ORIGINAL ARTICLE Reproductive epidemiology

Prevalence and risk factors forobstetric haemorrhage in 6730singleton births after assistedreproductive technology inVictoria AustraliaD.L. Healy1,2,8, S. Breheny2, J. Halliday6,7, A. Jaques6, D. Rushford3,C. Garrett3, J.M. Talbot5, and H.W.G. Baker3,41Department of Obstetrics and Gynaecology, Monash University, Level 5, Monash Medical Centre, 246 Clayton Road, Clayton, VIC 3168Australia 2Monash IVF, Richmond, Australia 3Department of Obstetrics and Gynaecology, University of Melbourne, Melbourne, Australia4Melbourne IVF Reproductive Services, The Royal Womens Hospital, Parkville, Australia 5Melbourne Assisted Conception Centre,Heidelberg, Australia 6Murdoch Childrens Research Institute, Melbourne, Australia 7Department of Paediatrics, University of Melbourne,Melbourne, Australia

8Correspondence address. Tel: 61-03-9594-5374; Fax: 61-03-9594-6389; E-mail: david.healy@med.monash.edu.au

background: Obstetric haemorrhages have been reported to be increased after assisted reproduction technologies (ART) but themechanisms involved are unclear.

methods: This retrospective cohort study compared the prevalence of antepartum haemorrhage (APH), placenta praevia (PP), placentalabruption (PA) and primary post-partum haemorrhage (PPH) in women with singleton births between 1991 and 2004 in Victoria Australia:6730 after IVF/ICSI, 24 619 from the general population, 779 after gamete intrafallopian transfer (GIFT) and 2167 non-ART conceptions ininfertile patients. Risk factors for haemorrhages in the IVF/ICSI group were examined by logistic regression.

results: The IVF/ICSI group had more APH: 6.7 versus 3.6% (adjusted OR 2.0; 95% CI 1.82.3), PP: 2.6 versus 1.1% (2.3; 1.92.9),PA: 0.9 versus 0.4% (2.1; 1.43.0) and PPH: 11.1 versus 7.9% (1.3; 1.21.4) than the general population. APH, PP and PA were as frequentin the GIFT group as in the IVF/ICSI group, but were less frequent in the non-ART group. Within the IVF/ICSI group, fresh compared withfrozen thawed embryo transfers (FET) was associated with more frequent APH (1.5; 1.21.8) and PA (2.1; 1.23.7) and the oddsratio increased with number of oocytes collected (1.02; 1.001.04). Endometriosis patients had more PP (1.7; 1.22.4) and PPH (1.3;1.11.6) than those without endometriosis. FET in articial cycles was associated with increased PPH (1.8; 1.32.6) compared with FETin natural cycles.

conclusions: Obstetric haemorrhages are more frequent with singleton births after IVF, ICSI and GIFT. The exploratory analysis offactors in the IVF/ICSI group, showing associations with fresh embryo transfers in stimulated cycles, endometriosis and hormone treatments,suggests that events around the time of implantation may be responsible and that suboptimal endometrial function is the critical mechanism.

Key words: IVF / obstetrics / antepartum haemorrhage / post-partum haemorrhage / assisted reproductive technology

IntroductionObstetric haemorrhage and primary post-partum haemorrhage (PPH)are important causes of maternal mortality worldwide (World HealthOrganisation, 2007). Antepartum haemorrhage (APH), placentapraevia (PP) and placental abruption (PA) are major obstetric compli-cations in all countries. Increased frequencies of these complications

with births after assisted reproduction technology (ART) have beenreported by several groups (Daniel et al., 2000; Perri et al., 2001;Smithers et al., 2003; Jackson et al., 2004; Kallen et al., 2005b;Shevell et al., 2005; Romundstad et al., 2006; Schieve et al., 2007).Although increased multiple births after ART were an importantfactor increasing obstetric haemorrhages, PP and PA are also knownto be more frequent with singleton deliveries after ART than after

& The Author 2009. Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved.For Permissions, please email: journals.permissions@oxfordjournals.org

Human Reproduction, Vol.25, No.1 pp. 265274, 2010

Advanced Access publication on November 7, 2009 doi:10.1093/humrep/dep376

by guest on February 24, 2014http://hum

rep.oxfordjournals.org/D

ownloaded from

natural conceptions in matched controls but the mechanisms involvedare not clear and deserve further study (Schieve et al., 2007).

We have previously reported a high frequency of APH and PP in IVFtwin pregnancies (Smithers et al., 2003). We speculated this was aconsequence of embryo transfer through the vagina and cervix, com-pared with in-vivo conception and implantation via the uterine tube.More recently, Romundstad et al. (2006) conrmed an increasedrisk of PP [adjusted OR 5.6, 95% condence interval (CI) 4.47.0]in 5581 singleton ART births. They also compared results of twobirths in the same woman who had both an ART and a natural con-ception and found that the risk of PP was also higher (OR 2.9) inthe ART pregnancy. These authors suggested the higher risk of PPwas largely attributed to factors related to the ART and that APHand PP may occur as a consequence of embryo transfer through thevagina and cervix.

Studying factors affecting obstetric haemorrhage after ART isimportant because ART accounts for 35% of all births in developedcountries and is increasing in use. Understanding the mechanismsmight lead to prevention not only of these haemorrhages in ART preg-nancies but also those in the general community. PPH has not pre-viously been reported in detail in ART patients (Kallen et al.,2005b), yet PPH is a leading cause of maternal death. Studying single-ton births is appropriate with the general move to elective singleembryo transfer to minimize the multiple pregnancy rate with ART.

Here we report a multi-centre study of obstetric haemorrhage andPPH in singleton births of patients who conceived with IVF or ICSI andcomparison groups from the general community, women who con-ceived with gamete intrafallopian transfer (GIFT), and infertilewomen who became pregnant without ART. Our strategy was asfollows: rstly, to replicate the ndings that the risks of these haemor-rhages are greater after IVF/ICSI than in the general population afteradjusting for confounding factors; secondly, to compare the frequen-cies of the haemorrhages after IVF/ICSI and GIFT to assess theeffect of manipulation and culture of the embryos in the embryologylaboratory as this does not occur with GIFT; thirdly, to compare thefrequencies of the haemorrhages after IVF/ICSI and non-ART con-ception in infertile women to examine the effect of infertility; andfourthly, to examine factors within the IVF/ICSI group that might berelated to haemorrhages.

Materials and MethodsThe project was approved by each Human Research and Ethics committeeof the hospitals involved and the Victorian Department of Human Services.

Data sourcesArt clinic databasesThe infertility clinics are required to keep comprehensive records of allprocedures that result in birth by Victorian State regulations and forreporting to the Fertility Society of Australias Reproductive TechnologyAccreditation Committee and all patients sign consent forms acknowled-ging this. The databases include information on clinical features, diagnosis,treatments and the embryos. Additional information about investigationsand lifestyle factors was added from the patients records.

Victorian birth dataIn the Victorian Department of Human Services, the Perinatal Data Collec-tion Unit (PDCU) oversees mandatory registration of all births of 20weeks gestation and over. There were approximately 862 000 singletonbirths in the study period (19912004). Miscarriages and terminationsof pregnancy at ,20 weeks gestation are not registered as births in Vic-toria unless a birth defect is recorded. Data on the existing maternalmedical conditions, previous pregnancies, gestation at birth, maternalcomplications of pregnancy and mode of delivery are provided on formsby midwives involved with the births and entered on computer centrally.Over 99.6% of all births in Victoria were reported to the PDCU in 1996(Riley and Grifn, 1997).

SubjectsThe ART data relate to women treated in the state of Victoria by any ofthree infertility services: Melbourne IVF, Monash IVF and the MelbourneAssisted Conception Centre (MACC) between January 1991 andDecember 2004. Only the subjects rst singleton birth during the studyperiod were included. This is a retrospective cohort study using recordlinkage to obtain the information on obstetric haemorrhages in infertilewomen conceiving after IVF or ICSI and three comparison groups:women conceiving in the general community and infertile womenconceiving by GIFT or without ART.

In vitro fertilization/intracytoplasmic sperm injectionFor this group, oocytes were collected by transvaginal ultrasound guidedoocyte collections under neurolept or general anaesthesia. ICSI was intro-duced in 1993. Treatment of male infertility was attempted with standardIVF before the introduction of ICSI. Stimulation regimens involved clomi-phene, human menopausal gonadotrophin and human chorionic gonado-trophin up to 1993. The oral contraceptive pill was used increasinglyafter 1995 to regulate the starting time of gonadotrophin treatment.Long down-regulation with gonadotrophin releasing hormone agonistswas used after 2001. Recombinant FSH was used after 1994. Gonado-trophin releasing hormone antagonists were used in increasing numbersof patients after 2000.

Culture media changed from human tubal uid to multi-stage media in2000. The protein supplement was human serum albumin until 2000 andhighly puried albumin thereafter. Embryo cryopreservation involved pro-panediol with embryo freezing on day 2, 3 or 5. Cryopreserved embryoswere thawed and transferred on the same day as their age after ovulationin natural or articial cycles until 1996. After September 1997, all cryopre-served embryos were thawed the day before transfer and cultured over-night so they would be on average 18 h more advanced than embryostransferred fresh. Frozen-thawed ET (FET) was usually performed innatural menstrual cycles but women with oligomenorrhoea or amenor-rhoea had articial cycles induced with cyclical estradiol and progesteronetreatment. Some other patients had luteal phase support with hCG orprogesterone. During the time of the study relatively low numbers ofembryos were transferred and the multiple pregnancy rate was ,20%.Embryo cryopreservation was common and approximately one third ofthe singleton births were from FET.

National health funding of ART covered about half of the patients costs.Patients without private health insurance or their own specialist gynaecol-ogist were classied as semiprivate and this is an indicator of socio-economic status.

General populationFor each IVF/ICSI subject, three records of singleton births were randomlyselected from the PDCU data matching on year of maternal age and yearof birth of the baby to balance the exposed and unexposed groups on

266 Healy et al.

by guest on February 24, 2014http://hum

rep.oxfordjournals.org/D

ownloaded from

maternal age and possible temporal changes in ascertainment of the obste-tric haemorrhages.

Gamete intrafallopian transferThe results of singleton births from GIFT were analysed to determine if thein-vitro fertilization and embryology laboratory procedures contributed tothe outcome, as these do not occur with GIFT. GIFT was performedunder general anaesthesia and involved laparoscopic placement of theoocytes and spermatozoa into the ampulla of a uterine tube via the m-brial end of the tube. Most of the GIFT was performed between 1990and 1998. Except at one centre (MACC), extra oocytes not transferredwere inseminated and embryos resulting were cryopreserved.

Non-ARTInfertile women who conceived without ART (non-ART group) were ana-lysed to determine the contribution of infertility itself to the outcomes.The women were classied as infertile because they were seen in theMelbourne or Monash clinics between 1991 and 2000 but were knownnot to have given birth after IVF, ICSI, GIFT or donor insemination atthe three Melbourne clinics. They were found in the Victorian PDCUbirth data to have given birth to a singleton baby within 5 years of regis-tration at the clinics. They may or may not have had ART treatment butany ART treatments did not result in a birth after 20 or more weeks ges-tation. Some patients may have been treated by ovulation induction orarticial insemination but most would have naturally conceived. The possi-bility that they had conceived by ART at other clinics interstate or overseaswas checked by inspection of the patient records. The three Melbourneclinics performed most of the ART procedures in the State during thetime of the study (.98% of clinical pregnancies, www.ita.org.au) andreproductive tourism was infrequent.

ExclusionsDonor oocyte and embryo recipients (n 339) were excluded. Bychance, some women in the IVF/ICSI, GIFT and non-ART groups whoalso had naturally conceived children were matched with other IVF/ICSIgroup subjects. Similarly some women with more than one birth werein the general population group. All these second and subsequent preg-nancies in the study groups were excluded so that each woman wasonly analysed once thus avoiding correlations between outcomes thatmight otherwise occur.

Primary outcome: haemorrhagesThe obstetric haemorrhages were dened using the Victorian PDCU elec-tronic user manual. APH was coded when bleeding of 15 ml or moreoccurred from the birth canal after 20 weeks gestation and before thebirth of the baby. This overarching category includes PP, PA and APHfor other reasons. PP was coded where the placenta was located overor very close to the cervical os as veried by independent ultrasoundexamination and conrmed at birth. APH may or may not have occurred.The standardized form sent to the PDCU does not differentiate betweenmarginal PP (grade I) from more major PP (grade II III) to grade IV PPwhich cover the internal cervical os. PA was coded where there was pre-mature separation of the placenta. There was evidence of retroplacentalbleeding and the amount of blood lost was always moderate or severewhether out of the vagina or internally. PPH was coded when there wasblood loss greater than 500 ml within 24 h of vaginal delivery or childbirthor greater than 750 ml after Caesarean section. The PDCU denitions areconsistent with the International Classication of Diseases (ICD10) codingsystem. In this study, six women had both PP and PA. Also 70 women withPP had no bleeding and were not classied as having APH.

Record linkageWith the aid of a computer programme, LinkageWiz, probabilistic recordlinkage was used to match the ART data to the PDCU common identifyingvariables: name, date of birth and postcode (Tayor, 1998). Once recordlinkage was complete, IVF clinical data were used to ensure matches to sin-gleton births were correct. Cross-checking between the Victorian ARTclinics was done to ensure the birth was not the result of ART atanother clinic. Data were extracted for rst singleton births within thestudy period of greater than 19 weeks gestation for the groups: IVF/ICSI (6730), population control (24 619), GIFT (779) and non-ART(2167). Stillbirths, late terminations and neonatal deaths (within 28 daysof birth) were included.

Data analysisThe background characteristics of the births in each of the four compari-son groups were described using means and ranges for quantitative vari-ables and percentages for categorical variables. The proportions ofAPH, PP, PA and PPH in each group were compared between the IVF/ICSI group and each of the general population, GIFT and non-ARTgroups using the x2 test. Logistic regression was performed usingGenStat 11th Edition (McCullagh and Nelder, 1989) to compare theodds of each complication between the IVF/ICSI and general populationgroups in unadjusted analyses and analyses adjusted for potential confoun-ders: age, year of birth of baby, marital status, parity, previous miscarriageor termination of pregnancy, country of birth and vertex presentation(Model 1). Also giving birth in tertiary, private or other maternity hospitalswas also included with the other factors in model (Model 2). For PPH, theeffect of adjusting for other complications of pregnancy: pre-eclampsia,APH, PP, PA, premature rupture of membranes and induced labour,were added to Models 1 and 2. Also, because of the different denitionsof post-partum haemorrhage for vaginal and Caesarean section deliveries,these were examined separately.

To determine which factors within the IVF/ICSI group were indepen-dently related to obstetric haemorrhages, causes of infertility, ovarianstimulation, fertilization procedure (IVF or ICSI), transfer of fresh orfrozen embryos, number of embryos transferred, average embryoquality, average cell number, luteal phase support and number of fetalhearts were examined by logistic regression to adjust for the potential con-founders listed above. Graphs were plotted to check the linearity andnature of the relationships between the data and the t of the models.Additional analyses were performed omitting factors which might be aneffect of the obstetric complication being studied, such as birth in a tertiarymaternity hospital, to determine if these affected the relationship betweenthe ART or infertility factor and the end-point. To study the possibleeffects of ovarian stimulation, oocyte numbers, fertilization rate andembryo numbers, and births resulting from fresh embryo transfer wereanalysed separately. Similarly, FET was analysed separately to examinefactors specic to this group such as the effects of natural and articialcycles and overnight culture of embryos after thawing.

Results

IVF/ICSI group compared with the controlgroupsThe main differences in characteristics in the subjects in the groups aresummarized in Table I. Compared with the IVF/ICSI group, thegeneral population group were more likely to be born in Asia or theMiddle East and migrated to Australia, to have previous pregnanciesand to have a vertex presentation, and less likely to be married,

Increased obstetric haemorrhage after ART 267

by guest on February 24, 2014http://hum

rep.oxfordjournals.org/D

ownloaded from

give birth in a private hospital, have an induced labour or have an Cae-sarean section. Compared with the IVF/ICSI group, the GIFT patientswere slightly younger, more often married, treated earlier in the studyperiod and less likely to have Caesarean sections. The non-ART groupwere younger, more often born in Asia or the Middle East, more likelyto have previous pregnancies, miscarriages and termination of preg-nancy, and less likely to use a private hospital and have a Caesareansection than the IVF/ICSI group. The IVF/ICSI group had higherrates of all the haemorrhagic complications than the general popu-lation (Table II). For GIFT, APH was more frequent than in the IVF/ICSI group, whereas frequencies of PP and PA were similar and post-partum haemorrhage was less frequent than in the IVF/ICSI group.The non-ART group had frequencies of these complications inter-mediate between those in the general population and the IVF/ICSIgroup (Table II).

The IVF/ICSI group had more of the obstetric haemorrhage com-plications than the general population group. The ORs changed littlefor APH, PP, PA and other APHs after the adjustments for confoun-ders were made (Table III). For post-partum haemorrhage, adjustingfor covariates reduced the OR. The factors producing the greatestchange in the OR were parity and Caesarean section. Analysing

vaginal deliveries and Caesarean section births separately producedsimilar results. Thus there is some evidence for an additional effectof IVF/ICSI on post-partum haemorrhage independent of theincreases in other maternal complications of pregnancy. The differencein the frequencies of the haemorrhages in Primiparous women in theIVF/ICSI group and in the general population group, by age, is shownin Fig. 1.

Factors within the IVF/ICSI group affectingthe risk of the haemorrhagic complicationsFresh versus frozen-thawed embryo transfersAPH was more frequent with fresh embryo transfer than with FET(Tables IV and V). Adjusting for other covariates made little differenceto the OR for fresh embryo transfer versus FET. The weak evidence(P 0.10.01) for effects of female infertility, tubal disease, threeor more embryos transferred and luteal phase support with hCGwas lost when fresh embryo transfer versus FET was included in theregression models. The data for PA were sparse (not shown) butthere was moderately strong evidence (P 0.01) that PA was morefrequent with fresh embryo transfer (49/422, 1.2%) than with FET

.............................................................................................................................................................................................

Table II Obstetric haemorrhages: APH, PP, PA and PPH in IVF/ICSI and general population, GIFT and Non-ART groups

Complication IVF/ICSI (6730) General Population (24 619) GIFT (779) Non-ART (2167)

APH 454 (6.7) 881 (3.6), P, 0.001 78 (10.0), P 0.001 98 (4.5), P, 0.001PP 174 (2.6) 273 (1.1), P, 0.001 21 (2.7), P 0.85 26 (1.2), P, 0.001PA 63 (0.9) 107 (0.4), P, 0.001 11 (1.4), P 0.20 13 (0.6), P 0.14PPH 746 (11.1) 1954 (7.9), P, 0.001 54 (6.9), P, 0.001 192 (8.9), P 0.003

APH, antepartum haemorrhage; PA, placental abruption; PP, placenta praevia; PPH, primary post-partum haemorrhage; GIFT, gamete intro-fallopian transfer.The number and (%) in each group is shown with the P-value for the comparison with the IVF/ICSI group.

.............................................................................................................................................................................................

Table I Characteristics of the subjects

Characteristic IVF/ICSI General population GIFT Non-ART

Number 6730 24 619 779 2167

Age 34 (2045) 34 (2050) 33 (2346) 33 (1947)

COB Asia or ME* 641 (9.5) 3100 (12.6) 54 (6.9) 273 (12.6)

Married 6326 (94.0) 20 345 (82.6) 766 (98.3) 2048 (94.5)

Gravidity 0.8 (016) 1.8 (022) 0.7 (06) 1.1 (013)

Parity 23 1345 (20.0) 13 880 (56.4) 174 (22.3) 530 (24.4)

.3 122 (1.8) 3204 (13.0) 8 (1.0) 39 (1.8)

Any abortions** 2013 (29.9) 8790 (35.7) 197 (25.3) 898 (41.4)

Tertiary hospital 1566 (23.3) 5381 (21.8) 207 (26.6) 488 (22.5)

Private hospital 3941 (58.6) 8528 (34.6) 437 (56.1) 1184 (54.6)

Year of birth 1999 (19912004) 1999 (19912004) 1995 (19912004) 1997 (19912004)

Vertex presentation 6143 (91.3) 23 126 (93.9) 703 (90.2) 1995 (92.1)

Induced labour 2138 (31.8) 6383 (25.9) 214 (27.5) 644 (29.7)

Caesarean section 2681 (39.8) 6689 (27.2) 236 (30.3) 763 (35.2)

The number and (%) or mean and (range) is shown.*COB: country of birth; ME: middle East **Abortions include miscarriages and terminations of pregnancy before 20 weeks gestation.

268 Healy et al.

by guest on February 24, 2014http://hum

rep.oxfordjournals.org/D

ownloaded from

(14/2503, 0.6%), OR 2.1 (1.23.7), Model 2 adjusted OR 2.1 (1.23.9). There was also some evidence that fresh transfers were associ-ated with increased PP, but with less post-partum haemorrhage, indi-cating increased rates of post-partum haemorrhage with FET(Table IV).

The results for fresh embryo transfer in FSH stimulated cycles (freshtransfers in natural cycles excluded) and for FET in natural cycles (arti-cial cycles excluded) are compared in Table V. There were more

APH and PA after fresh embryo transfer. There was also a trend tohigher PP with the fresh transfers but no evidence of a difference inthe frequency of post-partum haemorrhage with fresh embryo transferor FET.

Number of oocytes collected before fresh embryo transfersAnalysing the results of fresh embryo transfer alone produced moder-ate evidence for an increase in frequency of APH with the number of

.............................................................................................................................................................................................

Table III Odds ratios of obstetric haemorrhages: APH, PP, PA and PPH in the IVF/ICSI group compared with the generalpopulation group unadjusted and adjusted for possible confounding

OR (95% CI), P-value Model 1 Ad OR (95% CI), P-value Model 2 Ad OR (95% CI), P-value

APH 1.95 (1.732.19), ,0.001 1.98 (1.732.26)*, ,0.001 2.04 (1.782.33)*, ,0.001

PP 2.37 (1.952.87), ,0.001 2.27 (1.822.83)*, ,0.001 2.34 (1.872.92)*, ,0.001

PA 2.16 (1.582.96), ,0.001 2.02 (1.412.88)*, ,0.001 2.07 (1.442.98)*, ,0.001

Other APH 1.59 (1.361.85), ,0.001 1.67 (1.401.99)*, ,0.001 1.74 (1.462.07)*, ,0.001

PPH 1.45 (1.321.58), ,0.001 1.19 (1.071.32)$, ,0.001 1.28 (1.161.42)$, ,0.001

Vaginal PPH 1.46 (1.291.65), ,0.001 1.18 (1.031.35)#, 0.02 1.24 (1.081.42)#, 0.002

Caesarean PPH 1.23 (1.071.40), 0.002 1.18 (1.011.38)#, 0.03 1.31 (1.121.54)#, ,0.001

APH, antepartum haemorrhage; PA, placental abruption; PP, placenta praevia; PPH, primary post-partum haemorrhage.*Model 1 adjusted for age, year of birth, marital status, parity, any miscarriages or terminations of pregnancy and vertex presentation. Model 2 adjusted in addition for tertiary and privatehospital status.$Model 1 adjusted for age, year of birth, marital status, parity, any miscarriages or terminations of pregnancy, vertex presentation, pre-eclampsia, APH, PP, PA, premature rupture ofmembranes, induced labour, Caesarean section. Model 2 adjusted in addition for tertiary and private hospital status.#Model 1 adjusted for age, year of birth, marital status, parity, any miscarriages or terminations of pregnancy, vertex presentation, pre-eclampsia, APH, PP, PA, premature rupture ofmembranes, induced labour. Model 2 adjusted in addition for tertiary and private hospital status.

Figure 1 The effect of maternal age on the frequency of obstetric haemorrhage: APH, PP, PA and PPH in Primiparous women in the IVF/ICSI andgeneral population groups (error bars indicate the 95% CI). APH, antepartum haemorrhage; PA, placental abruption; PP, placenta praevia; PPH,primary post-partum haemorrhage.

Increased obstetric haemorrhage after ART 269

by guest on February 24, 2014http://hum

rep.oxfordjournals.org/D

ownloaded from

................................................................. ................................................................ .................................................................

..........................................................................................................................................................................................................................................................

Table IV Factors in IVF/ICSI group related to APH, PP and PPH

Factors Total APH PP PPH

n (%) OR (95% CI) Ad OR (95% CI)* n (%) OR (95% CI) Ad OR (95% CI)* n (%) OR (95% CI) Ad OR (95% CI)$

Not female infertility 3274 195 (6.0) 1 1 67 (2.0) 1 1 332 (10.1) 1 1

Female infertility 3456 259 (7.5) 1.28 (1.061.55) 1.27 (1.051.55) 107 (3.1) 1.53 (1.122.08) 1.55 (1.132.12) 414 (12.0) 1.21 (1.041.40) 1.14 (0.971.33)

Not endometriosis 5465 356 (6.5) 1 1 126 (2.3) 1 1 581 (10.6) 1 1

Endometriosis 1265 98 (7.7) 1.20 (0.951.52) 1.21 (0.951.53) 48 (3.8) 1.67 (1.192.34) 1.65 (1.182.32) 165 (13.0) 1.26 (1.051.52) 1.28 (1.061.56)

Not ovulatory 5742 380 (7.7) 1 1 145 (2.5) 1 1 610 (10.6) 1 1

Ovulatory disorder 988 74 (6.5) 1.14 (0.881.48) 1.14 (0.871.48) 29 (2.9) 1.17 (0.781.75) 1.14 (0.751.74) 136 (13.8) 1.34 (1.101.64) 1.27 (1.031.57)

Not tubal 5076 320 (6.3) 1 1 122 (2.4) 1 1 564 (11.1) 1 1

Tubal disease 1654 134 (8.1) 1.31 (1.0621.62) 1.34 (1.0821.67) 52 (3.1) 1.32 (0.951.83) 1.41 (1.001.98) 182 (11.0) 0.99 (0.831.18) 0.94 (0.781.13)

IVF 3340 232 (6.9) 1 1 95 (2.8) 1 1 376 (11.3) 1 1

ICSI 3390 222 (6.5) 0.94 (0.781.14) 0.89 (0.721.09) 79 (2.3) 0.82 (0.601.10) 0.72 (0.521.00) 370 (10.9) 0.96 (0.831.12) 0.94 (0.801.12)

Fresh transfer 4227 321 (7.6) 1.46 (1.1921.80) 1.46 (1.181.80) 123 (2.9) 1.44 (1.042.00) 1.40(1.001.95) 432 (10.2) 0.79 (0.680.92) 0.81(0.690.95)

FET 2503 133 (5.3) 1 1 51 (2.0) 1 1 314 (12.5) 1 1

Number transferred 1 1024 73 (7.1) 1.11 (0.861.45) 1.06 (0.811.38) 29 (2.8) 1.16 (0.771.75) 1.08 (0.711.65) 118 (11.5) 1.02 (0.831.27) 1.00 (0.801.25)

Number transferred 2 4807 310 (6.4) 1 1 118 (2.4) 1 1 542 (11.3) 1 1

Number transferred .2 899 71 (7.9) 1.24 (0.951.63) 1.37 (1.031.80) 27 (3.0) 1.23 (0.801.88) 1.42 (0.912.22) 86 (9.6) 0.83 (0.661.06) 0.88 (0.691.14)

Average cell number ,4 887 66 (7.4) 1.12 (0.841.50) 1.16 (0.861.56) 26 (2.9) 1.23 (0.781.95) 1.30 (0.822.08) 94 (10.6) 0.95 (0.741.21) 0.94 (0.731.21)

Average cell number 4 2804 188 (6.7) 1 1 67 (2.4) 1 1 312 (11.1) 1 1

Average cell number .4 2050 139 (6.8) 1.01 (0.811.27) 1.04 (0.821.30) 61 (3.0) 1.25 (0.881.78) 1.26 (0.881.79) 232 (11.3) 1.02(0.851.22) 1.07(0.891.29)

Unknown 989 61 (6.2) 0.91 (0.681.23) 1.01 (0.711.43) 20 (2.0) 0.84 (0.511.39) 1.02 (0.571.82) 108 (10.9) 0.98(0.781.23) 1.01(0.761.34)

Luteal phase hCG 1345 104 (7.7) 1.28 (1.011.62) 1.31 (1.031.67) 41 (3.0) 1.20 (0.831.72) 1.26 (0.871.82) 115 (8.6) 0.70 (0.570.86) 0.72 (0.580.90)

Progesterone 1200 93 (7.8) 1.28 (1.001.64) 1.26 (0.981.61) 26 (2.2) 0.84 (0.551.30) 0.82 (0.531.27) 138 (11.5) 1 1

None 4185 257 (6.1) 1 1 107 (2.6) 1 1 493 (11.8) 0.97(0.801.19) 0.93(0.761.15)

Fetal heart 1 6456 439 (6.8) 1 1 169 (2.6) 1 1 719 (11.1) 1 1

Fetal heart .1 274 15 (5.5) 0.79 (0.471.35) 0.79 (0.461.34) 5 (1.8) 0.69 (0.281.68) 0.69 (0.281.70) 27 (9.8) 0.87 (0.581.31) 0.86 (0.571.30)

APH, antepartum haemorrhage; PA, placental abruption; PP, placenta praevia.*Adjusted for age, year of birth, country of birth Asia and Middle East, marital status, parity, any miscarriages or terminations of pregnancy, vertex presentation, tertiary and private hospital status (Model 2).$Adjusted for age, year of birth, country of birth Asia and Middle East, marital status, parity, any miscarriages or terminations of pregnancy, vertex presentation, tertiary and private hospital status, pre-eclampsia, APH, PP, PA, premature rupture ofmembranes, induced labour, Caesarean section (Model 2).

270Healy

etal.

by guest on February 24, 2014 http://humrep.oxfordjournals.org/ Downloaded from

oocytes collected (,6 oocytes 51/897, 5.7%; 69 oocytes 93/1220,7.6%; 1013 oocytes 77/947, 8.1%; .13 oocytes 100/1163, 8.6%,P 0.009).

Infertility diagnosis: endometriosis and ovulatory disordersPP and post-partum haemorrhage were increased with endometriosis.Adding the other IVF/ICSI factors to the models did not change theOR greatly: PP 1.7 (1.22.4) and post-partum haemorrhage 1.3(1.11.6). Post-partum haemorrhage was also moderately stronglyassociated with ovulatory disorders and this continued after addingthe other IVF/ICSI factors to the model [OR 1.3 (1.11.6)].

Hormone treatmentPost-partum haemorrhage was less frequent when hCG was used forluteal phase support. Adding the other IVF/ICSI covariates did notchange the OR greatly [0.7 (0.60.9)].

Analysing only the frozen embryo transfer results showed strongevidence (P, 0.001) for an increase in post-partum haemorrhagewith FET in articial cycles compared with natural cycles [OR 2.2(95% CI 1.62.9), Adjusted OR 1.8 (1.32.6)]. There were 93 post-partum haemorrhages following 458 FET in articial cycles (20.3%) and221 post-partum haemorrhages in 2045 FET in natural cycles (10.8%).

Factors within the IVF/ICSI group without strong evidence of anassociation with the haemorrhagic complicationsOther factors analysed, including number of previous oocyte collec-tions or ETs, types of ovarian stimulation and drugs used, source ofsperm, overnight culture of embryos, maternal height, smoking andobesity (for which data were incomplete), showed no evidence ofassociation with APH, PP or post-partum haemorrhage (P . 0.1).

DiscussionThis large, multi-centre study shows increased rates of obstetric haem-orrhage in mothers of singleton ART babies. Our results add to infor-mation in national and international studies (Dickey, 2007; Halliday,2007; Anderson et al., 2008). It has been hitherto uncertainwhether the increases in these complications are caused by factorsintrinsic to the ART procedures or factors related to the character-istics of the ART-seeking population. We have interrogated our data

to address this question and examined possible mechanisms forthese adverse outcomes.

The GIFT women were different because this group was small innumber, younger, treated earlier and more often married than theIVF/ICSI group. They also had less Caesarean sections probablyrelated to the increasing rate of Caesarean section over the time ofthe study. The GIFT method of treatment has become less popularover time with improvements in the results of IVF. Although theyhad less post-partum haemorrhage than the IVF/ICSI group, theyhad the same or higher rates of APH, PP and PA (Table II). The sig-nicance of this is that in contrast to IVF and ICSI, embryos are notformed in vitro with GIFT. The oocytes and sperm are introducedinto the outer end of the uterine tube and fertilization and earlyembryo cleavage occurs in vivo. GIFT was also usually used incouples with unexplained infertility. The fact that APH is not less fre-quent with GIFT than with IVF and ICSI indicates that it is not the fer-tilization and embryo manipulation within the laboratory that isresponsible for the increased frequency of APH in the IVF/ICSIgroup. Furthermore, the adverse APH results with GIFT are also evi-dence that it is not the trans-cervical embryo transfer with low place-ment of embryos into the uterus that increases the risk of APH andparticularly PP, as previously suggested (Smithers et al., 2003;Romundstad et al. 2006). In contrast, as IVF/ICSI and GIFT areboth undertaken during a stimulated cycle, factors common to bothprocedures such as ovarian stimulation, or anaesthesia and surgeryfor oocyte collection may be implicated in the adverse outcomes.

The non-ART group had fewer of the haemorrhages than the IVF/ICSI group. From these data there is inconclusive evidence ofincreased rates of obstetric haemorrhages related to infertility;however, it is likely this group of patients had less severe infertilitythan did those in the IVF/ICSI group. It is possible but unlikely thatsome of these patients may have conceived by ART interstate or over-seas. As this would bias this groups results towards more haemor-rhages, we conclude there appears to be adverse effects of IVF/ICSIincreasing the haemorrhagic complications separate from theinfertility.

Having conrmed higher frequencies of APH and post-partumhaemorrhage with ART, we were interested to determine if the differ-ences in characteristics between subjects in the IVF/ICSI and generalpopulation groups would account for some or all of the effect.

.............................................................................................................................................................................................

Table V Obstetric complications in IVF/ICSI patients with fresh embryo transfer (embryo transfer) in a stimulated cycleor FET in a natural cycle [number (%), unadjusted and adjusted odds ratios and 95% CI with FET as reference]

Group APH PP PA PPH

Stimulated cycle: fresh embryo transfer(N 4058)

314 (7.7) 122 (3.0) 46 (1.1) 417 (10.3)

Unstimulated cycle: FET (N 2045) 105 (5.1) 44 (2.2) 9 (0.4) 221 (10.8)OR (95% CI), P-value 1.55 (1.231.94), ,0.001 1.41 (1.001.99), 0.05 2.59 (1.295.18), 0.007 0.94 (0.801.12), 0.52

Ad OR (95% CI), P-value 1.53 (1.221.93)*, ,0.001 1.37 (0.961.95)*, 0.08 2.56 (1.255.27)* 0.01 0.94 (0.781.22)$, 0.49

APH, antepartum haemorrhage; PA, placental abruption; PP, placenta praevia; PPH, primary post-partum haemorrhage, FET, frozen thawed embryo transfer.*Adjusted for age, year of birth, country of birth Asia and Middle East, marital status, parity, any miscarriages or terminations of pregnancy, vertex presentation, tertiary and private hospitalstatus (Model 2).$Adjusted for age, year of birth, country of birth Asia and Middle East, marital status, parity, any miscarriages or terminations of pregnancy, vertex presentation, tertiary and private hospitalstatus, pre-eclampsia, APH, PP, PA, premature rupture of membranes, induced labour, Caesarean section (Model 2).

Increased obstetric haemorrhage after ART 271

by guest on February 24, 2014http://hum

rep.oxfordjournals.org/D

ownloaded from

Although we matched the general population group and the IVF/ICSIsubjects by age of mother and date of birth, this left some importantdifferences between the two groups particularly in gravidity and parity.Also the IVF/ICSI group delivered by Caesarean section more often.Adjusting for confounding left the ORs for APH either unchanged orincreased (Table III). In contrast, for post-partum haemorrhage, theORs decreased with adjustments for the other factors particularlyparity, Caesarean section and other maternal complications(Table III). Thus some of the higher frequency of post-partum haem-orrhage after IVF/ICSI is explained by differences in the characteristicsof women having these treatments from those in the general commu-nity, particularly their lower parity and higher use of Caesareansection. However, there remained some evidence of a small excessof post-partum haemorrhage related to some other aspect of IVF/ICSI.

Having conrmed an increase in the obstetric haemorrhages specicto IVF/ICSI, we then analysed the IVF/ICSI group on its own toexplore which factors were related to the higher frequencies of thehaemorrhages. We found that APH, PP and PA were higher withfresh embryo transfer than with FET, and within the fresh embryotransfer treatments, APH increased with the numbers of oocytes col-lected. PP and post-partum haemorrhage were increased in subjectswith endometriosis. Post-partum haemorrhage was also more fre-quent in women with ovulatory disorders, but less frequent whenhCG had been used for luteal phase support. Post-partum haemor-rhage was more frequent with FET treatments in articial cyclesthan with FET in natural cycles. It is possible that some of thesenew ndings may be statistically signicant by chance alone and theyneed conrmation by analysis of separate datasets. Also it is difcultto know if these signicant associations reect direct mechanisms orresult from a peripheral relationship. For example the luteal phasehCG and FET in articial cycle effects maybe more related to thecharacteristics of the patients having these treatments than to thetreatments themselves. However, if these ndings are true theremay be important implications and some speculation on the possiblemechanisms, is useful for further studies.

Therewas a greater risk of APH after fresh embryo transfer in a stimu-lated cycle than FET in a natural cycle (Tables IV and V). This increase inAPH in general with fresh embryo transfer in stimulated cycles is particu-larly interesting. Differences in outcome of singleton births after freshembryo transfer and FET have been noted for birthweight, pretermbirth, perinatal death and some birth defects (Sutcliffe et al., 1995;Wen-nerholm et al., 1998; Aytoz et al., 1999; Kallen et al., 2005a; Olson et al.,2005; Belva et al., 2008; Shih et al., 2008). These differences are intri-guing, as they must result from either embryonic effects, with cryopre-servation improving the outcomes by exerting some selective effect forbetter embryos, or adverse effects occurring in the mother in freshembryo transfer cycles that are not present with FET. The latter maybe related to ovarian stimulation and oocyte collection.

Anaesthesia or monitored conscious sedation for oocyte collectionseems unlikely to explain the differences observed between stimulatedand natural ART cycles. The initial half-life of anaesthetic medicinessuch as propofol is only 24 min: embryo transfer is typically 25days after anaesthesia. Moreover, anaesthetic agents are not thoughtto act on the endometrium.

The increase in APH following fresh embryo transfer with the numberof oocytes is a particularly important indicator that ovarian stimulation is

involved. Oocyte number is related to the dose of FSH used and theovarian responsiveness of the woman. Estradiol and progesteronelevels at the time of implantation in fresh embryo transfer cycleswould be expected to be associated with oocyte number. It may bethat excessive ovarian hormone levels with multiple follicular develop-ments interfere with the endometrium during implantation and increasethe risk of APH and PA. We have recently shown that the endometrialprotein, pregnancy associated plasma protein A (PAPP-A) is low in preg-nancies after fresh embryo transfer in stimulated cycles (Amor et al.,2009). PAPP-A is a zinc binding matrix metalloproteinase. PAPP-Ahelps regulate extracellular matrix remodelling. This is vital in angiogen-esis and the establishment of placentation during the early weeks ofpregnancy. Our nding of increased APH after fresh embryo transferin stimulated ART cycles suggest that the high estradiol and progester-one concentrations, as a result of multiple folliculogenesis, produce sub-optimal angiogenesis and placentation.

The diagnosis of endometriosis was associated with more PP andpost-partum haemorrhage. Although PP was also associated withhigher PPH, there was evidence that the effect of endometriosisincreased the risk of both complications. We recently reportedother potential obstetric consequences of endometriosis for ART:ovarian endometriosis is associated with approximately doubledrates of preterm birth and small for gestational age babies (Fernandoet al., 2009).

The less frequent post-partum haemorrhage with luteal phase hCGalso suggests a disturbance of endometrial function around the time ofimplantation is involved in increasing PPH. In addition it suggests thatimproved management of luteal phase support in ART patients mayhave a benecial effect. With FET there was more PPHs with the arti-cial cycles. This may be related directly to the hormone treatmentbeing less effective in preparing the endometrium for implantationcompared with natural cycles or, alternatively, the ovulatory disorderin the patients requiring articial cycles may be responsible. Ovulatorydisorders were also associated with increased post-partumhaemorrhage.

The strengths of this study are the large study population anddetailed information available on the treatments from the three infer-tility services and birth outcomes from the State PCDU database. Theresults of IVF/ICSI patients were compared with those of populationcontrols, GIFT and non-ART conceptions in infertile women. Indepen-dent midwives, not ART clinic staff, recorded all obstetric data and thisshould minimize ascertainment bias. During the study period relativelylow numbers of embryos were transferred, the multiple pregnancyrate was less that 20% and there were few vanishing twins. The analy-sis of singleton births only and the high usage of FET is different frommost other data for this time with higher multiple pregnancy rates. Wecould therefore study the inuence of ovarian stimulation by compar-ing fresh embryo transfer with FET in a natural or non-stimulatedcycle. We also assessed the effects of different methods of stimulationand a marker of the effect of stimulation: the number of oocytes col-lected before a fresh embryo transfer. To avoid the problem of analy-sis of subsequent pregnancies in the same woman and the high withinsubject correlations with repeated complications in subsequent preg-nancies, for each woman only the rst birth within the study periodwas analysed.

There are a number of limitations. The quality of the data is variable.For example, there were inevitable time-related changes over the

272 Healy et al.

by guest on February 24, 2014http://hum

rep.oxfordjournals.org/D

ownloaded from

15 years of data collection, with differences in approach to diagnosisand procedures with the evolution of ART. There were not manywomen with PA. There was no information on covariates such asmaternal weight or smoking during pregnancy in the general popu-lation group and this was also incomplete in the IVF/ICSI group.Another problem with these data is that we have no information ofprevious pregnancy complications particularly previous Caesareansection, which increases the risk of PP in subsequent pregnancies.We were not able to analyse these factors. Nevertheless, as 78% ofthe IVF/ICSI patients were primiparous (Table I), we might haveexpected to see a higher rate of past Caesarean sections in thegeneral population group thereby producing more PP cases: weobserved the opposite. Thus it is not possible to treat the results asif they were from a clinical trial with random placement of couplesin separate treatment groups or from a hypothesis based study, andconclusions on cause and effect need to be guarded.

In conclusion, all the obstetric haemorrhages are more frequentafter IVF/ICSI than in the general population. For post-partum haem-orrhage, some of the increase is related to the IVF/ICSI patientshaving fewer previous pregnancies and using Caesarean sectionmore. As these complications are less frequent in non-ART con-ceptions in infertile women, the mechanism of the increase mustinvolve some aspects of the IVF/ICSI treatment. As APH is not lessfrequent with GIFT, embryology procedures and embryo placementthrough the cervix do not appear to be important. Exploratory analysisof factors in the IVF/ICSI group suggests events around the time ofimplantation are responsible for the increase in subsequent obstetrichaemorrhages. The increased APH with fresh embryo transfer andwith greater numbers of oocytes collected, and the protective effectof hCG luteal phase support and the adverse effect of ovulatory dis-orders and articial cycles for FET on PPH all indicate that suboptimalendometrial function around the time of implantation is critical.

Authors RolesAll authors have contributed to the conducting of this study. Themanuscript has been seen and approved by all authors. D.L.H.,H.W.G.B. and J.H. were the senior investigators involved in studydesign, method investigations, data analysis and preparation of themanuscript. S.B., A.J., D.R. and C.G. were involved in method inves-tigation, data checking and analysis and review of the manuscript.J.M.T. was involved in data provision.

AcknowledgementsThe authors thank Obi Ukoumunne, Odette Taylor, Gillian Wheatleyand Associate Professor James King at the Perinatal Data CollectionUnit for assistance. We acknowledge the nancial support from theBUPA Foundation and the Fertility Society of Australia.

FundingJ.H. Senior Research Fellowship is funded by the National Health &Medical Research Council (436904).

ReferencesAmor DJ, Xu JX, Halliday J, Francis L, Healy DL, Breheny S, Baker HW,Jaques A. Pregnancies conceived using assisted reproductivetechnologies (ART) have low levels of pregnancy associated protein-A(Papp-A) leading to a high rate of false positive results in rsttrimester screening for Down syndrome. Hum Reprod 2009;24:13301338.

Anderson AN, Goossens V, Ferraretti AP, Bhattacharya S, Felberbaum R,de Mouzon J, Nygren KG, (ESHRE) EI-mECESoHRaE. Assistedreproductive technology in Europe, 2004: results generated fromEuropean registers by ESHRE. Hum Reprod 2008;23:756771.

Aytoz A, Van den Abbeel E, Bonduelle M, Camus M, Joris H, VanSteirteghem A, Devroey P. Obstetric outcome of pregnancies afterthe transfer of cryopreserved and fresh embryos obtained byconventional in-vitro fertilization and intracytoplasmic sperm injection.Hum Reprod 1999;14:26192624.

Belva F, Henriet S, Van den Abbeel E, Camus M, Devroey P, Van der Elst J,Liebaers I, Haentjens P, Bonduelle M. Neonatal outcome of 937 childrenborn after transfer of cryopreserved embryos obtained by ICSI and IVFand comparison with outcome data of fresh ICSI and IVF cycles. 2008;23:22272238.

Daniel Y, Ochshorn Y, Fait G, Geva E, Bar-Am A, Lessing JB. Analysis of104 twin pregnancies conceived with assisted reproductive technologiesand 193 spontaneously conceived twin pregnancies. Fertil Steril 2000;74:683689.

Dickey R. The relative contribution of assisted reproductive technologiesand ovulation induction to multiple births in the United States 5 yearsafter the Society for Assisted Reproductive Technology/AmericanSociety for Reproductive Medicine recommendation to limit thenumber of embryos transferred. Fertil Steril 2007;88:15541561.

Fernando S, Breheny S, Jaques AM, Halliday JL, Baker G, Healy D. Pretermbirth, ovarian endometriomata, and assisted reproduction technologies.2009;91:325330.

Halliday J. Outcomes of IVF conceptionsare they different? Best Pract ResClin Obstet Gynaecol 2007;21:6781.

Jackson RA, Gibson KA, Wu YW, Croughan MS. Perinatal outcomes insingletons following in vitro fertilization: a meta-analysis. ObstetGynecol 2004;103:551563.

Kallen B, Finnstrom O, Nygren KG, Olausson PO. In vitro fertilization(IVF) in Sweden: risk for congenital malformations after different IVFmethods. Birth Defects Res A Clin Mol Teratol 2005a;73:162169.

Kallen B, Finnstrom O, Nygren KG, Olausson PO, Wennerholm U. In vitrofertilization (IVF) in Sweden: obstetric characteristics, maternalmorbidity & mortality. BJOG 2005b;12:15291535.

McCullagh P, Nelder JA. Generalized Linear Models, 2nd edn. London:Chapman and Hall, 1989.

Olson CK, Keppler-Noreuil KM, Romitti PA, Budelier WT, Ryan G,Sparks AE, Van Voorhis BJ. In vitro fertilization is associated with anincrease in major birth defects. Fertil Steril 2005;84:13081315.

Perri T, Chen R, Yoeli R, Merlob P, Orvieto R, Shalev Y, Ben-Rafael Z,Bar-Hava I. Are singleton assisted reproductive technologypregnancies at risk of prematurity? J Assist Reprod Genet 2001;18:245249.

Riley M, Grifn O. Validating a statewide data collection: differences ininformation technology resources between hospitals. Health Inf Manag1997;27:6768.

Romundstad LB, Romundstad PR, Sunde A, von During V, Skjaerven R,Vatten LJ. Increased risk of placenta previa in pregnancies followingIVF/ICSI; a comparison of ART and non-ART pregnancies in thesame mother. Hum Reprod 2006;21:23532358.

Increased obstetric haemorrhage after ART 273

by guest on February 24, 2014http://hum

rep.oxfordjournals.org/D

ownloaded from

Schieve LA, Cohen B, Nannini A, Ferre C, Reynolds MA, Zhang Z, Jeng G,Macaluso M, Wright VC, Massachusetts Consortium for AssistedReproductive Technology Epidemiologic Research (MCARTER). Apopulation-based study of maternal and perinatal outcomes associatedwith assisted reproductive technology in Massachusetts. Matern ChildHealth J 2007;11:517525.

Shevell T, Malone FD, Vidaver J, Porter TF, Luthy DA, Comstock CH,Hankins GD, Eddleman K, Dolan S, Dugoff L et al. Assistedreproductive technology and pregnancy outcome. Obstet Gynecol2005;106:10391045.

Shih W, Rushford DD, Bourne H, Garrett C, McBain JC, Healy DL,Baker HW. Factors affecting low birthweight after assistedreproduction technology: difference between transfer of fresh andcryopreserved embryos suggests an adverse effect of oocytecollection. Hum Reprod 2008;23:16441653.

Smithers PR, Halliday J, Hale L, Talbot JM, Breheny S, Healy DL. HighFrequency of Cesarean section, antepartum hemorrhage, placenta

previa, and preterm delivery in in vitro fertilization twin pregnancies.Fertil Steril 2003;80:666668.

Sutcliffe AG, DSouza SW, Cadman J, Richards B, McKinlay IA,Lieberman B. Minor congenital anomalies, major congenitalmalformations and development in children conceived fromcryopreserved embryos. Hum Reprod 1995;10:33323337.

Tayor R. Automatic Record Linkage. Sydney: The McGraw-Hill CompaniesInc., 1998, 184187.

Wennerholm UB, Albertsson-Wikland K, Bergh C, Hamberger L,Niklasson A, Nilsson L, Thiringer K, Wennergren M, Wikland M,Borres MP. Post-natal growth and health in children born aftercryopreservation as embryos. Lancet 1998;351:10851090.

World Health Organisation. WHO Recommendations for the Prevention ofPostpartum Haemorrhage. Geneva, Switzerland: WHO Press, 2007.

Submitted on May 20, 2009; resubmitted on September 10, 2009; accepted onOctober 1, 2009

274 Healy et al.

by guest on February 24, 2014http://hum

rep.oxfordjournals.org/D

ownloaded from