Obstetric and Perinatal Outcomes in Women with Eating Disorders

Introduction

Eating disorders are a major source of psychiatric morbidity in women of childbearing age. ¹ Up to 1% of women in developed countries will be diagnosed with an eating disorder during their lifetime. ² The current Diagnostic and Statistical Manual of Mental Disorders (DSM-4), lists three classes of eating disorders: anorexia nervosa, bulimia nervosa, and eating disorder not otherwise specified. ³ Women with eating disorders experience image distortion and exhibit an obsessive desire to be thin. These characteristics lead to abnormal dieting, purging behavior, and abuse of laxative and diuretics. ^{4 –9} These behaviors may cause physiologic complications, such as electrolyte and mineral imbalances, functional and structural abnormalities of internal organs, and severe metabolic alterations, ^{9 –11} as well as amenorrhea in postmenarchal women, which is a diagnostic criterion of anorexia nervosa. ⁵

Data about the effects of eating disorders on the course of pregnancy and delivery are scarce. However, eating disorders have been implicated as having a negative effect on pregnancy and perinatal outcomes. Some pregnancy and delivery complications have been shown to be more prevalent among patients with eating disorders, but other complications remain controversial. ^{12 –17} Higher rates of miscarriages, ^14,15 genitourinary tract infection, ¹⁶ microcephaly and smaller head circumference, ¹⁷ and cesarean deliveries ¹² were noted among patients with eating disorders. Whereas some authors reported higher rates of preterm delivery among patients with eating disorders, ^11,12,16 others have not found such an association. ^14,17 Kouba et al. ¹⁷ reported higher rates of small for gestational age (SGA) infants in women with eating disorders. Conversely, Ekeus et al. ⁷ did not find a statistically significant difference between women with and without eating disorders with regard to the main birth outcome measures that were investigated.

The purpose of this study was to investigate whether women with a history of an eating disorder have an increased risk for adverse obstetric and perinatal outcomes.

Materials and Methods

A population-based study comparing all singleton deliveries of women with and without eating disorders was conducted. Deliveries occurred during the years 1988–2009 at the Soroka University Medical Center, the sole hospital in the Negev, the Southern region of Israel, which serves the entire obstetric population. Thus, the study represents nonselective data. After approval of the local ethics institutional review board, cases were located through the clinic for eating disorders at the psychiatric department. The study group consisted of outpatients who were diagnosed by clinical interviews conducted by the medical staff of the eating disorder outpatient clinic as having eating disorders (either anorexia nervosa, bulimia nervosa, or eating disorder not otherwise specified) using the DSM-4 criteria. ³ Any comorbidity of eating disorders with chronic and severe psychiatric disorders, such as schizophrenia, was excluded. The eating disorders clinic offers its patients a variety of treatments, including individual psychotherapy, group treatments, nutritional advisors, and weight follow-up. A specific treatment is tailored to each patient. Data were collected from the perinatal database of information reported by an obstetrician directly after delivery.

Women lacking prenatal care and with multiple gestations were excluded from the study. A total of 127 births to women diagnosed with eating disorders occurred at our center during the study period. Of these, only 5 deliveries (3.9%) were of Bedouin ethnicity, a group that accounts for >50% of the births in our center. Therefore, it is reasonable to assume that eating disorders are underestimated in this ethnic group. Hence, we excluded deliveries of women of Bedouin ethnicity from the study group as well as from the control group. Accordingly, only births to Jewish women were included in the analysis.

The following clinical characteristics were analyzed. (1) Demographic and clinical characteristics—maternal age, parity, recurrent abortions, fertility treatments, gestational age, neonatal gender, and birth weight. These characteristics are related to the different pregnancy and perinatal outcomes that we evaluated in the study. (2) Obstetric risk factors and perinatal complications—previous cesarean delivery (CD), hypertensive disorders, gestational diabetes mellitus (GDM), suspected intrauterine growth restriction (IUGR), placenta previa, polyhydramnion, oligohydramnion, and premature rupture of the fetal membranes (PROM). (3) Labor and delivery complications—labor induction, malpresentation, nonprogressive labor in the first stage, nonprogressive labor in the second stage, placental abruption, CD, and postpartum hemorrhage. (4) Perinatal outcomes—congenital malformations, Apgar scores <7 at 1 and 5 minutes, and perinatal mortality.

Statistical analysis was performed with the SPSS package, version 14.0 (SPSS, Chicago, IL). Statistical significance of the categorical variables was tested using the chi-square or Fisher's exact test, as appropriate. For continuous variables, Student t test was used. Multivariable logistic regression models were constructed to control for confounders. Odds ratios (OR) and their 95% confidence interval (CI) were computed. A p value<0.05 was considered statistically significant.

Results

A total of 117,875 deliveries occurred during the study period. Of these, 122 (0.1%) were to women suffering from an eating disorder; this resulted in a rate of 10.3/10,000 deliveries. The diagnosis of an eating disorder was made in the eating disorders clinic of the department of psychiatry at our hospital, and we refer to it as lifetime prevalence diagnosis. Of these, 41 were diagnosed with anorexia nervosa, 62 had a diagnosis of bulimia nervosa, and 19 had an eating disorder not otherwise specified. Table 1 compares demographic and clinical characteristics of pregnancies of patients with and without eating disorders. Women with eating disorders had significantly higher rates of preterm deliveries (<37 weeks gestation) as well as significantly higher rates of fertility treatments. An increased risk of delivering a term low birth weight infant was noted among the women with eating disorders. Finally, nulliparity was significantly more common among women with eating disorders.

Obstetric risk factors and perinatal complications are presented in Table 2. Rates of IUGR were significantly increased among patients with eating disorders compared with controls. Table 3 shows the labor and delivery complications of patients with and without eating disorders. Higher rates of CD were found among patients with eating disorders.

The perinatal outcomes of patients with and without eating disorders are presented in Table 4. Increased rates of neonates born with a low birth weight were noted among patients with eating disorders. No other significant differences were noted between the groups.

Using three different multivariable logistic regression models (Table 5), controlling for parity and maternal age, preterm delivery, low birth weight at term, and cesarean section remained independently associated with eating disorders (adjusted OR 2.2, 95% CI 1.4-3.6, p=0.001; adjusted OR 2.8, 95% CI 1.4-5.5, p=0.003; and adjusted OR 1.9, 95% CI 1.3-2.9, p=0.002, respectively). After controlling for parity and maternal age, however, the higher rates of IUGR and fertility treatments noted among women with eating disorders lost their significance. (adjusted OR 1.5, 95% CI 0.7-3.2, p=0.348; adjusted OR 1.3, 95% CI 0.6-3.0, p=0.515, respectively).

Discussion

The major findings of our study are that pregnancies of women with eating disorders are associated with higher rates of adverse maternal and perinatal outcomes, such as fertility treatments, preterm delivery, low birth weight, IUGR, and CD. In accordance with our findings, previous studies reported higher rates of IUGR ^16,17 and low birth weight infants ^12,14,17 among patients with eating disorders. Conversely, Franko et al. ¹³ found that the majority of the women with eating disorders had pregnancies that resulted in infants of normal birth weight. Nevertheless, their ability to determine predictor variables was limited by the small group size, as they compared between groups with eating disorders. Moreover, their absence of a comparison group does not allow drawing definitive conclusions. ¹³

The association between eating disorders and preterm delivery is controversial. In our study, patients with eating disorders had a higher risk of delivering prematurely regardless of maternal age and parity. This finding, as well as a higher rate of IUGR in women with eating disorders, could be associated with low pregnancy body mass index (BMI) and low weight gain during pregnancy. ^18,19

There is controversy about the association between eating disorders and CDs. A previous study has shown that women with eating disorders have a higher risk of delivering by cesarean section. ¹² It is important to note that in that study, only the subgroups of bulimia nervosa and binge eating disorder were found to have a significant association with CD. Other studies did not report such an association. ^16,17 In our cohort, higher rates of CD were noted among patients with eating disorders, and in a multivariable logistic regression model, CD was independently associated with eating disorders. It could be interesting, therefore, to conduct further research to find additional factors that are associated with CD among women with eating disorders. Such factors would be able to better explain the relationship between eating disorders and CD.

Franko et al. ¹³ reported an increased rate of fetal malformations in patients with eating disorders. Kouba et al. ¹⁷ found an increased rate of small head circumference and microcephaly among patients with eating disorders, although no other fetal abnormalities were noted. In our study, we noted a difference in congenital malformations between women with and without eating disorders, although it was not significant (8.2% vs. 4.7%, p=0.072).

Finally, our study was the first to note significantly higher rates of fertility treatments among women with eating disorders. It is reasonable to assume that because of the severe metabolic alterations noted in women with eating disorders ^{9 –11} as well as amenorrhea in postmenarchal women, ⁵ increased risk of infertility will be noted among women with eating disorders. While including only the women who gave birth, we probably underestimated the higher rates of infertility among women with eating disorders, which lost their significance in the multivariable logistic regression model.

Our study offers several strengths. Our large sample size (achieved while combining all eating disorders) allowed studying a relatively rare diagnosis in our population and its association with several clinically important outcomes. Additionally, the comprehensive database allowed us to access pregnancy outcome information that was obtained in a prospective manner. It should also be noted that a large number of obstetric and perinatal outcomes were examined.

Our study has several inherent weaknesses, however, mostly due to its retrospective nature. One potential weakness is that some undiagnosed women with eating disorder are included in the general population group, an assumption that is reinforced by the fact that the prevalence of diagnosed eating disorders is smaller than described in the literature. ² Cases were located through the clinic for eating disorders at the psychiatric department. It could be expected that only the most severe cases reached the clinic, and it is, therefore, likely that a proportion of patients with milder or untreated eating disorders were missed. We did not discriminate our analysis according to the status of the disease (past or current) or the type of the disease because of the small sample size. Bulik et al. ¹² found no differences between women with active vs. remitted anorexia nervosa on any of the investigated measures. One study found that only 44% of women report their eating disorder to their physician, ²⁰ probably because of the shame and secrecy associated with eating disorders. ^20,21 Therefore, the prevalence of eating disorders is likely underestimated in the obstetric population. ¹⁶ Because the diagnosis of an eating disorder is confidential, the physicians, at the time of labor and delivery, were blinded to this diagnosis. Therefore, caregiver bias should be ruled out, thus increasing the clinical utility of our study.

Another limitation of our study was derived from its focus. The focus of our study was the obstetric and perinatal outcomes in a population of women with eating disorder and not on the characteristics of the eating disorder. Therefore, some information that is related to characterization of patients with eating disorder, such as the woman's weight or her lifetime minimum BMI, the duration of the disorder, or if the woman vomited during pregnancy as part of the disorder, was not available and therefore is not presented in the study. In addition, there are confounding variables that were out of our reach, such as social class and smoking history, that could have contributed to the analysis. Although many obstetric and perinatal variables were evaluated in the study, it would be interesting to evaluate other variables in further studies, such as the presence of hyperemesis gravidarum, infectious diseases, and cholestasis of pregnancy.

Finally, a weakness inherent to retrospective cohort studies is the potential for missing data. However, the data were reported by an obstetrician directly after delivery. Skilled medical secretaries routinely reviewed the information before entering it into the database. Coding was done after assessing the medical prenatal care records together with the routine hospital documents. Women lacking prenatal care were excluded from the study analysis. This makes this potential source of information bias less likely.

If an eating disorder is suspected or there is a known history of an eating disorder, it is important to have a nonjudgmental attitude toward the patient and to ask open questions that assess the patient's perception of her weight gain, body image, exercise, and eating patterns. ^5,10 Previous research suggests that a multidisciplinary approach involving a gynecologist, psychiatrists, and nutritionists may achieve better results in the management of parturients with eating disorders. ^5,10

Conclusions

Pregnant women with a diagnosis of an eating disorder appear to be at greater risk for certain perinatal complications. If the diagnosis of eating disorder is known, these patients should be considered at high risk and monitored closely during pregnancy to optimize maternal and fetal outcomes.