviation of the INR of 10% ofthe reference value is defined as clinically important(19, 20). We decided to consider an INR change outsidethe positive or negative limits of 10% of the referencevalue to be clinically relevant.

Centers 1 and 2 did the same experiments. In Center 1(Monovette, Hepato Quick, and STA Rack), for most ofthe experiments the results of the PT/INR test changedless than results in Center 2 (Venoject, Recombiplastin,

and Electra). We have no explanation for this differencebetween centers.

During storage time at room temperature, 46 C, and37 C (experiments 1a, 1b, and 1c), changes in PT/INRwere mostly negative until 6 h, but were positive after24 h at 37 C. Center 1 considered it acceptable to store atroom temperature and in the refrigerator for 24 h, but atCenter 2 only storage for up to 6 h is acceptable. In bothcenters, storage at 37 C longer than 6 h is not acceptable.

Mechanical agitation with the roller mixer (horizontalposition) at room temperature (experiment 2a) showed inCenters 1 and 2 an increase in changes after 6 h; after 24 h,all changes were positive and exceeded the 10% limit. Themechanism of this phenomenon is unknown. In Center 3(Vacutainer, Simplastin HTF, and Thrombolyzer), me-chanical agitation with the orbital shaker (vertical posi-tion) at 46 C and 37 C (experiments 2b and 2c) showedno relevant changes until 6 h. At 46 C, the changesremained negative after 24 h, and only a small numberexceeded the limit of 10%. However, with mechanicalagitation at 37 C, all changes were positive and exceededthe limit of 10%. Comparing the two experiments at 37 C(experiments 1c and 2c), we see that this temperaturealways changes the results from negative to positive andthat mechanical agitation will enhance this phenomenon.We assume that mechanical agitation in a horizontalposition stimulates coagulation more than when the tubeis in a vertical position. We cannot explain this phenom-enon. We recommend transporting the tubes vertically,and after transporting the tubes in a car to determine thePT/INR within 6 h after blood collection. The tempera-tures inside cars vary wide, but a high temperature mustbe avoided, not exceeding 37 C for 6 h.

When the samples were centrifuged immediately afterblood collection (experiment 3a), the results were stable atroom temperature at both Centers 1 and 2. Thus, it ispossible to carry out centrifugation immediately afterblood collection and store the samples at room tempera-ture. However, off-site blood collection does not alwaysoffer an opportunity to centrifuge immediately before thetubes are transported.

When we looked at the results obtained for tubes 1 and1a in all experiments in Centers 1, 2, and 3 (experiments 1athrough 3a), we saw that a 24-h delay at room tempera-ture or at 46 C caused a negative change in INR in allcenters, with larger changes in center 2 than in Centers 1and 3. These tubes were also centrifuged immediatelyafter blood collection, between 0.51 h, but the PT/INRwas determined several times in the same tube.

For determination of PT/INR, a temperature-con-trolled centrifuge is not required. The duration of centrif-ugation (5 or 10 min) and the platelet count did notinfluence the results.

Many different combinations of blood collection sys-tems, thromboplastin reagents, and coagulation metersare used. The influence of preanalytical variables thatchange the INR values by 10% must be minimized. In

Clinical Chemistry 51, No. 3, 2005 567

laboratories that have not investigated the influences ofthe preanalytical variables mentioned above on PT/INRresults, we recommend a maximum of 6 h of storage atroom temperature, 46 C, or 37 C or after mechanicalagitation. This recommendation does not agree with theNCCLS guideline (8 ). Storage at a temperature of 37 C ormechanical agitation for 24 h appears to be unacceptable.Storage at room temperature or at 46 C for 24 h beforedetermination of PT/INR may be acceptable, but thismust be determined by each laboratory for its combina-tions of assays and blood collection methods.

We thank the personnel at the laboratories who accuratelyperformed the experiments. We would particularly like tothank Masja de Punder and Afzal Kariman (s-Hertogen-bosch), Annelies van der Smissen-van Meel (Etten-Leur),and Ellen van Eekelen (Veldhoven). We also thank RocheDiagnostics for providing the thromboplastin reagentHepato Quick.

References1. Rosendaal FR, Cannegieter SC, van der Meer FJM, Briet E. A

method to determine the optimal intensity of oral anticoagulationtherapy. Thromb Haemost 1993;39:2369.

2. Van Geest-Daalderop JHH, Sturk A, Levi M, Adriaansen HJ. Om-vang en kwaliteit van de antistollingsbehandeling met cumarined-erivaten door de Nederlandse trombosediensten. Ned TijdschrGeneeskd 2004;148:7305.

3. WHO Expert Committee on Biological Standardization. Require-ments for thromboplastins and plasmas used to control oralanticoagulant therapy. 33rd report. World Health Organ Tech RepSer 1983;687:81105.

4. Loeliger EA, van den Besselaar AMHP, Lewis SM. Reliability andclinical impact of the prothrombin times in oral anticoagulantcontrol. Thromb Haemost 1985;53:14854.

5. Poller L. International Normalized Ratios (INR): 20 years. J ThrombHaemost 2004;2:84960.

6. Adcock DM, Duff S. Enhanced standardization of the InternationalNormalized Ratio through the use of plasma calibrants: a concisereview. Blood Coagul Fibrinolysis 2000;11:58390.

7. Van den Besselaar AMHP. Use of lyophilized calibration plasmas

and control blood for international normalized ratio calculation inexternal quality assessment of the prothrombin time. Am J ClinPathol 1994;102:1237.

8. National Committee for Clinical Laboratory Standards. Collection,transport and processing of blood specimens for coagulationtesting and general performance of coagulation assays. Approvedguideline H21A3, 3rd ed. Wayne, PA: NCCLS, 1998.

9. Brigden ML, Graydon C, McLeod B, Lesperance M. Prothrombintime determination. The lack of need for a discard tube and24-hour stability. Am J Clin Pathol 1997;108:4226.

10. Baglin T, Luddington R. Reliability of delayed INR determination:implications for decentralized anticoagulant care with off-sitesampling. Br J Haematol 1997;96:4314.

11. Adcock DM, Kressin DC, Marlar RA. The effect of time andtemperature variables on routine coagulation tests. Blood CoagulFibrinolysis 1998;9:46370.

12. Leeming DR, Craig S, Stevenson KJ, Taberner DA. The determina-tion of INR in stored whole blood. J Clin Pathol 1998;51:3603.

13. Davis KD, Danielson CFM, May LS, Han Z. Use of differentthromboplastin reagents causes greater variability in InternationalNormalized Ratio results than prolonged room temperature stor-age of specimens. Arch Pathol Lab Med 1998;122:9727.

14. Ridyard J, Bhavnani M, Seal LH. Laboratory control of oralanticoagulant therapy: preservation of prothrombin time speci-mens using a polypropylene collection system. Clin Lab Haematol1998;20:36972.

15. Grau E, Tenias JM, Olaso, MA, Ferrando I, Juan MT, Pastor E, et al.Monitoring oral anticoagulant treatment from plasma stored for upto 48 hours and frozen plasma. Haematologica 1999;84:6336.

16. Rao LV, Okorodudu AO, Petersen JR, Elghetany MT. Stability ofprothrombin time and activated partial thromboplastin time testsunder different storage conditions. Clin Chim Acta 2000;300:1321.

17. Froom, P, Abramova D, Bar-El, M, Barak, M. Reliability of delayedprothrombin time INR determinations in a central laboratory usingoff-site blood sampling. Clin Lab Haematol 2001;23:18992.

18. Nelson, S, Pritt A, Marlar RA. Rapid preparation of plasma forStat coagulation testing. Arch Pathol Lab Med 1994;118:1756.

19. Poller L, Barrowcliffe TW, van den Besselaar AM, Jespersen J,Tripodi A, Houghton D. Minimum lyophilized plasma requirementfor ISI calibration. European Concerted Action on Anticoagulation.Am J Clin Pathol 1998;109:196204.

20. Poller L. Screening INR deviation of local prothrombin time sys-tems. J Clin Pathol 1998;51:3569.

568 van Geest-Daalderop et al.: Preanalytical Variables and PT/INR Determination