Effect of pregravid obesity on perinatal outcomes in singleton pregnancies following in vitro fertilization and the weight-loss goals to reduce the risks of poor pregnancy outcomes: A retrospective cohort study.

Obesity is a major global health issue, and its severity is increasing in recent years. The worldwide proportion of women with a body mass index (BMI) of above 25 kg/m 2 has increased from 29.8% in 1980 to 38% in 2013, which is largely driven by new cases from Asia [ 1 ]. In Asia, the prevalence of obesity is very low previously, while it is increasing at an alarming rate recently, especially in China, Japan and India [ 2 ]. The number of Chinese obese people is below 0.1 million in 1975, while such number has reached 43.2 million in 2014, accounting for 16.3% of worldwide obesity [ 3 ]. As obesity and overweight have become one of the most important threats to human health in general, it has also become one of the most common medical conditions complicating pregnancies of women of reproductive age. Now it is not uncommon for overweight and obese women to seek fertility treatment, such as in vitro fertilization (IVF) [ 4 ]. Previous studies have found that the presence of excessive maternal adipose tissue is linked to a number of important adverse outcomes in spontaneous pregnancies. However, the effects of obesity on risks of maternal and fetal adverse outcomes in pregnancies following successful IVF remain largely unexplored. The 2013 American College of Obstetricians and Gynecologists strongly recommends preconception counseling for overweight and obese women about maternal and fetal risks in pregnancy and encourage them to undertake a weight-loss program [ 5 ]. Until now, there is insufficient data regarding the effects of weight loss on the risks of perinatal complications. The gold standard evidence to inform this counseling would come from randomized trials of preconceptional weight-loss interventions. However, such studies are difficult to conduct in IVF pregnancies. Therefore, population-based studies comparing the pregnancy outcomes of different women based on their pre-IVF BMI are important to provide weight-loss goals prior to conception with the aim to reduce perinatal complications. The aim of this study was to evaluate whether pre-IVF obesity independently predicts increased pregnancy complications after adjusting for important confounders. We also aimed to provide recommendations for Chinese women about the magnitude of weight loss prior to IVF for better perinatal outcomes.

A total of 3,962 singleton births were assessed for selected adverse pregnancy and birth outcomes. Table 1 lists the baseline characteristics of all participants. The obese population was significantly older than the normal-weight population and less frequently diagnosed with tubal factor. The proportion of women with PCOS in the obese population was significantly higher compared with the normal-weight group. The rate of “long agonist protocol” used in COH was significantly lower in overweight women compared with the other two groups. Parity was not significantly different across groups. Data were presented as median and interquartile range (IQR) or n (%). a. Pairwise comparisons revealed a statistically significant difference between the first and second BMI categories. b. Pairwise comparisons revealed a statistically significant difference between the first and third BMI categories. c. Pairwise comparisons revealed a statistically significant difference between the second and third BMI categories. In the unadjusted analyses ( Table 2 ), obesity was associated with increased risks of GDM, HDP, PPROM, placental abruption, PTB <37 weeks, CS, fetal macrosomia, LGA, LBW<2,500 g, NRDS, neonatal intensive care unit (NICU) admission and congenital anomalies. In the adjusted analyses ( Table 3 ), the significantly increased risk of LBW<2,500 g disappeared, whereas the following pregnancy complications remained significant after adjustment of age, parity, PCOS, and type of COH: GDM (aOR: 2.32, 95% CI: 1.58–3.40), GH (aOR: 3.08, 95% CI: 2.11–4.50), PE (aOR: 2.92, 95% CI: 1.19–7.20), polyhydramnios (aOR: 2.25, 95% CI: 1.14–4.47), PPROM (aOR: 2.92, 95% CI: 0.94–2.77), placental abruption (aOR: 4.51, 95% CI: 1.30–15.60), PTB <37 weeks (aOR: 1.68, 95% CI: 1.18–2.37), CS (aOR: 2.19, 95% CI: 1.63–2.95), fetal macrosomia (aOR: 2.19, 95% CI: 1.63–2.95), NRDS (aOR: 3.17, 95% CI: 1.23–8.19), LGA (aOR: 2.33, 95% CI: 1.85–2.94), NICU admission (aOR: 1.51, 95% CI: 1.04–2.29) and congenital anomalies (aOR: 1.63, 95% CI: 1.04–2.56). The risks of GDM, CS, LGA, fetal macrosomia and NICU admission were considerably increased in the overweight and obese women compared with the normal-weight ones, whereas the remaining selected adverse pregnancy and birth outcomes appeared to be significantly increased only in the obese women. In the subgroup analysis of organ specific malformations, there was a statistically significant increase for malformations of the urogenital system and congenital heart defects in the obese population. Compared with offspring of normal-weight mothers, the aOR for urogenital system malformations was 2.48 (95% CI: 1.13–7.14) for obese mothers, and that for congenital heart defects was 2.30 (95% CI: 0.64–8.27). The results were presented in Fig 2 . (adjusted for age, PCOS, parity and type of COH). PTB<37 weeks and PPH seemed to have a less significant association with BMI in the present study. We observed that obese women had a slightly higher rate of LBW<2,500g, but a lower rate of LBW<1,500 g. Even though both of them were not statistically significant between the obese group and reference group (REF), GA at birth was significantly lower in obese pregnancies not only compared with the normal-weight group but also the overweight group. There were no significant differences in rates of SGA, PP or perinatal mortality. To assess whether the increased risk of perinatal complications might be mediated by development of GDM and HDP, we performed logistic regressions for those outcomes with a significant association with pre-IVF BMI that was adjusted for development of GDM and HDP separately (in addition to age, PCOS, parity and type of COH) ( Table 4 ). There were no longer increased risks of PPROM (aOR: 1.94, 95% CI: 0.97–3.85, P = 0.060), NRDS (aOR: 2.59, 95% CI: 0.96–7.01, P = 0.061) and congenital anomalies (aOR: 1.54, 95% CI: 0.98–2.43, P = 0.061) once the development of GDM for obese women was adjusted when compared with the normal-weight women. The increased risk of NRDS was eliminated after adjustment of HDP development (aOR: 2.58, 95% CI: 0.95–7.01, P = 0.063). In the group of women without PCOS, GDM, GH, PE, polyhydramnios, placental abruption, CS, fetal macrosomia, LGA, NRDS and NICU admission were significantly more common in the obese group compared with the normal-weight group ( Table 5 ). However, the increased risks of PTB <37 weeks and placental abruption were no longer observed after adjustment of HDP. In patients with PCOS, GDM, HDP, PPROM, CS, fetal macrosomia and LGA were the outcomes that were significantly changed with the increase of BMI ( Table 6 ). Tables 7 – 10 show the comparisons of adverse perinatal outcomes between the target BMI group and BMI reduction group. There was a statistically significant difference for congenital anomalies between the group with a BMI of 30–32 and the group with a BMI of 27–29, representing a 10% reduction in BMI. Apart from congenital anomalies, there was no statistically significant difference regarding other pregnancy outcomes between the two groups. In contrast, women with a BMI of 30–32 were associated with higher risks of GH, fetal macrosomia and LGA when compared with the women with a BMI of 25–27, representing a 15% reduction in BMI. No significant difference was observed regarding the perinatal complications between the group with a BMI of 28–29 and group with a BMI of 25–26, representing approximately a 10% reduction in BMI. Rates of GH, CS and fetal macrosomia were significantly different between the group with a BMI of 28–29 and group with a BMI of 24–25, representing a 15% reduction in BMI. Pregravid BMI in the overweight range was associated with higher rates of GDM, CS, fetal macrosomia, LGA and NICU admission. BMI of 26–27 resulted in increased rates of CS, fetal macrosomia, and LGA when compared with BMI of 23–24, representing a 10% reduction in BMI. The same results were seen between BMI of 26–27 and BMI of 24–25, representing a 5% reduction in BMI. Rates of GDM, LGA and NICU admission were significantly higher among women with a BMI of 24–25 compared with those with a BMI of 21–22, representing a 10% reduction in BMI. Meanwhile, the rates of GDM and NICU admission in women with a BMI of 24–25 were still significantly higher than those in women with a BMI of 22–23, representing a 5% reduction in BMI.

Collectively, pregravid obesity served as an independent predictor of adverse birth outcomes in IVF pregnancies. Our results suggested that some risks could occur through development of HDP and GDM. It is hard for obese women to lose enough weight to normal BMI categories. We encouraged obese women to lose weight to a 10–15% reduction in pregravid BMI, which was useful to reduce the risks of some perinatal complications. For overweight women, just a 5% reduction in pregravid BMI was helpful. However, we used BMI definitions of WGOC in the present study, making the recommendations less applicable to general international population. Prospective studies are required to further demonstrate the weight-loss goals to reduce the risks of poor perinatal outcomes for women with high BMI. The study was approved by the institutional review board of the Reproductive Hospital Affiliated to Shandong University. The ethics board approval number is 201424. The data were anonymously analyzed, so no consent was required.

This retrospective cohort study was carried out at the Reproductive Medical Center affiliated to Shandong University. The Centre routinely collects pregnancy and delivery information from postpartum patients. Women who underwent their first IVF cycle and delivered a single live infant (vanishing twin and selective reduction were excluded) at ≥28 weeks of gestation were enrolled in the cohort. Those who had internal medical conditions, especially pre-IVF hypertension and mellitus diabetes, recurrent spontaneous abortion (defined as three or more previous spontaneous miscarriages), cervical incompetence or chromosomal abnormality were excluded from the present study. To eliminate age as an independent variable for IVF pregnancy, women aged 38 years or older were excluded from this study. Of the 4,670 charts identified with a singleton live birth, 356 subjects used donor sperm, 204 women were over 38 years of age, 42 women had internal medical conditions, 56 cases had chromosomal abnormality and underwent preimplantation genetic diagnosis, and 50 births did not meet inclusion criteria or contained insufficient information The flow chart was presented in Fig 1 . Eventually, a total of 3,962 women were included in the final analysis. Of these enrolled women, 584 subjects had polycystic ovary syndrome (PCOS), and 3,378 women did not. They were categorized into three groups according to their BMI, which was measured at the initial IVF consultation (weight [kilograms]/height [meters] 2 ). The WHO expert consultation has reviewed scientific evidence and suggested that Asian populations have different associations between BMI, percentage of body fat, and health risks compared with European populations. They conclude that the proportion of Asian people with a high risk of type 2 diabetes and cardiovascular disease is substantial at BMI lower than the existing WHO cut-off point for overweight (> or = 25 kg/m 2 ) [ 6 ]. Since only Asian women were included in the present study, it might be more reasonable to classify them according to the BMI definitions of the Working Group on Obesity in China (WGOC). BMI groups were defined as follows: normal weight (BMI<24.00 kg/m 2 ), overweight (BMI 24.00–27.99 kg/m 2 ) and obese (BMI≥28.00 kg/m 2 ). Most of the studies on effects of BMI on perinatal outcomes have focused on spontaneous pregnancies. Therefore, PCOS is often a confounding factor being omitted. However, it remains unclear whether the reported effects of obesity on pregnancy outcome are independent of the effects of PCOS. We therefore performed two subgroup analyses to isolate the effect of obesity from PCOS on pregnancy outcomes. For those outcomes significantly associated with pregravid obesity or overweight, additional analyses were carried out to compare the target BMI group and corresponding BMI reduction group. Because the National Institutes of Health (NIH) recommends a 10% reduction in body weight to confer health benefits outside of pregnancy [ 7 ], a 10% reduction in prepregnancy BMI was defined as the BMI reduction goal to make the weight-loss model. For example, the risks among women with a prepregnancy BMI of 30–32 were compared with risks among women with a BMI of 27–29, which represents approximately a 10% reduction in BMI. Four models were conducted as follows: group with a BMI of 30–32 vs. group with a BMI of 27–29; group with a BMI of 28–29 vs. group with a BMI of 25–26; group with a BMI of 26–27 vs. group with a BMI of 23–24; group with a BMI of 24–25 vs. group with a BMI of 21–22, and the BMI reduction group was approximately a 10% reduction of the target BMI group. Logistic regressions were performed between the target BMI group and BMI reduction group (the control group) separately. If there was a statistically significant increase in the risks of poor pregnancy outcomes between the target BMI group and BMI reduction group, the target BMI group was continuously compared with a smaller BMI reduction group, which presents only a 5% difference in prepregnancy BMI. For example, group with a BMI of 26–27 vs. group with a BMI of 24–25; group with a BMI of 24–25 vs. group with a BMI of 22–23; the BMI reduction group was approximately a 5% reduction of the target BMI group. On the contrary, if there was no statistically significant difference between the target BMI group and the control group, a stricter group which presents a greater magnitude of weight loss (a 15% difference in pre-IVF BMI) was defined as the control group. For example, group with a BMI of 30–32 vs. group with a BMI of 25–27; group with a BMI of 28–29 vs. group with a BMI of 24–25. The following adverse maternal and perinatal outcomes were examined: 1) gestational diabetes mellitus (GDM) was diagnosed via the oral glucose tolerance test (75 g, 2 h) [ 8 ], 2) hypertensive disorders of pregnancy (HDP), including gestational hypertension (GH) and pre-eclampsia (PE) as per the International Society for the Study of Hypertension in Pregnancy guidelines [ 9 ], 3) polyhydramnios was defined as amniotic fluid index (AFI) >24 cm, whereas oligohydramnios was defined as AFI <8 cm, 4) placenta previa (PP) refers to that the placenta partially or completely obstructs the internal orifice of the cervix by lying the lower uterine segment, 5) placental abruption was defined as the premature detachment of the placenta from the uterine wall before birth and after 20 weeks’ gestation, 5) postpartum hemorrhage (PPH) was defined as blood loss of more than 500 mL within 24 h after vaginal delivery or more than 1,000 mL after caesarean section (CS), 6) PPROM, and 7) mode of delivery (rate of CS). Birth outcome variables included gestational age (GA) at delivery (week), birth weight (g), birth height (cm), preterm birth (PTB)<32 and <37 weeks, low birth weight (LBW<1,500 and 4,000 g), small for gestational age and large for gestational age (SGA and LGA; 90th percentiles, respectively, according to Fenton 2013 growth curves [ 10 ]), neonatal respiratory distress syndrome (NRDS) (defined as one or more signs of increased work of breathing, such as tachypnea, nasal flaring, chest retractions and grunting), congenital malformations, and perinatal mortality (≤28 days). Statistical analysis was performed with SPSS 20.0. Descriptive statistical methods were used to summarize the study population. Participant characteristics were summarized using median and interquartile range (IQR) for continuous variables, and counted with percentages (%) for categorical variables. The Wilcoxon rank-sum test was used to evaluate differences between continuous variables, and Fisher’s exact test and X 2 were performed for categorical variables to compare data of the three BMI categories. For each outcome, logistic regression was used to estimate odds ratio (OR) and 95% confidence interval (CI). Initially, unadjusted ORs were calculated for all outcomes by fitting univariable logistic regression models. Then, multiple logistic regression models were constructed to examine the magnitude and significance of the independent effect of BMI by adjusting maternal age, parity, PCOS, and type of controlled ovarian hyperstimulation (COH). To demonstrate how obesity affected neonatal outcomes, PTB was also adjusted (in addition to age, parity, PCOS and type of COH) in logistic regression analyses. A P value of <0.05 was considered as statistically significant.

(XLS) Click here for additional data file.