Hydramnios, an excess of amniotic fluid, complicates approximately 1% of pregnancies, with a fairly constant prevalence among different populations.1–4 More than 30 years ago, Queenan and Gadow reported that 20% of cases of hydramnios were accompanied by a congenital fetal anomaly.5 Other etiologies, such as rhesus isoimmunization, poorly controlled diabetes, and multiple gestation, are readily detected through routine antibody screen, glucose tolerance testing, and sonography, respectively. For many clinicians, the question becomes whether the hydramnios is complicated by an abnormality or whether it is simply idiopathic, often the most common finding.5 From neonatal descriptions of anomalies associated with hydramnios, it is clear that a variety of organ system malformations can lead to this complication, some relatively subtle.3,6 There have even been cases of hydramnios and underlying aneuploidy in which no major malformation was noted in the neonatal period.7,8 Thus, one cannot expect to detect all anomalies in the setting of hydramnios, and counseling is necessarily somewhat limited.
The anomaly rate in pregnancies with hydramnios will depend on the population studied—higher in pregnancies referred with complications and lower in a general (nonreferred) population.1 Over the past decade, reported anomaly rates have ranged from 8% to 45%.2,3,9–11 The likelihood of aneuploidy has similarly varied, from 0.4% to more than 10%.2,3,8,11,12 Though pregnancies with hydramnios are routinely evaluated with targeted sonography, little information is available to assist with counseling when the ultrasound evaluation is normal. Our objective was to compare neonatal ascertainment of anomalies with their sonographic detection in a large, nonreferred cohort with hydramnios. The goal was to estimate the residual risk of a major anomaly or aneuploidy, according to degree of hydramnios present, if no abnormality was detected sonographically.
MATERIALS AND METHODS
This was a retrospective cohort study of singleton pregnancies with antepartum diagnosis of hydramnios between February 1, 1991, and September 30, 1997. Pregnancies were ascertained using a computerized ultrasound database, based on gestational age of at least 20 weeks and amniotic fluid index of 25 cm or more.10 After diagnosis of hydramnios, targeted sonography was performed in all cases. Hydramnios was categorized as mild, moderate, or severe based on amniotic fluid index of 25.0–29.9 cm, 30.0–34.9 cm, or 35.0 cm or more, respectively.9 Sonography was typically repeated every 4 weeks. Pregnancies were grouped according to worst degree of hydramnios to facilitate analysis.
Neonatal information was obtained using our computerized obstetric database, which contains selected pregnancy outcomes for all women who are delivered at our hospital. Though all singleton pregnancies with hydramnios were included, only those with fetal or neonatal anomalies were the objective of this analysis. Evaluation for major congenital anomalies is routinely carried out in the immediate neonatal period by attending faculty pediatricians, and karyotype analysis is performed at their discretion. Obstetric research nurses follow all infants suspected of having anomalies or aneuploidy on a daily basis and record this information in the obstetric database. The focus of the study was live-born singleton infants of 25 weeks' gestation or more, to permit verification of anomalies diagnosed in the immediate neonatal period. Information about fetal deaths (birth weight greater than 500 g) is also included in our database; however, malformation data may not be complete because of postmortem changes. For this reason, fetal deaths were analyzed separately, and malformations were considered present if detected antepartum, at delivery, or upon necropsy.
For study purposes, an anomalous infant was considered detected if at least one major anomaly present in the neonatal period had been noted during antenatal sonography. An anomaly was considered detected if at least one major abnormality of that organ system had been noted sonographically. The prevalence of anomalies and their ultrasound detection were analyzed according to degree of hydramnios and organ system(s) involved. An anomaly was considered major if potentially life threatening or requiring medical or surgical treatment in the immediate neonatal period. We included any cardiac septal defects that came to medical attention before hospital discharge in this category.
Aneuploidy was analyzed separately, according to presence or absence of major anomalies. In this way, we were able to evaluate the likelihood of either a major anomaly or aneuploidy, according to degree of hydramnios, even if no anomaly had been detected sonographically. Pregnancies complicated by diabetes were also analyzed separately. Gestational diabetes was diagnosed based on the presence of two or more abnormal values on a standard 100-g, 3-hour oral glucose tolerance test, using the thresholds of the National Diabetes Data Group.13 Statistical analyses were performed using χ2 test, Student-Newman-Keuls test, analysis of variance, and Mantel-Haenszel χ2 for trend (SAS system 8.0, SAS Institute, Cary, NC). P < .05 were judged statistically significant.
There were 93,332 singleton pregnancies delivered during the study period. Hydramnios was diagnosed during antepartum sonography in 708 pregnancies, and 672 (95%) were delivered at our institution, representing 0.7% of deliveries during the study period. Hydramnios was categorized as mild in 446 pregnancies (66%), moderate in 145 (22%), and severe in 81 pregnancies (12%). Demographic characteristics are presented in Table 1, stratified by the degree of hydramnios. No significant differences in maternal age, ethnicity, weight at delivery, or presence of diabetes were noted between the mild, moderate, and severe hydramnios groups. Pregnancies with severe hydramnios were delivered at an earlier gestational age and had correspondingly lower birth weights. The number of large for gestational age infants, using a nomogram from our population, did not differ between groups.14
Seventy-seven infants (11%) were noted to have one or more major anomalies in the immediate neonatal period. Shown in Table 2 is the number of anomalous infants and their sonographic detection, stratified by the degree of hydramnios. The prevalence of anomalous infants was higher in pregnancies with more amnioticfluid, exceeding 30% in cases of severe hydramnios (P < .001). The sonographic detection rate was 79% overall and did not differ significantly according to the degree of hydramnios (P = .4). The likelihood of a major anomaly, with hydramnios and a normal sonographic evaluation, was 1% in pregnancies with mild hydramnios, 2% with moderate hydramnios, and 11% with severe hydramnios (P for trend < .001). Thus, as hydramnios increased, the greater prevalence of anomalies coupled with a stable anomaly detection rate led to greater residual anomaly risk.
Ninety anomalies were identified in 77 infants. Figure 1 is a graph of neonatal anomaly prevalence and sonographic anomaly detection, shown according to organ system involved. At least 90% of anomalies of the central nervous system, thorax, skeleton, kidney, and ventral wall were identified antenatally, whereas the detection rate for cardiac malformations, gastrointestinal anomalies, and craniofacial defects was considerably lower (Figure 1). Table 3 lists organ system anomaly detection according to degree of hydramnios. Presented in Table 4 is a list of the specific anomalies not detected antenatally. The most frequent anomalies were cardiac septal defects, cleft palate, imperforate anus, and tracheoesophageal fistula. The ten infants with anomalies of more than one organ system were all detected.
Thirteen infants (2%, 95% confidence interval [CI] 1, 3) were aneuploid, and eight of these 13 infants were found to be anomalous in the neonatal period. In each of the 13 cases, hydramnios had been diagnosed after 25 weeks' gestation. Figure 2 is a chart showing which of the aneuploid infants had one or more structural anomalies present in the neonatal period, and how many of these anomalies were detected with antenatal ultrasound. Overall, the prevalence of aneuploidy among anomalous infants with hydramnios was 10% (95% CI 5, 19) with no significant difference according to degree of hydramnios (12% with mild, 11% with moderate, and 8% with marked hydramnios). Six aneuploid infants (46%) had no anomaly detected antenatally, five of whom were also found to have no major anomaly in the neonatal period, even after the aneuploidy had been characterized. These five infants included four with Trisomy 21 and one with 9p deletion. The remaining infant had Trisomy 21 with Tetralogy of Fallot; interestingly, cardiac outflow tracts had not been able to be visualized sonographically. These six aneuploid infants conferred a residual aneuploidy risk of 1% (95% CI 0.4, 2) in the setting of a normal sonogram.
Given the established associations between diabetes and hydramnios and between pregestational diabetes and fetal malformations, the cohort of women with diabetes was also evaluated. Of 672 pregnancies with hydramnios, 49 (7%) were also complicated by diabetes, including 13 (2%) with pregestational diabetes, 13 (2%) with gestational insulin-treated diabetes (Type A2), and 23 (3%) with gestational diet-controlled diabetes (Type A1). Anomalous infants were present in six diabetic pregnancies with hydramnios (12%), similar to the overall anomaly rate (11%). Three of these six anomalous infants were born to women with pregestational diabetes and three with gestational diabetes.
Fetal deaths were evaluated separately. There were 25 fetal deaths in pregnancies with the antenatal diagnosis of hydramnios. Anomalies were present in 15 cases (60%), in seven of 14 with mild hydramnios, four of four with moderate hydramnios, and four of seven with severe hydramnios. Karyotype data was not available for the fetal deaths. If fetal deaths are included with the overall group (697 pregnancies with hydramnios), the fetal death rate was 4% (95% CI 2, 5). The overall rate of a major anomaly or aneuploidy in a live-born or stillborn infant in pregnancies complicated by hydramnios was 14% (95% CI 11, 17) and was 9% for cases of mild hydramnios, 17% for moderate hydramnios, and 33% for severe hydramnios.
Using a MEDLINE search for the terms “hydramnios” and “fetal anomaly” from 1966 through 2001, this series is the largest population-based analysis of fetal anomalies in pregnancies with hydramnios to date. Whereas prior studies have evaluated anomaly prevalence and risk factors for adverse outcome,2,9,10 our goal was specifically to improve counseling for women with hydramnios in whom no anomaly is detected sonographically. The prevalence of anomalies in our cohort (11%) is fairly comparable with other large series and confirms reports of greater anomaly risk with worsening hydramnios.2–4,9–11 Our detection of anomalous infants (79%) was also comparable with other series.3,11 Golan et al reported detection of 61% of 28 anomalous infants in their series of 197 pregnancies with hydramnios.3 Zahn et al detected 73% of 15 anomalous infants in pregnancies with hydramnios.11 When moderate or severe hydramnios is encountered, the suspicion for anomalies is quite high, and we were somewhat surprised that the detection of anomalous infants did not vary according to degree of hydramnios. However, because the overall prevalence of anomalies does increase with the degree of hydramnios, the residual risk of anomaly similarly increases. From our data, we estimate that if the sonographic evaluation is normal, the likelihood of a major anomaly in the setting of mild, moderate, and severe hydramnios is approximately 1%, 2%, and 11%, respectively.
Ultrasound technology continues to improve, and any analysis large enough to establish meaningful anomaly detection rates will be hampered by the improvement in technology over the study period, in our case the 1990s.11 Prenatal series have understandably tended to focus on anomalies most amenable to sonographic detection. For example, one group that reported detection of all anomalies (in a referred population) included no cases of imperforate anus, tracheoesophageal fistula, facial clefts, or atrial septal defects.10 During the period of our study, reliable detection of a variety of major organ system anomalies in this setting has been described, with findings similar to ours. Golan et al reported that all major anomalies of the central nervous system, kidney, diaphragm, and ventral wall were identified prenatally, whereas facial clefts, cardiac septal defects, and imperforate anus were not.3 A reason why anomaly detection in our series did not vary according to degree of hydramnios may be that certain anomalies are challenging if not impossible to reliably image, regardless of one's index of suspicion.
Other limitations of this retrospective review should be mentioned. We did not stratify our findings according to fetal size. This was because our interest was how well anomalies are detected when varying degrees of hydramnios are present, rather than anomaly prevalence in the setting of other risk factors. The combination of fetal growth restriction and hydramnios is associated with both malformations and aneuploidy, and amniocentesis has been recommended in such cases.6,9 Some cases of hydramnios are also related to diabetes. However, Lazebnik and Many found that the anomaly rate was not significantly different between diabetic pregnancies with hydramnios and nondiabetic pregnancies with hydramnios, similar to our results.9
Should fetal karyotype analysis be offered when hydramnios is present? In the setting of a sonographic anomaly, we found an aneuploidy prevalence of 10%, comparable with the 9.6% prevalence reported by Stoll et al.7 When no anomaly is detected, the answer is less clear. Biggio et al reported that in such cases only one pregnancy of 370 contained an aneuploid infant.2 Though we encountered 13 aneuploid infants in 672 pregnancies (2%), if no anomaly was detected, the aneuploidy rate was only 1%. A caveat to this finding is that we did not have karyotype information available for the stillborn infants. Several groups in addition to ours have reported cases of hydramnios accompanied by aneuploidy in which the infant did not have a major malformation. Stoll et al described seven such infants, and Brady et al reported three others.7,8 We do not have information about maternal serum screening for aneuploidy or sonographic “soft signs” for Down syndrome in these cases, though such information would certainly be interesting. Although fetal karyotype may be offered in the setting of hydramnios, we suggest informing the woman that if no anomaly is detected sonographically, the aneuploidy risk is likely 1% or less. Further, in our series, none of the pregnancies with isolated aneuploidy developed hydramnios before 26 weeks' gestation, and in no case was the aneuploidy lethal in the immediate neonatal period.
Based on our findings, we suggest counseling women with mild or moderate hydramnios (amniotic fluid index below 35 cm) that targeted sonography may reduce their risk of an anomalous infant to 2% or less, comparable with the general population risk. In contrast, those with severe hydramnios remain at considerable risk (10% or more), and their anomalies may be less amenable to sonographic detection.
1. Cunningham FG, Gant NF, Leveno KJ, Gilstrap LG III, Hauth JC, Wenstrom KD, eds. Williams obstetrics. 21st ed. New York: McGraw Hill, 2001:817–9.
2. Biggio JR, Wenstrom KD, Dubard MB, Cliver SP. Hydramnios prediction of adverse perinatal outcome. Obstet Gynecol 1999;94:773–7.
3. Golan A, Wolman I, Langer R, David MP. Fetal malformations associated with chronic polyhydramnios in singleton pregnancies. Eur J Obstet Gynecol 1992;47:185–8.
4. Hill LM, Breckle R, Thomas ML, Fries JK. Polyhydramnios: Ultrasonically detected prevalence and neonatal outcome. Obstet Gynecol 1987;69:21–5.
5. Queenan JT, Gadow EC. Polyhydramnios: Chronic versus acute. Am J Obstet Gynecol 1970;108:349–55.
6. Sickler GK, Nyberg DA, Sohaey R, Luthy DA. Polyhydramnios and fetal intrauterine growth restriction: Ominous combination. J Ultrasound Med 1997;16:609–14.
7. Stoll CG, Alembik Y, Dott B. Study of 156 cases of polyhydramnios and congenital malformations in a series of 118,265 consecutive live births. Am J Obstet Gynecol 1991;165:586–90.
8. Brady K, Polzin WJ, Kopelman JN, Read JA. Risk of chromosomal abnormalities in patients with idiopathic hydramnios. Obstet Gynecol 1992;79:234–8.
9. Lazebnik N, Many A. The severity of polyhydramnios, estimated fetal weight and preterm delivery are independent risk factors for the presence of congenital malformations. Gynecol Obstet Invest 1999;48:28–32.
10. Damato N, Filly RA, Goldstein RB, Callen PW, Goldberg J, Globus M. Frequency of fetal anomalies in sonographically detected polyhydramnios. J Ultrasound Med 1993; 12:11–5.
11. Zahn CM, Hankins GDV, Yeomans ER. Karyotypic abnormalities and hydramnios. J Reprod Med 1993;38:599–602.
12. Phelan JP, Ahn MO, Smith CV, Rutherford SE, Anderson E. Amniotic fluid index measurements during pregnancy. J Reprod Med 1987;32:601–5.
13. National Diabetes Data Group. Classification and diagnosis of diabetes mellitus and other categories of glucose. Diabetes 1979;28:1039–57.
14. McIntire DD, Bloom SL, Casey BM, Leveno KJ. Birth-weight in relation to morbidity and mortality among newborn infants. N Engl J Med 1999;340:1234–8.