J Clin Pthol 199;43:679-684 Automtion nd prothrombin time: United Kingdom field study of two widely used ogulometers 679 J M Thomson, D A Tberner, L Poller United Kingdom Referene Lbortory for Antiogulnt Regents nd Control, Withington Hospitl, Mnhester M2 8LR J M Thomson D A Tberner L Poller Correspondene to: L Poller Aepted for publition 8 Februry 199 Abstrt Current performne in the prothrombin time (PT) of the two min United Kingdom ogulometer/thromboplstin systems ws ssessed in field survey Twenty bnorml smples overing wide spetrum of Interntionl Normlised Rtio (INR) were distributed to users of the KC4/KC1O nd Cog--Mte instruments Cogulometer results were ompred with those of the mnul method A substntil minority with eh system showed good greement with the mnul referene vlues There ws, however, onsiderble vrition between instruments, meningful in linil terms, evidened by vrying regression slopes nd lol system Interntionl Sensitivity Indies (ISI) For intense ntiogultion (3 to 4 5 INR) lrger dose of wrfrin is needed with the Cog--Mte thn with the KC instruments With mnul INR of 4 the KC instruments tended to give longer PT (men INR + 3); the Cog--Mte PT ws generlly shorter (men INR --1) With both systems the men norml PT were shorter thn the mnul but the degree of shortening did not prllel tht of the bnorml smples This effet undermines the use of simple prothrombin rtio nd of n INR vlue derived from it, bsed on mnul ISI The use of system relted ISI nnot, however, be reommended until lol instrument vribles re ontrolled About 5% of United Kingdom hospitl lbortories use ogulometer for prothrombin time (PT) testing nd bout hlf of these use one of two types of instrument; KC4/ KC1O (Amelung) nd Cog--Mte x 2/ x C (Orgnon Teknik) We hve previously shown in Ntionl xternl Qulity Assessment Surveys (NQAS) tht ogulometers hve n effet on the Interntionl Normlised Rtio (INR) whih vries not only mong the different types of ogulometer but lso within the instruments' These dt were obtined retrospetively from NQAS omprising two or three test plsms per survey, onduted over two yers The present prospetive study ws designed to obtin ontemporry omprehensive dt on the instrument effet on PT nd INR vlues Methods All entres using either the KC4/KC1O or the Cog--Mte x 2/ x C with the Mnhester Regent (MR) were invited to tke prt The study ws onfined to one thromboplstin beuse other regents re not used in suffiient numbers in the United Kingdom for sttistil nlysis with these instruments nd beuse our previous study hd shown tht there ws thromboplstin/instrument intertion' A rnge of 2 lyophilised test plsms ws provided for ll prtiipnts together with supply of MR Smples were tken from 1 stbilised ptients, who hd been treted with oumrin, with wide stter of Interntionl Normlised Rtios (INR) vlues, nd from 1 rtifiilly depleted smples giving similr spetrum of INR Beuse of the limited volumes of some of the ptient dontions, two seprte series of 2 bnorml test plsms were provided for the two test systems, but eh gve similr spn of INR vlues nd n identil men mnul PT of 47 seonds Nine of the plsms were ommon to eh set of 2 Prtiipnts were sked to test these smples using their ustomry KC or Cog-- Mte tehnique The stted men norml prothrombin time (MNPT) vlue t eh entre ws requested The KC study ws divided into two prts onsisting of 1 dsorbed plsms on the first osion nd 1 oumrin smples one month lter The purpose ws to determine the onsisteny of performne over time nd whether the type of defiient plsm influened the result For the Cog--Mte users ll 2 plsms were issued with MR on one subsequent osion The results obtined by prtiipnts with eh test plsm were ompred with those obtined by the mnul method The ltter ws determined from suffiient replite testings by six lbortory stff t the United Kingdom Referene Lbortory to ensure men oeffiient of vrition between 125% t the lower end of the INR sle nd mximum of 3-5% t higher vlues s the oeffiient of vrition of the PT depends on its degree of prolongtion23 The results of the PT of the 2 plsms with eh instrument were plotted ginst the mnul results to provide liner nd logrithmi orthogonl regression slopes The PT were trnsformed into INR in two different
68 Thomson, Tberner, Poller wys: () bsed on the lolly derived prothrombin rtio using the stted lol MNPT nd the Interntionl Sensitivity Index (ISI) of MR; nd (b) bsed on the prothrombin rtio derived using the mnul MNPT nd the MR ISI vlue The men mnul MNPT ws determined using the geometri men of 2 norml smples nd the ISI of the two bthes of MR were heked independently in full librtion exerise ginst the seond primry WHO Interntionl Referene Preprtion, humn plin BCT/253' For eh entre the perentge devitions between instrument nd mnul INR were lulted for eh test plsm using the lol instrument MNPT or the men mnul norml vlue provided with MR An pproximte ISI for eh lol instrument/system ws determined from the logrithmi orthogonl regression slopes of the PT results of the 2 test plsms nd the ISI of MR Results The mnul MNPT with the two bthes of MR ws 15-9 seonds This ws identil with the vlue provided by the mnufturer The stted ISI vlues of these bthes were onfirmed within the reommended 3% limit of the WHO librtion proedure5 All forty four entres using the KC returned results nd 25 of 27 of the Cog--Mte system users Figures l-ld show the rithmeti orthogonl regression slopes with the two systems The men vlues expressed s PT (seonds) for the two instruments, ompred with the mnul results, re given in fig l Figures lb nd I show the slopes for KC users on dsorbed nd oumrin plsms tested in onseutive months These were onsistent between dtes nd soure of plsm in 6% of the entres Both systems showed onsiderble stter of slopes (figs lb-id) The KC gve good overll greement with the mnul method only between 25-3 seonds wheres with the Cog--Mte there ws no point of oinidene between mnul nd instrument results With the KC, the PT with the bnorml plsms tended to be up to 25 seonds shorter thn the mnul but longer with higher vlues In ontrst, the Cog--Mte PT were onsistently shorter thn the mnul method over the rnge of test plsms, lthough the Figure I (A) Men rithmeti orthogonl regression lines of the KC (-----), Cog--Mte ( ), nd the line of oinidene ( ) between mnul nd instrument results (B) Individul entre's rithmeti orthogonl regression lines for KC with dsorbed plsms (C) Individul entre's rithmeti orthogonl regression linesfor KC with oumrin plsms (D) Individul entre's rithmeti orthogonl regression lines for Cog-Mte (dsorbed nd oumrin plsms ombined) -I (A - wo, w D I- - ;l 1 2 3 1 2 - ȧ5, I- x
::::-:: Comprison of two widely used ogulometers for mesuring prothrombin time 681 Figure 2 Distribution of 6 stted ogulometer MNPT 5 4 Cog--Mte x 2 x C 3 2 ii In z 6- KC4/KCIO 5-4 3 ::::: 2 - ^ I > m I::: 9 :X::- R :' ',:,- : -: :::::I U 12 13 14 15 16 Prothrombin time (s) Figure 3 Reltion between stted MNPT nd men bnorml PT for eh Cog--Mte perentge devition diminished with inresing PT Figure 2 shows tht the stted MNPT with both systems were shorter thn tht of the mnul method but with onsiderble vrition between instruments With the KC, the shortening ws lso observed with mild defets, but there ws prolongtion with intermedite nd more severe ogultion defets-tht is, r 3 "rossover" effet This is shown in the tble whih provides n ssessment of the devition in seonds from the mnul t three representtive levels of ntiogultion, pproximting to INR of 2-1, 312, nd 45 The perentge showing shortened PT with the KC fell bruptly s the ogultion defet inresed In ontrst, with the Cog--Mte the PT ws bout two seonds shorter thn tht of the mnul ross the rnge of test plsms The MNPT reported with most Cog--Mte instruments were < 15-9 seonds, but the degree of elertion did not prllel the observed shortening of the bnorml test smples (fig 3) The instrument effet on INR vlues, lulted from the lol MNPT or the mnul men provided by the mnufturer, were determined With the KC system, the men perentge devition obtined with the lol instrument MNPT ws + 15-2 but ws redued to + 4-76 when the mnul men vlue provided by the mnufturer ws used (p < 1) With the Cog--Mte, the differene ~-1 z 3u 3 l -4 to Differenes from mnul times (seonds) obtinedfrom individul regression lines Mnul KC4/KCIO Cog--Mte PT (seonds) x 21 x C (seonds) (G%) (seonds) (%) Men 3 --2-7 -2-5 -8-3 devition from 45 +1-6 +3-6 -2-4 -5-3 mnulpt 6 +4 +6-7 -18-3 Rnge of 3-3-3 to + 3-7 -6-1 to +2-9 differene from 45-4 9 to + 1 9-7-9 to + 7-4 mnul PT 6-6-4 to +2- -11l1 to + 12-9 Perentge 3 53-5 96-2 shorter thn 45 3-2 84-6 mnul PT 6 23-3 73-1
682 Thomson, Tberner, Poller 5 4 3 : 4: 51 4 3 oj: _ 4> 2 :,> 2- I, to 4 - i 13 14 15 16, 1 4, - w - 13-14 %15 :16-1 -1-2 Lol KC4/KC1 MNPT -2- Lol Cog--Mte MNPT 4 4 51 4 3- T 3- C C o -1 to 4-, 2 ' 4, 4, C 1 - @ 15 13 14 : 16 13 14 @ 15 16-1 -1- -2- Lol KC4/KC1 MNPT -2- Lol Cog--Mte MNPT Figure 4A-D Men perentge devition of ogulometer INR from mnulfor eh entre plotted ginst stted lol MNPT (A) When lol MNPT is used to derive INR for KC (B) When lol MNPT is used to derive INR for Cog--Mte (C) When mnufturer's stted mnul MNPT is used to derive INR for KC (D) When mnufturer's stted mnul MNPT is used to derive INR for Cog--Mte Figure 5 ffet of instruments on ISI vlues, disounting thromboplstin ISI between the two proedures ws smller nd not signifint The lol MNPT method to derive the INR overestimted the vlue by 8&3% wheres using the stted MNPT with MR, the INR were underestimted by 52% When the instrument INR is derived from the prothrombin rtio using the lol stted MNPT, with both the KC nd Cog--Mte, the shorter the reported lol MNPT the greter the perentge devition of the instrument INR from the mnul INR (figs 4 nd b) This negtive orreltion ws signifint t the 1% level for the two systems In ontrst, there ws no reltion between the devition nd the MNPT when the instrument INR ws derived from the mnufturer's stted mnul MNPT (figs 4 nd d) A wide rnge of pproximte ISI vlues for eh entre's utomted system, derived from the logrithmi orthogonl regression nlysis, is shown in fig 5, refleting the instrument differenes nd ignoring the thromboplstin effet
Comprison of two widely used ogulometers for mesuring prothrombin time Disussion The multientre field survey desribed in this report hs evluted the urrent performne of the two most widely used ogulometer systems in the United Kingdom nd hs shown onsiderble vribility within instruments with both systems s well s devitions from the mnul result In the British system for orl ntiogulnt ontrol reommended therpeuti rnges hve been derived from linil trils nd umultive experiene with the mnul PT tehnique For this reson the mnul tehnique hs been the referene PT method ginst whih utomted proedures hve been ssessed-the longstnding United Kingdom prtie Previous ogulometer studies hve been bsed on ssessments of single instruments of one or more types t one entre' or with groups of instruments by retrospetive nlysis of umultive dt obtined over n extended period in ntionl externl qulity ssessment This type of dtum is therefore limited nd does not relibly ssess performne in linil prtie Assessments of single instruments provide no indition of the vrition between instruments with multipliity of ogulometers of the sme type Retrospetive externl qulity ssessment bsed on limited dt on eh osion my not reflet urrent performne nd is inevitbly ffeted by instrument hnges nd djustments whih our over n extended period Our study shows tht substntil minority of the KC/Cog--Mte users obtined good greement with the mnul referene vlues independently estblished over the rnge of bnorml test plsms This indites tht lose pproximtion to the mnul result is possible using the two systems in ptients given oumrin The overll trend with the ogulometers, however, is to elerte the norml PT, lthough the extent vries onsiderbly mong entres The stted MNPT vlues for the two systems rnged between 13- nd 16-2 seonds with geometri men of 14-6 seonds for KC nd 14- for the Cog--Mte, ompred with 15-9 whih is the mnul geometri men for MR Unexpetedly, the elertion observed with the depleted smples did not prllel the shortening of norml vlues ssessed from the stted lolly derived MNPT With the KC t levels less thn therpeuti the PT ws generlly shorter thn tht of the mnul, but for vlues over 3 seonds the overll tendeny ws for results to beome progressively longer by omprison A different pttern ws observed with the Cog-Mte The trend to shorter PT beme reltively less pronouned with higher vlues The men perentge devition for INR lulted from the lol instrument MNPT for the KC4/KC1 ws + 15-2, nd tht for the Cog--Mte + 8 34 The men perentge devition lulted from the mnufturer's stted mnul MNPT ws + 4-76 for the KC4/ KC1O nd - 5-23 for the Cog--Mte These disrepnies, unless orreted, ombined with the degree of vrition of lolly reported 683 MNPT to undermine onfidene in rtio bsed ontrol system A further importnt observtion relted to the derivtion of the INR is tht the shorter the stted lol MNPT the greter the disgreement between instrument nd mnul INR The negtive orreltion of the lol MNPT nd the perentge devition from the mnul INR (figs 4 nd b) lerly shows tht the derivtion of INR vlues using the lolly derived norml vlue with either test system does not orret for the instrument effet Idelly, therefore, the instrument should be djusted to give lol MNPT s lose s possible to tht of the mnul The vrition in the MNPT is most likely to depend on the instrument Irrespetive of the use, the use of the lolly derived instrument MNPT is shown in this study to introdue onsiderble error in derivtion of INR vlues The bove effets on the norml vlues nd the differenes in orthogonl regression slopes of the bnorml smples given in figs 1 nd 2 might indite the need for djustment of the mnully derived ISI to tke into ount the instrument effets The extent of vribility between instruments illustrted by the differenes in slopes nd MNPT show tht it is urrently impossible to ssign system speifi ISI pplible to ll instruments of the sme type The gret degree of vrition between the entres n be guged by the instrument speifi lol ISI vlues determined from logrithmi orthogonl regression slopes (fig 5) In linil terms the men devition from the mnul INR vlue with both instruments is not gret, the respetive figures being -1 with the Cog--Mte nd + 3 with the KC for mnul INR of 4- Slightly more orl ntiogulnt would be presribed with Cog-Mte ontrol nd less with KC ontrol With some Cog--Mte nd KC instruments, however, effets on dosge my be muh more notieble thn these men figures suggest The derivtion of n ISI vlue bsed on lol full librtion of single instrument system might orret these dosge problems The limittions of this pproh re the exessive demnds on norml nd ptient smples for full librtion nd the inherent error of single librtion t one entre Furthermore, librtion exerise neessrily entils prllel observtions using the mnul referene tehnique This my not be fesible in ll entres The two prt KC study ddressed two dditionl issues-whether performne of the ogulometers ws onsistent over period of time nd whether rtifiilly dsorbed bnorml plsms gve the sme order of result s smples from ptients given oumrin The overll findings onfirmed both these Thus the vilbility of lyophilised bnorml plsms librted by the mnul method in PT nd INR, together with lyophilised norml smples, my provide mehnism for djustment of instruments to give results omprble with those of the mnul tehnique This should improve the uneptble vribility
684 Thomson, Tberner, Poller urrently found within instruments in this study with the two mjor ogulometer systems used in the United Kingdom nd might then llow system bsed ISI to be developed 1 Poller L, Thomson JM, Tberner DA ffet of utomtion on prothrombin time test in NQAS survey J Clin Pthol 1989;42:97-1 2 Poller L, Tberner DA, Thomson JM, Drby KV A survey of the prothrombin time results in NQAS exerises 198-1987 J Clin Pthol 1988;41:361-4 3 Tberner DA, Poller L, Thomson JM, Drby KV ffet of interntionl sensitivity index (ISI) ofthromboplstins on preision of interntionl normlised rtios (INR) J Clin Pthol 1989;42:92-6 4 World Helth Orgnistion xpert Committee on Biologil Stndrdistion: 34th Report WHO Tehnil Report Series Genev: WHO, 1984:19 5 World Helth Orgnistion xpert Committee on Biologil Stndrdistion: 33rd Report WHO Tehnil Report Series Genev: WHO, 1983:1-15 6 D'Angelo A, Seveso MP, D'Angelo SV, et l Comprison of two utomted ogulometers nd the mnul tilt tube method for the determintion of prothrombin time Am J Clin Pthol 1989;92:321-8