SHORT COMMUNICATION: Adverse maternal outcomes in multifetal pregnancies


Dr M. Walker, Department of Obstetrics and Gynecology, Omni Research Group, Ottawa General Hospital, General Campus, 501 Symth Road, Box 241, Ottawa, Ontario, Canada K1H 8L6.


In this retrospective cohort of 165,188 singleton pregnancies and 44,674 multiple-fetal pregnancies in Canada from 1984 to 2000, we compared the incidence of maternal complications. Multiple gestation pregnancies were associated with significant increases in cardiac morbidity, haematologic morbidity, amniotic fluid embolus, pre-eclampsia, gestational diabetes, postpartum haemorrhage, prolonged hospital stay, the need for obstetric intervention, hysterectomy and blood transfusion. Multiple gestation pregnancies are associated with an increased risk of morbidity for the mother. This should be taken into consideration in antenatal care of these women.


The incidence of twinning and higher order multiples has increased dramatically over the last two decades.1 The greatest contributor to this explosion in multiple gestations has been delayed fertility and the use of assisted reproductive technology.1

It is well known that multiple gestation is related to increased risk of perinatal morbidity and mortality and consequently long term health sequelae.1 The impact of multiple gestation on women's health is less known. A few previous studies of the issue suffer from various methodological limitations, including small sample size, inadequate control group and inappropriate outcome measures.2,3

Pregnancy is associated with a number of physiologic changes for adaptation and accommodation of the fetus.2–4 There is a 50% increase in plasma volume and a corresponding increment in cardiac output. This allows for increased blood volume for uterine perfusion for delivery of nutrient and oxygen exchange. There are changes in coagulation factors with an increase in procoagulants and a decrease in anticoagulants.5 This hypercoaguable state has a protective effect to the mother during delivery by decreasing haemorrhage. These physiologic changes, although having inherent advantages for mother and fetus, can lead to adverse maternal outcomes such as venous thromboembolism.

Multiple gestation, having increased placental and fetal mass, is likely to have increased physiologic responses compared with singleton pregnancies.2–4 For instance, stroke volume and heart rate are both increased in multiple gestation suggesting that there is a decrease cardiac reserve. These exaggerated physiologic changes in multiple pregnancies are potentially associated with an increase in adverse maternal outcomes such as pulmonary oedema, venous thromboembolic disease and even death. The objective of this article is to determine whether there is an increase in adverse maternal outcomes in multiple gestation pregnancies.


We extracted all records for obstetric deliveries in Canada (except for Quebec) from 1984 to 2000 contained in the Discharge Abstract Database collected by the Canadian Institute of Health Information. During the study period, Canadian Institute of Health Information coded diagnoses according to the International Classification of Diseases 9th Revision (ICD-9)6 and coded procedures according to the Canadian Classification of Diagnostic, Therapeutic, and Surgical Procedures.7 About 70% of all obstetric deliveries in Canada were recorded by the Canadian Institute of Health Information. Most of the obstetric deliveries in the province of Quebec and parts of Manitoba and Nova Scotia were not included in the Canadian Institute of Health Information. In a previous study, we demonstrated that obstetric deliveries recorded by the Canadian Institute of Health Information form a reasonably representative sample of all births in Canada.8

We selected all pregnant women with twin or higher order multiples as the exposed subjects. For each woman with a multiple gestation, we selected the next four available women with a single gestation (non-exposed) matched by institution of hospitalisation for delivery (same; individual match), year of discharge (within two years except fiscal 1984/1985; individual match), age (within five years; frequency match) and postal codes (first three digits; frequency match). Relative risks and 95% confidence intervals were estimated as the effect measures for multiple gestation, using women with singleton pregnancy as the reference.


During the 16 years of study period, a total of 4,395,968 obstetric deliveries were recorded in the Canadian Institute of Health Information's Discharge Abstract Database, of which 44,674 were multiple gestation. These were matched to 165,188 singleton deliveries according to the aforementioned criteria. The matching was highly successful, with mean ages quite similar between the two groups (Table 1).

Table 1.  Characteristics of study subjects.
Fiscal yearNo. of deliveriesNo. of sampled hospitalsNo. of multiplesMean age of mothers with multiplesNo. of singletonsMean age of mothers with singletons
1984190,903147159827.4 (4.8)605627.3 (4.6)
1985/1986407,456170360128.0 (4.8)13,34427.9 (4.7)
1987/1988474,324201421528.4 (4.8)15,72628.3 (4.7)
1989/1990578,636210538428.6 (4.9)20,16928.5 (4.8)
1991/1992592,013208575428.9 (4.9)21,29428.8 (4.8)
1993/1994563,135196573429.1 (5.1)21,21929.0 (5.0)
1995/1996556,135194605729.6 (5.2)22,32029.5 (5.1)
1997/1998524,562194610030.1 (5.3)22,40429.9 (5.2)
1999/2000508,804179623130.1 (5.5)22,65630.0 (5.3)

There were significant increases in maternal cardiovascular morbidity, haematologic morbidity, amniotic fluid embolus, pre-eclampsia, gestational diabetes, postpartum haemorrhage, prolonged hospital stay, the need for obstetric intervention, hysterectomy and blood transfusion (Table 2). There was also a higher incidence of maternal death in hospital and stroke. Vaginal tearing and uterine rupture were less common in multiples (Table 2).

Table 2.  Comparison of maternal outcomes between mothers with a multifetal pregnancy (n= 44,674) and mothers with singleton pregnancy (n= 165,188). Values are presented as n (%).
OutcomeMultiplesSingletonsRelative risk (95% CI)
In-hospital death5 (0.011)9 (0.005)2.05 (0.69–6.13)
Pre-eclampsia4407 (9.87)5872 (3.56)2.78 (2.67–2.88)
Gestational diabetes1633 (3.66)5387 (3.26)1.12 (1.06–1.18)
Myocardial infarction37 (0.083)37 (0.022)3.70 (2.34–5.83)
Heart failure (left ventricular)7 (0.016)2 (0.001)12.94 (2.69–62.30)
Stroke1 (0.002)1 (0.001)3.70 (0.23–59.12)
Venous thromboembolism94 (0.21)131 (0.08)2.65 (2.04–3.46)
Pulmonary oedema54 (0.12)28 (0.017)7.13 (4.52–11.25)
Amniotic fluid embolus5 (0.011)14 (0.008)1.32 (0.48–3.67)
Uterine rupture21 (0.047)153 (0.093)0.51 (0.32–0.80)
Vaginal tearing140 (0.31)1108 (0.67)0.47 (0.39–0.56)
Postpartum haemorrhage3903 (8.74)7680 (4.65)1.88 (1.81–1.95)
Operative vaginal delivery13,120 (29.37)26,298 (15.92)1.84 (1.81–1.88)
Caesarean delivery20,106 (45.006)34,383 (20.81)2.16 (2.13–2.19)
Hysterectomy60 (0.13)97 (0.059)2.29 (1.66–3.16)
Blood transfusion37 (0.083)82 (0.050)1.67 (1.13–2.46)
>4 days in hospital27,279 (61.062)41,242 (24.97)2.45 (2.42–2.47)


We found increases in common obstetric complication such as pre-eclampsia and gestational diabetes as well as rarer, but potentially more serious, morbidity such as pulmonary oedema, venous thromboembolic disease and myocardial infarction. Pre-eclampsia, thromboembolic disease and postpartum haemorrhage, which had relative risks of 2.7 or greater in multiples, are three of the leading causes of direct maternal death in industrial countries.9 Maternal in-hospital death was increased in multiple gestations but because of small numbers the confidence interval crossed one.

Uterine rupture and vaginal tearing were less common in multiples. These findings are expected and give internal validity to the results of this study. Many of these pregnancies, due to either malpresentation or higher order multiples, would have elective caesarean sections thereby reducing these outcomes.3

Surgical interventions, either caesarean section or operative vaginal delivery, occurred in 75% of multiple pregnancies. These women were more likely to require hysterectomy or receive a blood transfusion. This high degree of operative intervention should be incorporated into risk management strategies and patients should be counselled appropriately.

Limitations in our data should not be overlooked. As in any observational study, imbalance between study groups is an inherent problem. To reduce the risk of imbalance between the comparison groups, we used a tight control strategy by matching for institution of hospitalisation for delivery, year of discharge, age and postal codes. As it is apparent from Table 1, the matching results were quite satisfactory. Matching by these factors could have removed a large extent of the confounding by comorbidity, referral bias and socio-economic status. However, the matching may be insufficient, and residual confounding may still have occurred. Our study used administrative data, which lack clinical details. For example, no information on infertility treatment is available in our data. As a result, we could not separate the effect of infertility or infertility treatment from the effect of multiple pregnancy. The last two decades, through assisted reproductive technologies, have seen an explosion in the number of multiple gestation pregnancies.1 However, assisted reproductive technologies cause increases mostly of triplets or higher order of multiples,1 while in our data, most of the multiples were twins, the impact of assisted reproductive technologies should be minimum.

The administrative data are prone to a certain degree of coding errors.10 However, coding errors are likely to have occurred in a random fashion, which would tend to attenuate the observed effects.11 This is especially true in our study as coding was done by in-house clerks who transcribe chart information into ICD-9 codes. They were completely blinded to the hypothesis that multifetal pregnancies would have more complications. It is unlikely that there would be any diagnostic suspicion bias. Moreover, we have conducted a preliminary assessment of the validity of the ICD-9 codes, using a linked mother and infant records acquired from Med-Echo (a similar hospital discharge database from the Canadian province of Quebec). Using preterm birth defined by gestational age <37 completed weeks as the ‘gold standard’, we found a concordance of 93% for the ICD-9 code-based diagnosis of preterm labour (data available upon request). Although this result may not necessarily be applicable to other diagnostic codes or other provinces, such high agreement suggests an acceptable quality of coding.

Observational studies remain the only source of current information for the adverse outcomes in mothers with multiple pregnancy. Our study compares favourably with previous studies in terms of sample size and population coverage. The increased medical morbidity, obstetric complications and surgical intervention have important implications to affected mothers as well as to the health care system in general. These factors must be taken into consideration in antenatal care of women with multiple gestation pregnancy. In conclusion, multiple gestation pregnancies are associated with an increased risk of morbidity for the mother.

Accepted 7 May 2004