Maternal obesity and excess of fetal growth in pre-eclampsia

Authors

  • S Rasmussen,

    Corresponding author
    1. Department of Clinical Science, University of Bergen, Bergen, Norway
    2. Department of Obstetrics and Gynaecology, Haukeland University Hospital, Bergen, Norway
    • Correspondence: S Rasmussen, Department of Obstetrics and Gynaecology, Haukeland University Hospital, N-5021 Bergen, Norway.

      Email Svein.Rasmussen@mfr.uib.no

    Search for more papers by this author
  • LM Irgens,

    1. Department of Global Public Health and Primary Care, University of Bergen, Bergen, Norway
    Search for more papers by this author
  • J Espinoza

    1. Department of Obstetrics and Gynecology, Baylor College of Medicine and Texas Children's Hospital Pavilion for Women, Houston, TX, USA
    Search for more papers by this author

  • Linked article: This article is commented on by Lisonkova S and Joseph KS, pp. 1358 in this issue.

Abstract

Objective

To assess whether the reported excess of large for gestational age (LGA) neonates in pre-eclamptic women delivering at term is attributable to maternal obesity.

Design, setting and population

Population-based observational study including 77 294 singleton pregnancies registered in the Medical Birth Registry of Norway between 2007 and 2010.

Methods

Comparison of birthweight percentiles and z-scores between women with and without pre-eclampsia.

Main outcome measures

Odds ratio (OR) of LGA and z-scores of birthweight in relation to pre-eclampsia.

Results

Pre-eclamptic women delivering at term had increased risk of having LGA neonates. Unadjusted ORs with 95% confidence interval (95% CI) of LGA above the 90th and 95th birthweight centiles were 1.4, 95% CI 1.2–1.6 and 1.6, 95% CI 1.3–1.9, respectively. The excess of LGA persisted after including gestational diabetes and diabetes types 1 and 2 in a multivariate analysis (corresponding ORs 1.3, 95% CI 1.1–1.5 and 1.4, 95% CI 1.2–1.7), but disappeared after adjusting for maternal prepregnant body mass index (ORs 1.1, 95% CI 0.9–1.2 and 1.1, 95% CI 0.9–1.3).

Conclusions

This study suggests accelerated fetal growth in a subset of pre-eclamptic women delivering at term. The excess of LGA neonates is attributable to maternal obesity among pre-eclamptic women delivering at term. The maternal obesity epidemic may lead to an increased prevalence of both pre-eclampsia and LGA neonates among women delivering at term.

Introduction

Impaired placentation seems to play an important role in the pathogenesis of pre-eclampsia.[1] Accumulating evidence indicates that products released by the ischaemic placenta including anti-angiogenic factors, may lead to endothelial dysfunction resulting in the clinical manifestation of pre-eclampsia.[2-4] In addition, it has been proposed that due to the limited perfusion of the placenta the fetuses are often growth restricted.[5-8] Moreover, fetal growth restriction may predate the diagnostic findings of pre-eclampsia in the same pregnancy, as demonstrated by longitudinal studies indicating that fetal growth restriction with limited placental perfusion is associated with a three-fold increased risk of pre-eclampsia.[9]

Large population based studies[10-12] have also reported excess of large-for-gestational-age (LGA) fetuses in pre-eclampsia and have challenged the hypothesis that placental dysfunction is the only subjacent mechanism of disease in pre-eclampsia. These observations have even raised the question whether pre-eclampsia has more than one aetiological entity.[13, 14] Recently, is has been proposed that a relative uteroplacental ischaemia due to a mismatch between limited uteroplacental blood flow and increased fetal demand for nutrients,[15, 16] may be involved in the pathogenesis of late-onset pre-eclampsia.[17, 18] It is possible that the increasing prevalence of maternal obesity may lead to excessive fetal size and increased prevalence of pre-eclampsia in pregnant women at term.

The aim of this population-based study was to assess whether the reported excess of LGA in late onset pre-eclampsia is attributable to obesity.

Methods

During 2007–10 a total of 248 056 births were notified to the Medical Birth Registry of Norway. During this study period there were 77 294 singleton births with gestational age 22–44 weeks and data on maternal body mass index (BMI), which were included (they represent 33% of all singleton births with gestational age 22–44 weeks, n = 235 572). Established in 1967, the Medical Birth Registry of Norway is based on compulsory notification of all liveborn neonates and stillbirths after 16 weeks of gestation, in 1999 extended to 12 weeks of gestation. Medical data are collected in a pregnancy record, which is kept by each woman and brought to the delivery unit at the time of delivery. Selected items of data are transferred by the midwives to the Medical Birth Registry notification form by simply checking boxes.[19] Transition to on-line electronic recording, representing an extract of the clinical record, began in 2006 and is now implemented in most delivery units in Norway.

The dependent variable was infant size in terms of gender-specific, birth order-specific (1 and 2+) and gestational age-specific birthweight empirical centiles (<5th, <10th, >90th, and >95th centiles) based on all 679 009 singleton births with gestational age 22–44 weeks notified to the Medical Birth Registry in 1999 to 2010. Additionally, gender-, birth order- and gestational age-specific z-scores of birthweight were calculated by polynomial multilevel regression because some women were included in the database more than once from different pregnancies. Gestational age was based on ultrasound scanning between 17 and 19 weeks of gestation. In pregnancies lacking such data (3%) gestational age was based on the last menstrual period.

Independent variables comprised maternal age (<20, 20–24, 25–29, 30–34, 35–39 and >39 years), marital status (married/cohabiting, other), smoking at the beginning of pregnancy (not, occasionally, daily, smoking pattern not specified), chronic maternal disease (asthma, chronic hypertension, chronic renal disease, rheumatoid arthritis, maternal heart disease, thyroid disease and diabetes mellitus types 1 and 2), pregnancy complications (pre-eclampsia, placental abruption, gestational diabetes), and maternal prepregnant BMI, categorised according to the definitions of the World Health Organization into <18.5 kg/m2 (underweight), 18.5–24.9 (normal), 25.0–29.9 (overweight) and ≥30.0 or more (obese).[20] From 2007 onwards data on maternal BMI were registered in 33% of all births notified to Birth Registry in Norway. To assess any bias in relative risk estimates of exposure–outcome associations we performed separate analyses in the subgroups with and without BMI.

In Norway, the diagnosis of pre-eclampsia is in accordance with the recommendations of the American College of Obstetricians and Gynecologists,[21] which defined pre-eclampsia as the presence of systolic blood pressure of ≥140 mmHg or a diastolic pressure of ≥90 mmHg on at least two occasions 6 hours apart after 20 weeks of gestation with proteinuria defined as excretion of ≥0.3 g/day, equivalent to at least 1+ on a urine reagent strip.

Severe pre-eclampsia is likely to result in a preterm birth. Therefore, we compared infants' size in the following groups: ≥37 weeks of gestation, late preterm (34–366/7 weeks) and early preterm (<34 weeks) births in mothers with and without pre-eclampsia. Associations between the infant size and pre-eclampsia were assessed by logistic multilevel regression analysis while adjusting for possible confounders (maternal age, marital status, chronic maternal disease, pregnancy complications, smoking habits and maternal BMI). As the birthweight centiles were birth-order-specific, we did not adjust for birth order. Possible co-linearity between type 2 diabetes or gestational diabetes and maternal BMI was assessed using a variance inflation factor in linear regressions with z-scores of birthweight as the outcome variable and BMI and type 2 diabetes or gestational diabetes as independent variables. Differences between groups in maternal characters and z-scores of birthweight were assessed by exact two-sided Pearson chi-square test and analysis of variance, respectively.

The statistical analysis was carried out with SPSS version 18.0.3 (Statistical Package for the Social Sciences; SPSS Inc, Chicago, IL, USA) and the MLwiN programme version 2.27 (MLwiN, Centre for Multilevel Modelling, University of Bristol, Bristol, UK).

Results

In the study population of 77 294 pregnancies, there were 2057 (2.7%) cases of term pre-eclampsia (gestational age ≥37 weeks), 350 (0.5%) cases of late preterm pre-eclampsia (34–36 weeks) and 216 (0.3%) cases of early preterm pre-eclampsia (<34 weeks).

Women with pre-eclampsia tended to be older, heavier, nulliparous and have a higher frequency of gestational diabetes and chronic disease, including diabetes mellitus type 1 or 2, than those without pre-eclampsia. Also, there were fewer smokers among women with pre-eclampsia (Table 1).These differences were more marked in early preterm pre-eclampsia (Table 1). There was no significant difference in marital status.

Table 1. Distribution of maternal characteristics in early and late pre-eclampsia and non-pre-eclamptic pregnancies
Pre-eclampsiaNo pre-eclampsia
<34 weeks (= 216)34–36 weeks (= 350)37 +  weeks (= 2057)(= 74 671)
n (%) n (%) n (%) n (%)
Maternal age (years)
<207 (3)12 (3)65 (3)1845 (2)
20–2430 (14)64 (18)382 (19)11 610 (16)
25–2961 (28)105 (30)681 (33)23 912 (32)
30–3469 (32)103 (29)591 (29)23 763 (32)
35–3943 (20)53 (15)281 (14)11 536 (15)
>396 (3)13 (4)57 (3)2003 (3)
Birth order
1114 (53)202 (58)1252 (61)32 366 (43)
264 (30)98 (28)515 (25)26 163 (35)
3+38 (18)50 (14)290 (14)16 142 (22)
Marital status
Married/cohabiting194 (90)314 (90)1855 (90)68 381 (92)
Other22 (10)36 (10)202 (10)6290 (8)
Smoking
Daily31 (14)56 (16)334 (16)11 559 (15)
Occasionally2 (1)8 (2)44 (2)1820 (2)
Pattern not specified20 (9)30 (9)216 (11)7621 (10)
Chronic maternal disease 42 (19)48 (14)266 (13)7139 (10)
Asthma18 (8)19 (5)129 (6)3817 (5)
Chronic hypertension17 (8)15 (4)66 (3)391 (1)
Chronic renal disease2 (1)4 (1)16 (1)548 (1)
Rheumatoid arthritis1 (1)1 (0.3)12 (1)329 (0.4)
Maternal heart disease2 (1)5 (1)13 (1)683 (1)
Thyroid disease6 (3)6 (2)51 (3)1748 (2)
Body mass index before pregnancy (kg/m2)
<18.5, underweight2 (1)9 (3)44 (2)3112 (4)
18.5–24.9, normal85 (39)171 (49)905 (44)46 334 (62)
25–29.9, overweight68 (32)96 (27)544 (26)16 640 (22)
30+, obese61 (28)74 (21)564 (27)8585 (12)
Diabetes mellitus type 1 8 (4)16 (5)30 (2)276 (0.4)
Diabetes mellitus type 2 3 (1)4 (1)19 (1)176 (0.2)
Gestational diabetes 4 (2)12 (3)55 (3)1015 (1)

Women with preterm pre-eclampsia (<37 weeks of gestation) were less likely to have heavy newborns (birthweight above the 90th or 95th centile), particularly those with early preterm pre-eclampsia (<34 weeks) (Table 2). In contrast, among women with term pre-eclampsia rates of birthweight above the 90th and 95th weight centiles were significantly higher than in term births without pre-eclampsia (Table 2). Unadjusted odds ratios (ORs) of LGA above the 90th and 95th birthweight centiles were 1.4, 95% confidence interval (95% CI) 1.2–1.6 and 1.6, 95% CI 1.3–1.9. However, after adjusting for maternal BMI, the excess of birthweight above the 90th and 95th centiles in women with term pre-eclampsia disappeared (OR 1.1, 95% CI 0.9–1.2 and 1.1, 95% CI 0.9–1.3), while the excess risk persisted after adjusting for pregestational and gestational diabetes without including BMI in the model (ORs for LGA above the 90th and 95th birthweight centiles 1.3, 95% CI 1.1–1.5 and 1.4, 95% CI 1.2–1.7).

Table 2. Odds ratios with 95% CI of birthweight centiles according to pre-eclampsia and week of gestation at delivery
Category of gestational age and pre-eclampsia (PE) n Total (%)Unadjusted OR (95% CI)Adjusteda OR (95% CI)Adjustedb OR (95% CI)
  1. a

    Adjusted for maternal age (<19, 20–24, 25–29, 30–34, 35–39, 40+, years), asthma, chronic hypertension, chronic renal disease, rheumatoid arthritis, maternal heart disease, thyroid disease, smoking (no, occasionally, daily, pattern not specified), and diabetes (no, type 1, type 2, gestational).

  2. b

    Adjusted for maternal age (<19, 20–24, 25–29, 30–34, 35–39, 40+, years), asthma, chronic hypertension, chronic renal disease, rheumatoid arthritis, maternal heart disease, thyroid disease, smoking (no, occasionally, daily, pattern not specified), diabetes (no, type 1, type 2, gestational), and maternal body index (continuous).

<5th birthweight centile
<34 weeks, no PE24675 (3.6)ReferenceReferenceReference
34–36 weeks, no PE1052478 (4.2)ReferenceReferenceReference
37+ weeks, no PE344971,518 (4.8)ReferenceReferenceReference
<34 weeks, PE11216 (5.1)1.5 (0.8–2.5)1.8 (0.8–3.7)1.3 (0.4–3.7)
34–36 weeks, PE46350 (13.1)3.4 (3.2–3.6)3.9 (2.7–5.7)3.5 (3.5–3.6)
37+ weeks, PE2222057 (10.8)2.4 (1.5–3.8)2.7 (2.4–3.2)2.4 (1.5–3.9)
<10th birthweight centile
<34 weeks, no PE54675 (8.0)ReferenceReferenceReference
34–36 weeks, no PE2032478 (8.2)ReferenceReferenceReference
37+ weeks, no PE705871,518 (9.9)ReferenceReferenceReference
<34 weeks, PE40216 (18.5)2.6 (2.4–2.8)2.8 (1.8–4.3)2.7 (2.6–2.9)
34–36 weeks, PE96350 (27.4)4.2 (3.8–4.6)4.5 (3.4–5.9)4.5 (4.0–4.9)
37+ weeks, PE3492057 (17.0)1.9 (1.7–2.1)1.9 (1.7–2.1)1.9 (1.7–2.1)
>90th birthweight centile
<34 weeks, no PE76675 (11.3)ReferenceReferenceReference
34–36 weeks, no PE2392478 (9.6)ReferenceReferenceReference
37+ weeks, no PE607671,518 (8.5)ReferenceReferenceReference
<34 weeks, PE5216 (2.3)0.2 (0.1–0.4)0.2 (0.1–0.4)0.1 (0.1–0.2)
34–36 weeks, PE26350 (7.4)0.8 (0.6–1.0)0.6 (0.4–1.0)0.5 (0.4–0.7)
37+ weeks, PE2332057 (11.3)1.4 (1.2–1.6)1.3 (1.1–1.5)1.1 (0.9–1.2)
>95th birthweight centile
<34 weeks, no PE30675 (4.4)ReferenceReferenceReference
34–36 weeks, no PE1242478 (5.0)ReferenceReferenceReference
37+ weeks, no PE294871,518 (4.1)ReferenceReferenceReference
<34 weeks, PE2216 (0.9)0.2 (0.1–0.6)0.2 (0.04–0.7)0.2 (0.1–0.3)
34–36 weeks, PE16350 (4.6)0.9 (0.7–1.3)0.7 (0.4–1.3)0.6 (0.4–1.3)
37+ weeks, PE1342057 (6.5)1.6 (1.3–1.9)1.4 (1.2–1.7)1.1 (0.9–1.3)

To control for the effect of gestational age, z-scores were calculated for women with and without pre-eclampsia in singleton births with gestational age 22–44 weeks. To increase the sample size in this analysis, 679 009 singleton births in 1999 to 2010 were included. A comparison between these two groups demonstrated a significant (< 0.001) displacement to the left of the birthweight distribution in pre-eclampsia (Figure 1). The displacement was more significant in preterm pre-eclampsia (Figure 1B,C). In contrast, the birthweight distribution in term pre-eclampsia was still to the left of non-pre-eclamptic women, but it was broader because of an excess occurrence of both small for gestational age (SGA) and LGA neonates (Figure 1A).

Figure 1.

Distribution of z-scores of birthweight in pre-eclamptic and non-pre-eclamptic pregnancies according to gestational age at birth. A total of 679 009 singleton births in 1999–2010 with gestational age 22–44 weeks.

Figure 2 shows the birthweight distribution in BMI groups with term births. In women with pre-eclampsia with normal BMI (18.5–24.9 kg/m2) no excess occurrence of large newborns was observed.

Figure 2.

Distribution of z-scores of birthweight in term (37 weeks of gestation or more) pre-eclamptic and non-pre-eclamptic pregnancies. Additionally, a subgroup of pre-eclamptic women with body mass index 18.5–24.9 kg/m2 (normal) is shown. Singleton births in 2007–10.

We assessed any evidence of selection bias in the subgroup with BMI data by performing separate analyses corresponding to Table 2 in the women with and without BMI data. Adjusted ORs of the 5th, 10th, 90th and 95th percentiles were similar in the subgroups with and without BMI (data not presented).

Discussion

Main findings

This population-based study demonstrates an excess of LGA neonates born at term after a pre-eclamptic pregnancy. The excess of LGA neonates was independent of gestational diabetes and diabetes mellitus types 1 and 2, but was attributable to an excess of maternal obesity among pre-eclamptic women delivering at term.

Strengths and limitations

Strengths of the study include its large size and its population-based design, which eliminate selection and recall bias. Birthweight and several variables that could potentially affect birthweight have previously been validated with satisfactory results including gestational age, birth order, pre-eclampsia and diabetes.[22-24] A limitation was that data on maternal BMI were available only in a subset of women. However, the statistical analysis demonstrated that this group of women was representative of the total population. Our database lacked data on ethnicity. However, women of different ethnicity represent only a small proportion of the population. Moreover, previous reports have indicated that women of foreign ethnicity living in Norway tend to more often have macrosomic newborns in diabetic as well as non-diabetic pregnancies and less often in pre-eclampsia.[25] Hence, adjusting for ethnicity would not dilute the association between pre-eclampsia and large newborn size.

Our observation of an excess of LGA neonates in pre-eclamptic term deliveries in the unadjusted analysis is consistent with previous population-based studies.[10, 12, 26] However, in a smaller study the authors did not find an excess of LGA associated with pre-eclampsia even before adjusting for BMI.[27] These conflicting reports may be the result of a selection bias in the latter study performed at a single tertiary-care centre.

Interpretation

Consistent with prior reports.[10, 12, 26] the present study demonstrated that term pre-eclampsia is associated with an excess of both SGA and LGA neonates. In addition, our population-based study provides evidence that the excess of LGA neonates in term pre-eclampsia can be attributed to maternal obesity. The excess of SGA in preterm pre-eclampsia supports the prevailing hypothesis that chronic uteroplacental ischaemia, due to placental vascular insults, shallow trophoblast invasion of the spiral arteries or abnormal fetal–placental circulation, plays an important role in preterm pre-eclampsia, fetal growth restriction or both.[18, 28-31]

It has been suggested that the excess of large fetuses in term pre-eclampsia indicates that placental dysfunction plays a minor role in this subset of pre-eclamptic pregnancies.[10, 12, 26] Morphological and experimental studies are in agreement with these results. Egbor et al.[30] reported that early onset pre-eclampsia is associated with abnormal placental morphology, whereas placentas from late-onset pre-eclampsia are morphologically similar to placentas from gestationally age-matched non-pre-eclamptic pregnancies. Hence, it is possible that in a subset of women with late onset pre-eclampsia the disease process may be associated more with maternal factors including obesity than with placental factors. Earlier reports have indicated an increased uteroplacental perfusion, as determined by placental clearance of dehydroisoandrosterone sulphate[32] and increased maternal cardiac output in late-onset pre-eclamptic pregnancies.[33] These observations suggest an increased uteroplacental perfusion in late-onset pre-eclampsia, presumably as a compensatory mechanism due to a mismatch between uteroplacental blood flow and increased demand for nutrients. However, additional studies are required to confirm this hypothesis.

In the present study, we found that in term pre-eclampsia the excess of LGA did not persist after adjusting for maternal BMI as a continuous variable. This observation suggests that maternal obesity accounts for the association between term pre-eclampsia and LGA. Our results are consistent with the report of Xiong et al.[10] that adjusting for high and low maternal weight reduced the effect of term pre-eclampsia on LGA. These results are also consistent with a solid body of evidence indicating that maternal obesity is a risk factor for both pre-eclampsia[31, 34-36] and LGA.[16, 37-40] In obese women, even in the absence of pregestational diabetes or gestational diabetes, it is possible that fetal overgrowth may lead to a mismatch between increasing fetal needs for nutrients and the ability of the placenta to keep up with these needs. This in turn may lead to fetal signalling to increase the maternal blood pressure in an attempt to compensate for a relative uteroplacental ischaemia.[17, 18] The mechanisms of disease in lean and obese women with pre-eclampsia remain unclear. However, a two-stage hypothesis has been proposed whereby the first stage is characterised by abnormal fetal trophoblast invasion of the maternal spiral arteries, and the second stage is characterised by widespread inflammation and dysfunction of the maternal endothelium associated with the clinical presentation of pre-eclampsia.[31] Obese women may be at higher risk for pre-eclampsia as a result of the oxidative stress associated with increased circulating concentration of lipid peroxides.[31, 34] Several studies have reported that maternal obesity is associated with fetal macrosomia, even in the absence of maternal diabetes.[16, 37-40] The combination of increased nutrients to the fetus and fetal hyperinsulinaemia in obese women may explain the increased frequency of LGA in obese women without diabetes.[15, 16] Moreover, although glucose is a major source of energy in the fetus, other maternal nutrients may influence fetal growth. Evidence in support of this view is the observation that when maternal glucose concentrations are properly controlled in women with gestational diabetes, maternal triglyceride and free fatty acids are independent predictors for the delivery of a LGA neonate.[41] This is in keeping with our results that maternal obesity may account for the excess of macrosomic neonates in pre-eclamptic women delivering at term.

Conclusions

The results of this study suggest an excess of both accelerated and impaired fetal growth in pre-eclamptic pregnancies. The excess of LGA neonates can be attributed to an excess of maternal obesity among pre-eclamptic women delivering at term. We anticipate that the increasing prevalence of obesity may lead to higher rates of both pre-eclampsia and the delivery of LGA neonates in pregnant women at term, further straining the resources allocated for antenatal care. Future longitudinal studies should further explore the associations of maternal obesity with fetal macrosomia and late onset pre-eclampsia. Such studies would require analyses of obese and lean women with respect to fetal growth, preferably including data on placental perfusion and morphology.

Disclosure of interests

None declared.

Contribution to authorship

SR prepared the analytical database, conducted the analyses and wrote the report. SR, LMI and JE discussed core ideas and study design and edited the report. All authors are guarantors of the paper.

Disclaimer

This study has used data from the Medical Birth Registry of Norway. The interpretation and reporting of these data is the sole responsibility of the authors, and no endorsement by the Medical Birth Registry of Norway is intended nor should be inferred.

Details of ethics approval

Based on anonymised registry data, the Regional Ethics Committee exempted this study from ethical review.

Funding

No external funding.

Need to evaluate empirical evidence before formulating theory

  • S Lisonkova, KS Joseph

  • Department of Obstetrics and Gynaecology, University of British Columbia, Vancouver, BC, Canada

Linked article This article is a mini commentary on Rasmussen S et al., pp 1351–7 in this issue. To view this article visit http://dx.doi.org/10.1111/1471-0528.12677.

Professor Rasmussen and colleagues attempted to answer the causal question: Does pre-eclampsia increase the risk of large-for-gestational-age (LGA) infants? This issue is important because previous findings of an excess of macrosomic and LGA infants among women with pre-eclampsia have intrigued researchers and influenced pre-eclampsia theory.

Numerous studies have shown that women with pre-eclampsia are at high risk for small-for-gestational-age infants. This is consonant with the widely accepted theory that pre-eclampsia is caused by an aberrant development of the placental vasculature. However, this fundamental aspect of the pathophysiology of pre-eclampsia needed to be expanded to accommodate the unexpected finding of a simultaneous excess of LGA infants among women with pre-eclampsia (Maulik J Matern Fetal Neonatal Med 2003;13:145–6). The combination of fetal growth restriction and excess fetal growth appeared to support a previous theory that pre-eclampsia involves heterogeneous processes; one originating in abnormal placentation, and the other involving maternal factors and enthothelial injury but no placental dysfunction (Ness et al. Am J Obstet Gynecol 1996;175:1365–70). Whereas this theory regarding the placental and maternal origins of pre-eclampsia is supported by several empirical findings, the apparent excess of LGA infants among women with pre-eclampsia cannot serve to bolster this proposition. As Rasmussen et al. show, the association between pre-eclampsia and LGA is spurious and entirely explained by the confounding effect of maternal prepregnancy body size.

This cautionary tale should give us pause; empirical evidence needs to be assessed carefully before it can become the underpinning for theory. Fundamental epidemiological principles should always receive proper attention; adjustment for confounders and care in avoiding residual confounding are key issues.

Another case in point is the empirical finding that preterm infants of women with pre-eclampsia have lower perinatal mortality than preterm infants born to women without pre-eclampsia. This paradoxical observation is in fact part of a more general phenomenon: perinatal mortality rates are relatively lower among low birthweight and preterm babies of several vulnerable groups including women who smoke. From a prognostic (noncausal) standpoint, the lower risk of perinatal death among preterm babies of women with pre-eclampsia and women who smoke is accurate. However, from a causal perspective, the association between pre-eclampsia/smoking and perinatal death is probably spurious; neither pre-eclampsia nor smoking confers benefit on the fetus or infant. The phenomenon is best explained with the fetuses-at-risk formulation, which shows that perinatal mortality rates are higher among women with pre-eclampsia and those who smoke at all gestational ages (Lisonkova et al. Am J Obstet Gynecol 2013;209:544.e1–544.e12). Under the fetuses-at-risk model, perinatal deaths at any gestation are viewed as incident cases that occur among the population of fetuses at risk of perinatal death at that gestation. This viewpoint contrasts with the traditional gestational age-specific perinatal mortality rate calculation in which rates are based on perinatal deaths occurring among births at a particular gestational age.

Pre-eclampsia has been called a disease of theories (Higgins et al. Curr Opin Obstet Gynecol 1998;10:129–33). Carefully evaluated empirical evidence should guide the refinement of these theories, a process that will eventually uncover the aetiology of this puzzling and complex disease.

Disclosure of interests

Neither author has anything to disclose.

Ancillary