British Journal of Obstetrics and GynaecologyJanuary 1996, Vol. 103, pp. 39-47

ECPPA: randomised trial of low dose aspirin for theprevention of maternal and fetal complications in high

risk pregnant womenECP PA (Estudo Colaborativo para Preveno da Pr-eclampsia com Aspirina) Collaborative Group*

* Collaborators and participating centres are Iisted on pages 45-46

Objective To determine the effectiveness of low dose aspirin in women at high risk of adverseoutcomes associated with pre-eclampsia.

Design A collaborative randomised trial comparing the effects of low dose aspirin (60 mg) withplacebo on pre-eclampsia and other materno-fetal complications associated with hypertension.

Setting Twelve teaching maternity hospitais and 182 obstetricians' offices in Brazil.

Subjects One thousand and nine women considered to be at high risk for the development of pre-eclampsia, or its complications, entered the study between 12 and 32 weeks of gestation. Theywere randomly allocated to receive aspirin (498 women) or placebo (511 women) until delivery,and follow up was obtained for 96 %.

Resu1ts There were no significant differences between the treatment groups in the incidence ofproteinuric pre-eclarnpsia (6'7 % aspirin-allocated compared with 60 % placebo-allocatedwomen), of preterm delivery (22'3 % compared with 261 %), of intrauterine growth retardation(8'5 % compared with 101 %), or of stillbirth and neonatal death (7'3 % compared with 60 %),nor were there significant differences in the incidence of proteinuric pre-eclampsia in anysubgroup of women studied, including those who had systolic blood pressures of 120 mmHgor above at entry (8'5 % compared with 73 %) or those who were chronically hypertensive(10'0 % compared with 71 %). Aspirin was not associated with a significant excess of maternalor fetal bleeding.

ConcIusion The results of this study do not support the routine prophylactic administration of lowdose aspirin in pregnancy to any category of high risk women (even those who have chronichypertension or who are considered to be especially liable to ear1y onset pre-eclarnpsia).

INTRODUCTIONPre-eclampsia is a major cause of maternal andfetal morbidity and mortality', and placentalischaemia is considered to have a central role inthe pathogenesis of these complications". Pre-eclampsia is associated with deficient intravascularproduction of prostacyclin and with excessiveproduction of thrornboxane". There is also evi-dence of activation of the clotting system and earlyinvolvement of platelets". This has led to the use ofantiplatelet regimens (usually low dose aspirin) inan attempt to prevent ar delay the developmentand progression of the condition.Some small trials of antiplatelet therapy in

pregnancy have reported large reductions in the

Correspondence: Dr A. N. Atallah, Department of Medicine, DClinica Medica, Escola Paulista de Medicina, Rua Botucatu 740,CEP 04023-900 So Paulo, Brazil.

RCOG 1996 British Journal of Obstetrics and Gynaecology

incidence of pre-eclampsia with the use of \owdose aspirin (sometimes with the addition ofdipyridamole''"). But these findings have notgenerally been confirmed by more recent largerandomised controlled trials'"!'. Despite this it hasbeen suggested, usually after retrospective data-dependent subgroup analysis, that the benefits ofantiplatelet prophylaxis may still be of use incertain restricted groups of women. For example,the CLASP investigators concluded that low doseaspirin might be beneficial for those at especiallyhigh risk of early onset pre-eclampsiall. Similarexploratory analyses in another study'" led to thesuggestion that low dose aspirin was efficacious inprimigravid women presenting with systolic bloodpressures of 120 mmHg ar more. Pre-eclampsiaand its sequelae are relatively common in Brazil,and an observational study conducted there foundthat about half of the chronically hypertensive

39

40 ECPPA COLLABORATIVE GROUP

pregnant women had severe materno-fetalcomplications attributable to hypertension'". Thepresent report is of the Estudo Colaborativo paraPreveno da Pr-eclampsia com Aspirina(ECPPA). This multicentre randomised controlleddouble-blind trial was designed to determinewhether low dose aspirin is effective in women atparticularly high risk of adverse outcomes asso-ciated with pre-eclampsia.

METHODSOne thousand and nine women were recruited intothe trial from 12 university teaching hospitaIs and182 obstetricians' offices throughout Brazil be-tween December 1989 and March 1993. The studywas approved by the Ethics Committee Board ofEscola Paulista de Medicina, So Paulo.

EligibilityWomen were eligible if they were between 12 and32 weeks of gestation and, in the opinion of theresponsible clinician, were at sufficient risk of pre-eclampsia or its sequelae for the use of low dose

aspmn to be contemplated, but without clearindications for or against its use. Women might beconsidered at sufficient risk for a number ofreasons, including chronic hypertension detectedbefore or during pregnancy, primigravidity(especially with other risk factors, such as young orold age), diabetes, renal disease, a history of pre-eclampsia or intrauterine growth retardation(IUGR) in a previous pregnancy or evidence oftheir presence in the current pregnancy. Contra-indications included an increased risk of bleeding,asthma, allergy to aspirin, gastric ulcer, andplacenta praevia. Consent to participate was soughtfrom eligible women.

RandomisationEntry to the study was attained by telephoning acentral 24 h service at Escola Paulista de Medicinain So Paulo. Baseline details of the women (Table1) were recorded directly on computer-generatedrandomisation lists prepared by the Clinical TrialService Unit, Oxford University. Only after com-plete baseline information had been provided was

Table 1. Pre-randomisation characteristics of women studied. Figures in parentheses are percentages unless otherwise stated.

No. (%) in allocated treatment group

Aspirin(n = 498)

Placebo(n=511)

Woman's age (years)< 2020-2930-39~ 40

Estimated duration of gestation (weeks)< 12*12 ~ 20> 20 ~ 28> 28

Systolic blood pressure (mmHg)< 120120-139~ 140

Diastolic blood pressure (mmHg)< 9090-109~ 110

Other features of current pregnancyProteinuria andjor facial oedemaEvidence of IUGR

Obstetric and medical historyPrimigravidMultiparous, no fetal lossMultiparous, with fetal lossChronic hypertensionDiabetes or hyperglycaemia

27-5 (SD 7-4) 27-5 (SD 7-4)79 (16) 84 (16)

218 (44) 228 (45)172 (35) 174 (34)29 (6) 25 (5)22-1 (SD 6-2) 22-4 (SD 6-0)18 (4) 20 (4)

186 (37) 161 (32)194 (39) 233 (46)100 (20) 97 ( 19)127-3 (SD 20-5) 126-8 (SD 20-5)153 (31) 159 (31)171 (34) 183 (36)174 (35) 169 (33)81-3 (SD 15-0) 80-3 (SD 14-8)

314 (63) 333 (65)155 (31) 159 (31)29 (6) 19 (4)

21 (4) 27 (5)34 (7) 28 (5)

221 (44) 250 (49)188 (38) 175 (34)89 (18) 86 (17)

242 (49) 231 (45)25 (5) 37 (7)

* Women randomised before 12 weeks of gestation were to start treatment at 12 weeks,

RCOG 1996 Br J Obstet Gynaecol103, 39-47

a specific numbered trial treatment pack allocated.Women could be randomised before an estimatedgestational age of 12weeks, but in such cases wereinstructed not to start taking the tablets before the12th week. After randomisation, no woman wasexcluded from the trial, irrespective of whethertreatment was dispensed ar taken. For the pur-poses of analysis, women remained in the treatmentgroup to which they had been originally allocated(i.e., intention to treat analyses are reported).

TreatmentWomen were assigned calendar-packed treatmentwith either one 60 mg film coated aspirin tabletdaily ar a placebo tablet, identical in appearance,containing microcrystalline cellulose and comstarch. The dose of aspirin was chosen to besufficient to inhibit platelet aggregatiori'", and wasone that had been reported to prevent pre-eclampsia" while keeping side effects to a minimum.Women were asked to take the study treatmentevery day until delivery, unless advised otherwise.Other aspirin-containing preparations were to beavoided, with paracetamol recommended whenanalgesia was necessary. The actual contents ofthe allocated study treatment were not revealed,even after delivery, unless there was a c1earmedicalreason for the treatment to be made known. Drugstability was confirmed at intervals throughout thestudy by testing a sample of the study treatmentpacks.

Follow upA very simple single page follow up form wascompleted after hospital discharge of both motherand baby (ar at six weeks postpartum, ifeither hadnot been discharged). Brief details were to berecorded of proteinuria developing during thepregnancy, the highest recorded blood pressure(other than during labour), 1UGR, fetal loss orany maternal or neonatal bleeding. The mode ofdelivery, birthweight, whether live birth, stillbirthar neonatal death and any neonatal complicationswere also to be recorded. The duration of tablettaking was assessed crudely by recording theapproximate date when study treatment wasstopped and, in addition, a small random sampleofwomen in the study were interviewed about theircompliance. Efforts were made to check andcorrect any incomplete and inconsistent datawherever possible.

Outcome measuresThe main prespecified outcome measures were:estimated duration of pregnancy; maximum

RCOG 1996 Br J Obstet Gynaecol103, 39-47

ECPPA: A RANDOMISED TRIAL OF LOW DOSE ASPIRIN 41

maternal blood pressure recorded after entry;crude birthweight; stillbirth and neonatal death;maternal and fetal complications related tobleeding; blood transfusion. The study outcome ofproteinuricpre-eclampsia required thedevelopmen tof hypertension plus the detection of protein in theurine after randomisation. Hypertension wasdefined for those with baseline diastolic pressurebelow 90 mmHg as a rise of at least 25 mmHg to90 mmHg ar higher; for those with initial diastolicpressure of 90 mmHg ar above, an increment of15mmHg was required 11. Preterm delivery wasdefined as delivery before 37 weeks of estimatedgestation and 1UGR as birthweight below thethird centile for sex and estimated gestationalmaturity'". Stillbirths included ali deaths at arafter 24 weeks of gestation and neonatal deathsincluded all deaths after birth, up to 28 days.

Comparisons and statistical methodsThe main comparison was to be of all womenallocated aspirin against ali those allocatedplacebo. In addition, subsidiary comparisons weremade of the results subdivided according togestational age, parity and the existence of chronichypertension at randomisation. Further explora-tory analyses were conducted in response to someof the findings of CLASPl1 and other studies'".Statistical analyses involve simple comparisons oftotal numbers affected. Standard methods wereused to calculate the apparent ratio of the odds ofan outcome occurring in the aspirin group com-pared with the odds in the control group, along withits confidence interval: 95 % for principal analyses,and, to take account of the number of com-parisons, 99% for subgroup analyses14.15. Alter-natively, the reduction (or increase) in the odds ofthe event in the aspirin group and its standarddeviation (SD) are cited; an odds ratio of 0,8, forexample, corresponds to an odds reduction of20% (such odds reductions are slightly larger thanthe corresponding risk reductions).In the high risk women to be studied in ECPPA,

it should have been possible to detect reliably areduction in the incidence of pre-eclarnpsia ofabout three quarters (as suggested by the resultsavailable when this study was designed) in a studyof about 600 women. Such an effect seemed,however, to be too much to hope for and so thestudy was designed to detect somewhat smallerbenefits. Resources were available to continuerecruitment until March 1993 (when the study wasstopped in ignorance of the results), by which time1009 women had been randomised. No interimanalyses of ECPPA were conducted during re-

42 ECPPA COLLABORATIVE GROUP

cruitment and the results remained concealed untilafter data collection had been completed. Thereassuring data monitoring committee reports tothe steering committee of the larger CLASP studywere, however, provided to the principal inves-tigator of ECPPA during the study.

RESULTSOne thousand and nine women were randomised,with good balance between the treatment groupsfor the main pre-randomisation characteristics(Table I). Of the women enrolled, 16% wereunder 20 years of age, 38 % were at 20 weeks ofgestation or earlier, 47 % were primigravidae,47 % had chronic hypertension and 6 % had ahistory of diabetes mellitus or hyperglycaemia.Post-delivery follow up forms were obtained for

96 % of the randomised women (476 allocatedaspirin and 494 allocated placebo), and thesewomen had 985 infants or fetallosses (482 aspirincompared with 503 placebo). Reported compliancewith study treatment was good, with no differencebetween the groups allocated aspirin or placebo.Of the 967 randomised women for whom the dateof stopping trial tablets was known, 90 % stopped

Entrycharacteristic

(a)PROTEINURIC PRE-ECLAMPSIAEvents/Women Odds ratio & CI

Aspirin Placebo (Aspirin: Placebo)

Gestation

,; 20 weeks 16/192 8/172

> 20weeks 16/284 22/322

Parity

nulliparae 8/210 10/241

multlparae 24/266 20/253

Chronic hypertension

yes 23/231 16/224

no 9/245 14/270

Ali women entered: 32/ 476(6.7%)

30/494(6.1%)

only after 75% ofthe time between randomisationand delivery had been completed, and 69 %stopped after 95 % of this time. Interviews with arandom sample of 88 women in the studysupported the overall estimate of compliance, with88% of the sample confirming that they had takenmore than 75 % of the scheduled study tablets.

Incidence of proteinuric pre-eclampsiaProteinuric pre-eclarnpsia in ECPPA was recordedin 67 % of women allocated aspirin versus 61 %of those allocated placebo (Fig. Ia). Although thisrepresents an 11% (SD 28) increase in the odds ofdeveloping proteinuric pre-eclampsia, this diff-erence is not conventionally significant and is stillconsistent with a reduction of as much as one-quarter (as well as with more than a doubling inrisk). There was an absence of good evidence thatthe effect on proteinuric pre-eclampsia differedamong the different subgroups of women studied,inc1uding women with evidence of chronic hyper-tension (10'0% compared with 7'1%; Fig. Ia).or those who had systolic blood pressures of120 mmHg or over at entry: 28 (8' 5 %) of 331 such

(b)PRETERM DELIVERYEventslWomen Odds ratio & CI

Aspirin Placebo (Aspirin: Placebo)

53/192 49/172

53/284 80/322

41/210 50/241

65/266 79/253

56/231 70/224

501245 59/270

106/476 129/494~

~ 19%501411%SO28 (22.3%) (26.1%) reductionincrease (2p=0.2)(2p=0.7)

0.5 1.0 1.5 0.5 1.0 1.5Aspirin Asplrin Aspirin Aspirinbetter worse better worse

Fig. 1. Effects of aspirin on (a) proteinuric pre-eclampsia developing after randomisation and (b) preterm delivery. The outcome ofproteinuric pre-eclampsia required the development of hypertension and proteinuria after randomisation, and preterm delivery wasdefined as delivery before 37 weeks of estimated gestation (as in Cl.Af'!'). Odds ratios (. = area proportional to amount ofinformation contributed "') and 99% confidence intervals (Cl : horizontalline) are plotted for certain subgroups ofthe study population.A black square to the left of the solid verticalline suggests a benefit (but this is significant at 2P < 001 only if the whole Cl is to theleft of the solid verticalline). The overall results for ali women (and 95% Cl) are represented by diamonds, with the observed reductionsor increases in the odds of the outcome developing given to the right of the solid verticalline. X2 tests for differences between the effectsobserved in the different subgroups were ali nonsignificant.

ReOG 1996 Br J Obstet Gynaeco/103, 39-47

Entrycharacteristic

(a)INTRAUTERINE GROWTH RETAROATIONEvents/Babies Odds ratio & CI

Aspirin Placebo (Aspirin: Placebo)

Gestation

520 weeks 13/196 9/174

> 20 weeks 28/286 421329

Parity

nulllparae 14/214 19/249

mulliparae 27/268 321254

Chronic hypertension

yes 26/233 26/226

no 15/249 25/2n

Ali babies: 41/482 51/503 18%50 20(8.5%) (10.1%) reduction

(2p=0.4)

ECPPA: A RANDOMISED TRIAL OF LOW DOSE ASPIRIN 43

(b)STILLBIRTHS ANO NEONATAL OEATHSEvents/Babies Odds ratlo & CI

Aspirin Placebo (Aspirin: Placebo)

121196 9/174

231286 21/329

10/214 11/249

25/268 19/254

221233 17/226

13/249 13/2n

35/482 30/503(7.3%) (6.0%) 23%5029

Increase(2p=0.4)

0.5 1.0 1.5 0.5 1.0 1.5Aspirin Aspirin Asplrln Asplrlnbetter worse better worse

Fig. 2. Effects of aspirin on (a) intrauterine growth retardation and (b) stillbirth and neonatal death. Intrauterine growth retardationwas defined as birthweight below the third centile for sex and estimated maturity, and stillbirths and neonatal deaths as deaths at orafter 24 weeks gestation and up to 28 days after birth. Symbols and conventions as in Fig. 1. X2 tests for differences between the effectsobserved in the different subgroups were all nonsignificant.

women allocated aspirin compared with 25 (7'3 %)of 343 allocated placebo.There was also a lack of support for the

hypotheses generated by CLASP of a reduction inearly onset pre-eclampsia: among women whowere delivered before 32 weeks, pre-eclampsiaoccurred in 3/28 patients in the aspirin group andin 3/27 patients in the placebo group; amongwomen who were delivered between 32 and 37weeks, pre-eclampsia occurred in 10/78 patients inthe aspirin group, compared with 13/102 patientsin the placebo group. Among women who weredelivered after 37 weeks, pre-eclampsia occurredin 19/370 patients in the aspirin group and in14/365 in the placebo group (X2 test for trend =0,64, 2P = 0-4). Sixty women with a history ofdiabetes or hyperglycaemia were followed up;proteinuric pre-eclampsia developed in none of the24 women allocated aspirin and in only 3 out of 36of those allocated placebo (NS).There was no difference between the treatment

groups in the medians ofthe highest blood pressuresrecorded after randomisation and before labour(140/90 mmHg among both those allocated as-pirin and those allocated placebo). Proteinuriawithout hypertension severe enough to be definedas pre-eclampsia was slightly less common amongaspirin-allocated women (41 (8'6 %) comparedwith 52 (10'5 %); 2P = 0'3), while hypertension

RCOG 1996 Br J Obstet Gynaecol103, 39-47

without associated proteinuria was slightly morecommon (68 (14'3 %) compared with 56 (l1'3 %);2P = 0'2).

Duration of pregnancyThe mean duration of pregnancy was about twodays longer among aspirin-allocated women thanamong placebo-allocated women (38'08 weeks(SD 3-48) compared with 3778 weeks (SD 3'71)),but this difference was not statistically significant.The likelihood of preterm delivery, that is before37 weeks of estimated gestation, was lower amongwomen allocated aspirin (22'3 % compared with261 %: Fig. 1b). However, although the odds ofdelivering preterm was 19% (SD 14) lower amongaspirin-allocated women, this difference was notsignificant (95 % CI: 40 % reduction to 9 %increase). As was the case for proteinuric pre-eclampsia, the effects on preterm delivery did notappear to differ significantly in the differentsubgroups studied.

BirthweightThe mean birthweight of all babies bom to womenallocated aspirin was 30218 g (SD 763'3) com-pared with 29650 g (SD 756'0) in the placebogroup, but this slight increase of 568 g (SD 48'7)was not statistically significant. Aspirin was asso-

44 ECPPAcaLLABaRATIVEGRaUp

Table 2. Effects of aspirin on delivery type, bleeding and fetalloss after randomisation. Values are shown as n (%).

Aspirin Placebo

Pregnancies with data n = 476 n = 494Fetal outcomes n = 482 n = 503

Labour and deliveryCaesarean section 291 (61'1) 301 (60'9)Forceps delivery 36 (7-6) 36 (7'3)

Maternal bleedingPlacenta I abruption 5 (1'1 ) 7 (l A)Other antepartum bleed 6 (1'3) 8 ( 1,6)Postpartum bleed 3 (0,6) 6 ( 1'2)Transfusion 7 ( 1'5) 7 (l A)

Fetal bleedingIntraventricular haemorrhage 6 (1'2) 3 (0'6)Other neonatal bleeds 3 (0'6) 2 (OA)

Fetal lossesLosses < 24 weeks 4 (0'8) 6 (1'2)Stillbirths (~ 24 weeks) 28 (5'8) 23 (4'6)eonatal deaths 28 days) 7 (1'5) 7 (I A)

ciated with a slightly smaller prapartian of babieswith lUGR (8'5% compared with 101%: Fig.2a), but, again, this difference was not significant,either overall or in any of the subgroups studied.

Stillbirths and neonatal deathsFaur (0'8 %) fetallosses occurred before 24 weeksof gestation in the aspirin group and 6 (12%)accurred in the placebo graup (Table 2). Therewere 28 stillbirths plus 7 neonatal deaths (35 total:7'3%) in the aspirin group and 23 plus 7 (30:60%) in the placebo group (Fig. 2b). This 23%(SD 29) increase with aspirin is nat statisticallysignificant and the 95% confidence interval iswide. There was no apparent difference in theeffect in the variaus subgroups of women studied,nor were there any significant differences in thenumber of stillbirths and neonatal deaths asso-ciated with pre-ec1ampsia, maternal hypertensionar lUGR (21 (4-4%) compared with 26 (5'2 % orin thase associated with maternal or neonatalbleeding (50'04%) compared with 80'59%.

Other outcomesThere were no significant differences between thetreatment graups in delivery by caesarean sectionar forceps, nor were there any significant dif-ferences in placental abruptions or other antepartum bleeds (Table 2). All bleeds after deliverywere not explicitly recorded and were incompletelyreported (overall rate of 09% compared with26% in CLASP), but maternal transfusions weresystematically sought, and there was no differencebetween the treatment groups in the numbers

transfused. Two maternal deaths were reported inthis study: one in the aspirin-allocated group wasattributed to the HELLP syndrome and the otherin the placebo group was due to a car crash at 24weeks of pregnancy. No significant differences inthe incidence of intraventricular haemorrhages orother bleeds in the babies were observed.

DISCUSSIONThe incidence of proteinuric pre-ec1ampsia inECPPA was similar to that reported in previousstudies, but stillbirths and neonatal deaths weremore common (6'6 % compared with 28% inCLASpll, 16% in the American study'" and23% in the Italian study"). Chronic hypertensionwas present at entry in 47% of the ECPPApatients (compared with 20% in CLASP), and47 % were primigravidae (compared with 28% inCLASP). The perinatal mortality rate among thelarge chronically hypertensive group in ECPPAwas 85%, which may reflect uterine vascularlesions caused by chronic hypertension.Despite the high risk population studied in

ECPPA, the effects of aspirin on adverse outcomesappear to be much less promising than thosesuggested by the results of the first small trials, andsimilar to those of the more recent larger trials(Fig. 3: updated from CLASPll). Possible explana-tions for the discrepancy between the results ofthe small trials and the larger trials have beendiscussed in detail in the report of the CLASPstudy. In particular, it seems likely that this maybe due, at least in part, to publication bias, withsmall trials with unpromising results being lesslikely to be published than those with particularlypromising results, and with some methodologicalproblems!" in at least one small tria\.It was recently suggested " that the results of the

larger trials may have been diluted by their broadentry criteria and by the wide variations in carebetween the different participating countries.However, it has not been possible, either in CLASPor in ECPPA, to identify any particular categoryof women-inc1uding those in a substudy ofCLASP who were angiotensin II sensitive (as inone particularly promising small study"), or thosewith e\evated blood pressure at entry (as in a posthoc subgroup analysis of one of the recent largertrials10)-in whom the reduction in proteinuricpre-ec1ampsia was as great as that reported in theprevious small trials (Fig. 3a). Moreover, althoughit had been suggested from exploratory analyses ofCLASPll that aspirin may be justified for those atespecially high risk of early onset pre-ec1ampsia,this is not supported by the results from ECPPA.

RCOG 1996 Br J Obstei Gynaecol103, 39-47

Trial categoriesortrial

(a)PROTEINURIC PRE-ECLAMPSIANo 01 Antlplatelet Control Odds ratlo & 95,4 CItrlals therapy therapy (Antlplatelet: Placebo)

Small trlals

Wlth data 11 10/ 319 50/ 284(3.1%) (17.6%)

7 -/ 308 -/ 228

5 tOO/2697 139/2524

313/4659 3521 4650

32/ 476 30/ 494

7 445/7832 521/7668( 5.7%) ( 6.8%)

18 455/8151 571/7952( 5.6%) ( 7.2%)

Wlthout data

Larger trials

Before CLASPCLASPECPPA tAli larger trlalsAli trlals wlth data

ECPPA: A RANDOMISED TRIAL OF LOW DOSE ASPIRIN 45

(b)STILLBIRTHS ANO NEONATAL OEATHSNo 01 Antlplatelet Control Odds n.tlo & 95% CItrlals therapy therapy (Antlplatelet : Placebo)

12 51 306(2.0%)

-I 3206

10/ 289(3.5%)

-/ 223

5 48/ 2697

129/4810

35/ 482

38/2524

1361 4821

30/ 503

1%5010Inc .(2p:O.9)

1%5010reduct10n(2p:O.9)

17%506 7reductlon(2p=0.006)

23%506 19reductlon(2p=0.00006)

212/ 7989( 2.7%)

20417848( 2.6%)

218/ 8295( 2.6%)

214/ 8137( 2.6%)

o o 1.5Antlplatelettherapyworse

0.5Antlplatelettherapybatter

Test lor heterogenelty batween:

- ali trlals wlth data: X'" = 40.2 (p = 0.0004)X', = 25.7 (p < 0.000001)X'. = 11.8 (p = 0.07)

- smaller trials versus largar trials:

- alllarger trials:

0.5 1.0Antlplatelettherapybatter

1.0 1.5Antlpletelettherapyworse

X~, = 10.3 (p = 0.6)

X~ = 1.3 (p = 0.3)

X~ = 6.3 (p = 0.4)Fig. 3. Overview of effects reported from ali randomised trials of antiplatelet therapy in pregnancy on (a) proteinuric pre-eclampsia and(b) stillbirths and neonatal deaths. Symbols and conventions as in Fig. 1. Available results from smaller trials (i.e. those that includedfewer than 200 women) and from larger trials were combined using standard overview methods'" and stratified odds ratios plotted.Details of the trials included are given in CLASpll

Similarly, there were no subgroups m thesegenerally higher risk Brazilian women, in whomclear effects on lUGR, stillbirths or neonataldeaths could be demonstrated. Consequently, theECPPA results support the conclusion!' that ifsome special category of women exists that maybenefit substantially from aspirin, it must composea much smaller and more select group than hadpreviously been thought to be the case.As in the previous trials, the results of ECPPA

are generally reassuring as regards maternal andneonatal complications, with no significantexcesses of placental abruption, other antepartumhaemorrhage, transfusion, or mortality due tobleeding. Prior to ECPPA, a systematic overviewof the available results from alI randomised trialsof antiplatelet therapy indicated a 25 % reductionin the incidence of pre-eclampsia'", in contrastwith the reduction of about three-quarters sug-gested by the first small trials. The addition ofECPPA reduces still further the apparent size ofthis benefit (Fig. 3a), and if the small 'hypothesisgenerating' trials are excluded, the apparentreduction in the larger trials combined is only17% (SD 6). ln absolute terms, these more modestproportional reductions imply that antiplatelettherapy would typically prevent proteinuric pre-ec1ampsia in about 1 woman per 100 treated (withconfidence interval ranging from about zero to

RCOG 1996 Br J Obstet Gynaecol103, 39-47

about 2 per 100). ECPPA and some previous largescale trials!' do suggest that aspirin may prevent afew preterm deliveries per 100 women treated. But,overall, there is no evidence of an effect of aspirinon the incidence of stillbirths and neonatal deaths(218 (2'6 %) aspirin-allocated compared with 214(2'6 %) placebo-allocated deaths: Fig. 3b).ln conc1usion, as was suggested by the other

large trials, the present randomised placebo-controlled study in 1009 women, with very goodfollow up and compliance, does not support thewidespread routine use of aspirin for the pre-vention of pre-eclampsia or other hypertensivecomplications, even in the very high risk pregnantwomen of a developing country.

AcknowledgementsThe most important acknowledgement is to thehundreds of women who took part in ECPPA andto the doctors who collaborated throughout Brazil.The following investigators participated in thestudy.

Data management, analysis and writing committee:A N Atallah, R Collins, B Farrell, H Handoll.

Principal Investigator: A N Atallah, Americana: AFreitas Jr, L Kinsu, O Fukushima, Andradina: MAmorim, Araatuba: R Eduardo, Araraquara: ADurante, C Vieira, J Filho, Arararas: L Davolos,

46 ECPPA COLLABORA TIVE GROUP

Asis: R Zambotti, Bariri: C Negro, Batatais: OAlves, Bauru: P'Tobias, Belo Horizonte: C Freire,C Almeida, Birigui: J Neto, L Bertechini, Botu-catu: I Maest, J Peraoli, Marilza V Rudge, ISilva, S Filho, I Calderon, O Abujamra Jr, Brag-ana: A Neto, W Muniz, Jos Carlos Pinton, Mde Mello, Cachoeira Paulista: A Ferreira Jr,Campinas: C Nogueira, A Sanches, A Mariano, CFerraz Costa, R de Lacerda, A Elias, R Yoshiassu,S Lustre, S Ximenes, P de Godoy, Campinas: A deMeio, Canguu: D de Campos, Carapicuiba: DBezerra, Cataguases: F Cesrio, Caxias do Sul: DTessari, Cruzeiro: M Kisse, W Sria, Erechim-RS: A Teixeira, Fernandoplis: A Flumignan,Franca: M Marcolini, Guaruj: E Rimi, Guar-ulhos: C Simes, V de Arujo, Itatiba: CPavanelri, Jaboticabal: L Martins, Jacare: JNeto, C Bianco, C Antonlia, Jundia: E Gennari,Limeira: Z Vinhal, Lins: J Leo, Londrina: FSobrinho, Marlia: J Prado, Mirandoplis: ZSouza, Mogi: J Magalhes, R Esteves, Natal-RN:H de Oliveira, Novo Hamburgo--RS: L Jaeger,Novo Horizonte: R Melchiori, Olimpia: J Minari,Osasco: E Santana, I Machida, Ourinhos: FCesme, H de Carvalho, Pejuara-RS: L daSilva, Pindamonhangaba: A Wolff, Piracicaba: CNegretti, Porto Alegre: I Belli, L Napp, A Mene-ghetti, E Chaves, S Costa, C de Quadros Kroeff, SEspinosa, Presidente Prudente: D Campos, JTsello, Ribeiro Preto: A Matthes, C dos SantosJr, G Duarte, Rio Claro: G Neto, W de MatosRezende, Rio Grande--RS: P Gonalves, HRivoire, S. Bernardo: LC Joo, A Gradella, S.Jos do R. Preto: R Bertazzo, I Moraes, J Dria,M Trevisan, N Gabriel, Santa Maria-RS: J daSilva Ethur, F Jobim, Santo Andr: R Serre, SSanforlin, J da Silva, J Neves, O Ferraro, Santos:Ma dos Santos, A Ribeiro, R de Freitas, SoCaetano: A Adans, So Leopoldo--RS: I Plentz,So Paulo: J de Andrade, H Nightingale, JRebello, J Bencic, C da Costa Alves, C Camp-many, A Pereira, A Kataguiri, L de Campos, JMasonetto, D Bentivegna, M de Oliveira, MRusso, A Celestini, A Azevedo, H Lippi, FZanotto, F Simon, G Frehse, H Halbe, L Primon,P David, P Franco, P Pirozzi, D Klotzel, LAwoke, L de Figueiredo, R Theodozio, I Con-ceio, I Daniel, J Kublikowski, J Adalaft Neto,G Kenji, S Guimares, S Leung, T Gollop, VFreitas, B Carvalho, G Porto, J dos Santos, AAndrade, E Cavalcante, A Henrich, W Arl, HAri, T de Oliveira, D Silvestrini, E de Souza, A deArajo, A Allegrini, L Sakamoto, L Takano, RMattar, W Taborda, H Paraventi, M Miyazawa,N Sass, R de Souza Mesquita, M Lemos, M

Bevilqua, I Wulkan, G Paramo, J Mottola Jr, MScott, Sertozinho: A Sde, R Clemente, Soro-caba: L Neto, Sorocaba: C Barros, Sorocaba: NBressan, Taubat: M de Assis, X Mazzini.ECPPA Co-ordinating Centre: A N Atallah, EClarizia, M L Duarte, H Gonzalez, A Pantoja, MMesquita.Study Monitoring Committee: I Chalmers, RCollins, O Delascio (late), J A Grisso, R Peto, MZugaib.The Clinical Trial Service Unit (CTSU), NuffieldDepartment of Medicine, University of Oxford,UK provided technical support and encourage-ment throughout the duration of the study. Thestudy was principally funded by Sterling Drugs (PTribble, S Weisman), who also donated speciallypackaged aspirin and matching placebo andby Escola Paulista de Medicina, CNPQ andINCLEN, Inc. The study was, however, designed,conducted, analysed, and interpreted indepen-dently of the commercial sponsor.

ReferencesI Davies AM. Epidemiology ofhypertensive disorders ofpregnancy.

Buli World Health Organ 1979; 57: 373.2 Redman CWG. Current topic: pre-eclampsia and the placenta.

Placenta 1991; 12: 301-308.3 Bussolino F, Benedetto C, Massobrio M, Camussi G. Maternalvascular prostacyclin activity in pre-eclampsia. Lancet 1980; 2:702.

4 Redman CWG, Bonnar J, Beilin L. Early platelet consumption inpre-eclampsia. BM] 1978; I: 467-469.

5 Beaufils M, Uzan S, Donsimoni R, Colau .IC. Prevention of pre-eclampsia by early antiplatelet therapy. Lancet 1985; I: 840-842.

6 Wallenburg HCS, Dekker GA, Makovitz JW, Rotmans P. Low-dose aspirin prevents pregnancy-induced hypertension and pre-eclampsia in angiotensin-sensitive primigravidae. Lancet 1986; I:1-3.

7 Schiff E, Peleg E, Goldenberg M el ai. The use of aspirin toprevent pregnancy-induced hypertension and lower the ratio ofthromboxane A2 to prostacyclin in relatively high risk pregnancies.N Engl J Med 1989; 321: 351-356.

8 McParland P, Pearce JM, Chamberlain GVP. Doppler ultrasoundand aspirin in recognition and prevention of pregnancy-inducedhypertension. Lancet 1990; 335: 1552-1555.

9 Italian Study of Aspirin in Pregnancy. Low-dose aspirin inprevention and treatment of intrauterine growth retardation andpregnancy-induced hypertension. Lancet 1993; 341: 396-400.

10 Sibai BM, Caritis SN, Thom E et ai. Prevention of pre-eclampsiawith low-dose aspirin in healthy, nulliparous pregnant women.New Engl] Med 1993; 329: 1213-1218.

11 CLASP (Collaborative Low-dose Aspirin Study in Pregnancy)Collaborative Group. CLASP: a randomised trial of low-doseaspirin for the prevention and treatment of pre-eclampsia among9364 pregnant women. Lancet 1994; 343: 619-629.

12 Atallah AN, de Souza Mesquita MR, Duarte ML et ai. Estudoprospectivo "cohort ' de gestantes com hipertanso arterialcrnica.v Bras Nefro11990; 12: 113-120.

13 Benigini A, Gregorini G, Frusca T et ai. Etfect oflow-dose aspirinon fetal and maternal generation of thromboxane by plateletsin women at risk of pregnancy-induced hypertension. N Engl JMed 1989; 321: 357-362.

RCOG 1996 Br J Obstet Gynaecol 103, 39-47

14 Peto R, Pike MC, Armitage P et ai. Design and analysis ofrandomized clinical trials requiring prolonged observation or eachpatient l l. Analysis and examples. Br J Cancer 1977; 35: 1-39.

15 Antiplatelet Trialists' Collaboration. Collaborative overview ofrandomised trials of antiplatelet therapy-I: Prevention of death,myocardial infarction, and stroke by prolonged antiplatelettherapy in various categories ofpatients. BMJ 1994; 308: 81-106.

RCOG 1996 Br J Obstet Gynaecol103, 39-47

ECPPA: A RANDOMISED TRIAL OF LOW DOSE ASPIRIN 47

16 Atallah AN, Shinar R. Pre-ec1ampsia and prostaglandins. LanceI1990; 1: 1267.

17 Beilin L. Aspirin and pre-ec1ampsia. BMJ 1994; 308: 1200-1251.

Received 27 April 1995Accepted 13 September 1995

British Journal of Obstetrics and GynaecologyJanuary 1996, Vol. 103, pp. 48-53

Detection of fetal fibronectin as a predictor of pretermdelivery in high risk asymptomatic pregnancies* s. C. Leeson Senior Registrar, *M. J. A. Maresh Consultant, ** E. A. Martindale Registrar,tT. Mahmood Registrar, * A. Muotune SHO, tt N. Hawkes SHO, * K. J. Baldwin Registrar

* Department of Obstetrics and Gynaeeology, Saint Mary's Hospital, Manehester, ** Department of Obstetrics andGynaeeology, Bolton General Hospital; t Department of Obstetrics and Gynaeeology, North Manehester General Hospital;

tt Department of Obstetrics and Gynaeeology, Gloueester Royal Hospital

Objective The study was designed to determine whether fetal fibronectin would predict deliverybefore 37 weeks in women at high risk of preterm delivery.

Study methods Forty-three women considered at risk ofpreterm delivery were recruited antenatallyinto a blind longitudinal study. Quantitative assays of fetal fibronectin were obtained fromsequential high vaginal swabs taken fortnightly from 24 to 34 weeks of gestation. Fibronectinconcentrations of 005 Ilg/ml or more were considered as positive.

Results Results were calculated by swab and by subject. The sensitivity of an individual fibronectinswab in predicting preterrn delivery within 14 days of testing was 71 % and the specificity was93 %. The overall positive predictive value was 31 % and the negative predictive value was99 %. The sensitivity of the fibronectin swab in predicting delivery before 37 weeks was 17%and the specificity was 93 %. The positive predictive value was 50% and the nega tive predictivevalue was 73 %. For a woman who has had a positive swab the sensitivity in predicting pretermdelivery within 14 days of testing was 80 % and the specificity was 83 %; a woman was countedas positive only if the final swab was positive and preceded delivery by not more than 14 days.The positive predictive value was 36% and the negative predictive value was 97%. For awoman who has had a positive swab the sensitivity in predicting delivery before 37 weeks was54 %. The specificity, the positive predictive value and the nega tive predictive value were 85 %,64 % and 79 %, respectively. Women were counted as positive if any swab in the samplingsequence was positive. Fibronectin swabbing when calculated by patient did predict pretermdelivery within 14 days of testing (P = 0'01) and before 37 weeks (P = 0'01). Analysis of theaccuracy of predicting delivery from 7 to 28 days after sampling revealed that the bestprediction for delivery was within the following 14 days.

Conclusion Serial fetal fibronectin assessment from 24 to 34 weeks of gestation anticipated pretermdelivery within 14 days oftesting and before 37 weeks for high risk asymptomatic women. Suchtesting should be performed every two weeks.

INTRODUCTIONPreterm delivery is the leading cause of newborninfant mortality in the developed world. Most ofthese deaths occur in pregnancies ending before29 weeks'. A substantial morbidity accompaniespreterm delivery as well as iatrogenic morbidityassociated with prolonged neonatal intensive care.Long term neurological problems are more difficultto evaluate. Apart from the human cost, thefinancial cost of providing inpatient care forpreterm babies and providing support in cases oflifelong handicap is great.

Correspondence: Dr S. C. Leeson, Department of Obstetricsand Gynaecology, Saint Mary's Hospital, Hathersage Road,Manchester M 13 9PT, UK.

48

Prediction ofpreterm labour is unreliable. Usinga history of previous preterm delivery, twinpregnancy, uterine abnormality, low socio-econ-omic status and cervical incompetence predicts, atbest, 50% of preterm deliveries". The use of serialvaginal examination, screening for reduced fetalbreathing movements on ultrasound scanning orscreening for recurrent contractions with externaltocography provides little additional sensitivity".Fetal fibronectin is part of a family ofubiquitous

dimeric glycoproteins present predominantly inplasma and extracellular matrix and whichinfluence cell adhesion, motility, tissue repair andcoagulation+". There are more than 20 isoforms ofthe molecule. Fetal fibronectin has a molecularweight of about 450,000 daltons and is produced

RCOG 1996 Britisli Journal of Obstetrics and Gynaeeology