Hydramnios complicates 0.4–1.9%1–3 of all pregnancies and is associated with maternal diabetes and other diseases, placental diseases such as chorioangioma, and various fetal abnormalities. The many associations make appropriate diagnostic evaluation for hydramnios controversial. Although several reports1–16 documented an increased incidence of congenital anomalies and poor perinatal outcomes in pregnancies complicated by hydramnios, most were relatively small, thus not conducive to accurate risk estimates. However, many clinicians recommend routine diagnostic procedures such as targeted ultrasound examination and amniocentesis based on such studies.4–13
Besides the lack of precise fetal risk data, which would support costly and invasive procedures, few data exist on the risk to ongoing pregnancy once fetal structural abnormalities and aneuploidy have been ruled out. Medically appropriate and cost-effective monitoring strategies cannot be implemented without such information. The purpose of this investigation was to determine the risk of adverse perinatal outcomes in a large series of hydramnios cases.
Materials and Methods
With Institutional Review Board approval, we reviewed the computerized records of all ultrasound examinations done at our institution from 1986 to 1996 to identify all women with singleton gestations in which amniotic fluid (AF) volumes were assessed. Women who received care in our clinics typically had at least two sonographic examinations during their pregnancies: one before 20 weeks to confirm estimated gestational age, and one at or after 20 weeks to evaluate growth and anatomy. Additional examinations were done as necessary when pregnancy complications developed (eg, size greater than dates) or in high-risk pregnancies.
The study group consisted of pregnancies complicated by hydramnios after 20 weeks' gestation. Hydramnios was defined by objective measure of a single vertical pocket, with hydramnios diagnosed when the depth was greater than 8 cm; by measurement of a four-quadrant amniotic fluid index (AFI) as described by Phelan et al,14 with hydramnios diagnosed when AFI was 25 cm or greater; or by the sonographers' subjective impressions. Multiple gestations and oligohydramnios were excluded. Controls included all singleton pregnancies evaluated during the same period that had normal AF volume on ultrasound examination after 20 weeks' gestation.
When hydramnios was diagnosed, the women had repeat indirect Coombs testing, consideration of repeat 50-g glucose tolerance testing (GTT), comprehensive ultrasound examination, and consideration of amniocentesis, depending on ultrasound findings and physician discretion. Pregnancy outcomes were cross-referenced for all cases and controls to our Obstetric Automated Records System, which included complete obstetric and medical records for all women in our system. Standard prenatal care for women in our system includes a 1-hour 50-g GTT at 26–28 weeks. Women with results exceeding 135 mg/dL have a 3-hour 100-g test, with gestational diabetes defined according to ACOG criteria.15 Neonatal records were reviewed in cases of hydramnios to confirm or deny major structural abnormalities.
Cases and controls were compared for structural fetal anomalies, aneuploidy, fetal growth restriction (FGR), cesarean rate, frequency of diabetes, and perinatal mortality. Women in the hydramnios group were then categorized according to diabetes status, and outcomes were compared. Differences were evaluated using Fisher exact test or χ2 test as appropriate. Continuous variables were analyzed using the Student t test. Confounding variables known to influence perinatal outcome were analyzed in a multiple logistic regression model to assess the influence on perinatal outcome. Statistical significance was defined as P < .05.
Ultrasound examinations were done on 40,065 women from 1986 through 1996 for whom delivery and outcome information was available. After exclusions, 370 women with hydramnios and 36,426 controls with normal AF were identified. The incidence of hydramnios was 1%. Table 1 displays the demographic data for the control and study groups.
Table 2 shows the pregnancy outcomes of women with hydramnios and controls. Perinatal mortality was more than three times higher in women with hydramnios than controls; major structural anomalies were more than 25 times more frequent in the hydramnios group. Of the 18 perinatal losses in the hydramnios group, 11 were stillbirths and seven were neonatal deaths.
Major structural anomalies were detected in the infants of 31 women with hydramnios. Central nervous system (CNS) anomalies including hydrocephaly, anencephaly, holoprosencephaly, encephalocele, and myelomeningocele were diagnosed in 14 pregnancies. Gastrointestinal anomalies including omphalocele, duodenal atresia, bowel obstruction, and diaphragmatic hernia were diagnosed in ten pregnancies. Four pregnancies had cardiovascular malformations including structural heart defects, hydrops, and pulmonary edema. The three remaining pregnancies had multiorgan system malformations including the CNS, gastrointestinal system, and genitourinary system.
Nine of 18 perinatal deaths in pregnancies with hydramnios had associated anomalies that likely contributed: three cases of CNS anomalies including hydrocephaly, holoprosencephaly, and anencephaly; two with cardiovascular malformations including structural heart defects and hydrops; and four with multisystem anomalies including diaphragmatic hernia, hydrocephaly, genitourinary malformations, and structural cardiac defects. In the remaining nine deaths without anomalies, there were multiple causes including umbilical cord accident (three cases), severe preeclampsia (one case), intrauterine infection (one case), abruption (one case), Pena Shokeir phenotype (one case), intrapartum head entrapment (one case), and neonatal Escherichia coli sepsis (one case). Excluding cases with anomalies, the perinatal mortality rate of women with hydramnios was still more than twice that of controls (37 per 1000 versus 14 per 1000; P < .001).
Women with hydramnios had a significantly higher cesarean rate, incidence of diabetes, and average birth weight. Although the average gestational age at delivery was lower in women with hydramnios than in controls, it is probably not clinically significant. However, in pregnancies with perinatal mortality, the average gestational age at delivery was 34.5 weeks in hydramnios cases compared with 27.9 weeks in controls. Diagnosis of aneuploidy in either group was rare, and no statistically significant difference in incidence was found. Review of the indications for cesarean deliveries found a fivefold increased incidence of umbilical cord prolapse, a 14-fold higher incidence of fetal anomalies, and a sevenfold higher incidence of suspected fetal macrosomia in women with hydramnios compared with controls. There were no increases in the frequencies of abruption, placenta previa, nonreassuring fetal heart rate (FHR) tracing, malpresentation, or failure to progress.
A multiple logistic regression analysis was done to control for potential confounding variables known to influence perinatal outcome, including maternal weight at first prenatal visit, tobacco use, race, male fetal gender, maternal age, parity, and hypertension. Table 3 shows the results of this analysis. Women with hydramnios still had increased perinatal mortality and anomaly rates; however, the lower incidence of FGR on initial analysis was not statistically significant once these variables were included.
We then evaluated the association of diabetes and hydramnios. The incidence of hydramnios was significantly higher in diabetic women (5.8%) than in nondiabetic women (0.84%) (P < .001). Among diabetic women with hydramnios, half had gestational diabetes and half had diabetes that predated pregnancy. Women with gestational diabetes had a 3.9% incidence of hydramnios, compared with 11.6% among women with nongestational diabetes (P < .001). The analysis of outcome data in nondiabetic (n = 299) versus diabetic (n = 71) women with hydramnios is displayed in Table 4. None of the diabetic women had perinatal losses or anomalies, but the nondiabetic women had excess perinatal mortality and anomaly rates. No significant differences were noted in the rate of aneuploidy or FGR in these groups.
The rate of cesarean delivery in diabetic women was nearly twice the rate in nondiabetic women and was more than fourfold higher than in controls. This did not appear to be solely dependent on birth weight because the average birth weight was lower in diabetic women than in euglycemic women with hydramnios. The mean gestational age at delivery was similar (37.3 weeks in diabetic women and 37.9 weeks in nondiabetic women). Review of the indications for cesarean deliveries found a fourfold increase in nonreassuring FHR tracings and suspected fetal macrosomia was 12 times higher. There were no differences in rates of abruption, placenta previa, umbilical cord prolapse, or fetal malpresentation.
In uncomplicated pregnancies, AF volume peaks at 24–26 weeks and remains relatively stable until 38 weeks, when it begins to decline slowly; at 40 weeks it declines at a rate of approximately 33% per week.3,17 Because of the presumed association between AF disturbance and perinatal morbidity and mortality, noninvasive evaluation of AF volume has become routine and has steadily evolved since the advent of ultrasonography. Subjective operator impression, deepest vertical pocket measurement, and measurement of a four-quadrant AFI are all accepted estimation techniques.
Although a great deal is known about the causes and prognosis of low AF volume, much less is known about hydramnios. The cause of hydramnios remains idiopathic in most cases, and the precise incidence of associated maternal and fetal disease is unknown. Despite the relative lack of data, many clinicians have viewed hydramnios as a prognostic indicator of increased risk of pregnancy complications and have recommended extensive evaluation of these pregnancies, often including multiple comprehensive ultrasound examinations, repeat diabetes screening, and amniocentesis for fetal karyotype analysis.
Landy et al10 reviewed 158 cases of hydramnios, of which 86 could be evaluated in detail. In 59 cases, no underlying factors were identified and all sonographic fetal examinations were normal. At delivery, one infant had trisomy 18 and eight had structural abnormalities not previously recognized. Because of the 14% false-negative rate of ultrasound detection of anomalies and the frequent association of some anomalies with trisomies, these authors recommended amniocentesis as an adjunct to evaluation.
Barnhard et al4 evaluated 49 women with hydramnios. Two infants (4.1%) had chromosomal abnormalities and six (12.2%) had structural abnormalities, compared with 0.12% and 1.8%, respectively, in the control group. They noted FGR in both fetuses with chromosomal abnormalities. Based on these findings, they recommended that comprehensive ultrasound examinations be done in all cases of hydramnios, but that karyotype analysis could be reserved for those with FGR or anomalies known to be associated with aneuploidy. In a review of 102 cases of hydramnios at the Mayo Clinic, Hill et al1 discovered 13 anomalous fetuses and 13 perinatal deaths, for an overall perinatal mortality rate of 128 per 1000 births. Even when corrected for lethal anomalies, the mortality rate was 59 per 1000, four times higher than in their general population. The study group included a large percentage of referrals, so the results might have been biased toward the most severe cases or those with anomalies.
We were able to access all ultrasound examinations done at one institution over a 10-year period, enabling us to generate incidence rates of various complications in our population and to establish one of the largest series of hydramnios cases. Hydramnios occurred in 1% of more than 36,000 ultrasound examinations, over which time we have used AFI measurement, deepest vertical pocket measurement, and subjective sonographer opinion to estimate AF volume. Although sequential use of these methods might lead to a trend toward higher diagnosis rates with one over another, we have used all three in tandem and have not noted any increase or decrease in the frequency of diagnosis with one over another. Owen et al reported our institution's experience with sonographic assessment of AF volume, noting excellent correlation between objective AFI measurement and subjective sonographer impression (Owen J, Davis RO, Brumfield CG, DuBard M. The correlation between the amniotic fluid index and a subjective estimate of the amniotic fluid volume in 31,000 sonographic evaluations. Am J Obstet Gynecol 1995;172. Society of Perinatal Obstetricians, Abstract 64).
Our demographic analysis showed that in our population, hydramnios was more likely in older gravidas, multiparas, smokers, women with chronic hypertension, and whites. Even after controlling for these variables, the increase in perinatal mortality and fetal anomalies persisted in pregnancies complicated by hydramnios. We were unable to identify a single etiologic factor for the perinatal mortality, although we ruled out two potentially contributing factors. Prematurity alone did not account for these losses because the mean gestational age at delivery for women with hydramnios and perinatal mortality was 34.5 weeks. Fetal anomalies were not solely responsible either because the perinatal death rate was still more than twice that of controls after all cases with anomalies were excluded.
We did not find a significantly increased incidence of aneuploidy. The rate of aneuploidy in our subjects and controls was low, at one in 370 and one in 3643, respectively. Although not all women with hydramnios had genetic amniocentesis, all newborns were evaluated by skilled pediatricians and neonatologists, in consultation as necessary with pediatric geneticists, making it unlikely that any cases of aneuploidy were missed. Because the amniocentesis-associated pregnancy loss rate is one in 200,18 we cannot recommend routine fetal karyotype analysis for all cases of hydramnios. It might be warranted when ultrasound examination shows structural abnormalities or severe FGR.
We found it interesting that there were no cases of perinatal mortality, anomalies, or aneuploidy in the 71 pregnancies complicated by diabetes and hydramnios, whereas nondiabetic women with hydramnios had a perinatal mortality rate of 60 per 1000 births and a 10.4% incidence of fetal anomalies. This difference might indicate that pregnant women with diabetes benefit from close antepartum surveillance that may identify anomalies early (allowing consideration of pregnancy termination) and prevent adverse outcomes in ongoing pregnancies. Our policy is to perform targeted ultrasound examinations at 18–22 weeks' gestation and begin antepartum fetal surveillance by 34 weeks' gestation, with amniocentesis for lung maturity analysis by 37–38 weeks in all insulin-dependent diabetic women. This management allows intervention before any evidence of fetal compromise or death.
Diabetic women with hydramnios had a cesarean rate of 70.4%, compared with 41.5% for nondiabetic women. Although the cesarean rate for diabetic women was extraordinarily high, both rates were dramatically increased over the control rate of 16%. Suspected macrosomia in part explains the high rate of abdominal delivery for diabetic women with hydramnios. It has been our policy to consider cesarean for macrosomic infants of diabetic mothers (defined as a sonographically derived estimated fetal weight greater than 4400 g) because of the increased risk of shoulder dystocia; however, this policy increased our cesarean rate by less than 1%. It seems likely that various other factors have an effect. The cause of the higher frequency of nonreassuring FHR tracings is not apparent and could be investigated in future analyses.
Data on long-term infant outcomes were not available; therefore, mortality after day 30 of life was not considered. A MEDLINE search covering 1966–1998 using the key words “hydramnios,” “long-term outcome,” “mortality,” and “childhood outcome” failed to identify any articles on this issue, which may be an area worthy of future study. Whether hydramnios persisted or resolved might also influence outcome. We did not examine it in this study, and a MEDLINE search using the key words “hydramnios,” “persistence,” and “resolution” identified fewer than five manuscripts on this topic. This also may be an issue worthy of further study.
Most of the current recommendations for the evaluation of hydramnios are reasonable. If hydramnios is diagnosed, a thorough evaluation for fetal or maternal factors is indicated, which at minimum should include comprehensive ultrasound examination and diabetes screen. If the comprehensive ultrasound examination does not identify fetal disease, the low risk of fetal aneuploidy does not justify genetic amniocentesis. The increase in the perinatal mortality rate might justify closer antenatal surveillance.
1. Hill LM, Breckle R, Thomas ML, Fries JK. Polyhydramnios: Ultrasonically detected prevalence and neonatal outcome. Obstet Gynecol 1987;69:21–5.
2. Cardwell MS. Polyhydramnios: A review. Obstet Gynecol Surg 1987;42:612–7.
3. Phelan JP, Martin GI. Polyhydramnios: Fetal and neonatal implications. Clin Perinatol 1989;16:987–94.
4. Barnhard Y, Bar-Hava I, Divon M. Is hydramnios in an ultrasonographically normal fetus an indication for genetic evaluation? Am J Obstet Gynecol 1995;173:1523–7.
5. Brady K, Polzin WJ, Kopelman JN, Read JA. Risk of chromosomal abnormalities in patients with idiopathic polyhydramnios. Obstet Gynecol 1992;79:234–8.
6. Desmedt EJ, Henry OA, Beischer NA. Polyhydramnios and associated maternal and fetal complications in singleton pregnancies. Br J Obstet Gynaecol 1990;97:1115–22.
7. Glantz JC, Abramowicz JS, Sherer DM. Significance of idiopathic midtrimester polyhydramnios. Am J Perinatol 1994;11:305–8.
8. Golan A, Wolman I, Langer R, David MP. Fetal malformations associated with chronic polyhydramnios in singleton pregnancies. Eur J Obstet Gynecol Reprod Biol 1992;47:185–8.
9. Golan A, Wolman I, Sagi J, Yovel I, David MP. Persistence of polyhydramnios during pregnancy—its significance and correlation with maternal and fetal complications. Gynecol Obstet Invest 1994;37:18–20.
10. Landy HJ, Isada NB, Larsen JW. Genetic implications of idiopathic hydramnios. Am J Obstet Gynecol 1987;157:114–7.
11. Stoll CG, Alembik Y, Dott B. Study of 156 cases of polyhydramnios and congenital malformations in a series of 118,265 consecutive births. Am J Obstet Gynecol 1991;165:586–90.
12. Varma TR, Bateman S, Patel RH, Chamberlain GVP, Pillai U. The relationship of increased amniotic fluid volume to perinatal outcome. Int J Gynaecol Obstet 1988;27:327–33.
13. Zahn CM, Hankins GDV, Yeomans ER. Karyotypic abnormalities and hydramnios: Role of amniocentesis. J Reprod Med 1993;38:599–602.
14. Phelan JP, Smith CV, Broussard P, Small M. Amniotic fluid volume assessment with the four-quadrant techniques at 36–42 weeks' gestation. J Reprod Med 1987;32:540–2.
15. American College of Obstetricians and Gynecologists. Diabetes and pregnancy. ACOG technical bulletin no. 200. Washington DC: American College of Obstetricians and Gynecologists, 1994.
16. Ben-Chetrit A, Hochner-Celnikier D, Ron M, Yagel S. Hydramnios in third trimester of pregnancy: A change in the distribution of accompanying fetal anomalies as a result of early ultrasonographic prenatal diagnosis. Am J Obstet Gynecol 1990;162:1344–5.
17. Brace R. Amniotic fluid dynamics. In: Creasy R, Resnik R, eds. Maternal-fetal medicine: Principles and practice. 3rd ed. Philadelphia: WB Saunders, 1994:106–7.
18. Elias S, Simpson JL. Amniocentesis. In: Milunsky A, ed. Genetic disorders and the fetus. 3rd ed. Baltimore: The Johns Hopkins University Press, 1992:33–57.