Accuracy of fetal echocardiography in the routine detection of congenital heart disease among unselected and low risk populations: a systematic review

Authors


Abstract

Objective  To determine among unselected and low risk populations the accuracy with which fetal echocardiography during the second trimester detects congenital heart disease.

Design  A systematic review of studies that assess the accuracy of fetal echocardiography.

Setting  District General Hospital and Tertiary referral centres.

Population  Women during the second trimester attending for ultrasound assessment.

Methods  General bibliographic databases (e.g. MEDLINE, EMBASE) and specialist computerised databases (e.g. Cochrane Library, National Research Register), grey literature, manual searching of reference lists of primary and review articles and personal contact with experts were used to identify studies. Studies were included if fetal echocardiography among unselected or low risk pregnant women was compared against a postnatal reference standard. Data were extracted on quality, study design and characteristics, and accuracy data to construct 2 × 2 tables. Data were synthesised qualitatively, and sensitivity and specificity with 95% confidence intervals were calculated.

Main outcome measure  Sensitivity and specificity.

Results  Five primary studies met the inclusion criteria and comprised 60,901 subjects. One study assessed the accuracy of fetal echocardiography among an unselected population and four studies among low risk populations. All eligible studies found that fetal echocardiography helped to correctly diagnose normal fetus among unselected and low risk populations. Correct diagnosis of babies for congenital heart defects was higher among the unselected population (85%) than among the low risk populations (range from 35% to 86%); however, the potential for ascertainment bias and the choice of reference standard limits the validity of this finding. The variation in sensitivity across studies was not explainable by clinical factors such as scanning regime, operator skill and equipment.

Conclusions  The evidence from this review about the accuracy of fetal echocardiography does not lend support to its routine use among unselected and low risk populations during the second trimester to detect congenital heart disease.

Introduction

Congenital heart disease, malformation of the heart or great arteries, is the most common form of major birth defect, being six times more common than chromosomal abnormalities and four times more common than neural tube defects.1 Approximately 8 out of 1000 babies are born in the UK each year with congenital heart disease, nearly 1% of live births.2–4 Congenital heart disease accounts for approximately 40% of perinatal deaths due to congenital anomalies,5 more than 20% of deaths in the first month of life6 and the majority of deaths from congenital defects in childhood.5,7,8 These defects therefore place a significant economic burden on the NHS and society and the psychological wellbeing of affected families.

Antenatal diagnosis of congenital heart disease allows planning and delivery of timely and appropriate neonatal care.7–9 This is increasingly important with accumulating evidence that early diagnosis may have the potential to save babies' lives.1,10–15 Early knowledge of congenital heart disease also allows further monitoring, testing for known associated non-cardiac structural and chromosomal anomalies16–19 and parental counselling about pregnancy management options including termination.

Various methods of antenatal ultrasound assessment of a baby's heart are currently available. The ‘four-chamber view’ is the most basic assessment. This allows a general examination of the main structure of the heart and the atrioventricular junctions. ‘Basic fetal echocardiography’ is a more extensive antenatal ultrasound examination of the heart and its associated structures, through additional assessment of the ventricular outflow tracts. ‘Extended fetal echocardiography’ is also available. During this examination, two-dimensional scanning of the heart and its associated structures is supplemented by spectral and colour-flow Doppler, and on occasion, M-mode scanning to assess blood flow within the heart. Fetal echocardiography has been used as a diagnostic tool to identify congenital heart disease in high risk groups for some years,20,21 and has been found to have sensitivities ranging from 60% to 100%.10,22–24 Its routine use in unselected obstetric populations including those at low risk is uncommon and more controversial.19,25–27 Against this background, a systematic review was conducted to determine the accuracy of fetal echocardiography to detect congenital heart disease to inform whether it should be included in a routine 20-week anomaly scan for an unselected obstetric population including newborn at low risk for developing cardiac defects.

Methods

To locate suitable studies, electronic searches of databases were performed from 1990 to July 2002. This included general bibliographic databases: MEDLINE, EMBASE, Science Citation Index—Expanded. Specialist computerised databases were also searched: Cochrane Library, National Research Register, Medical Research Council Research Register, Health Technology Assessment projects Database, National Co-ordinating Centre for Health Technology Assessment, Health Services/Technology Assessment Text, Turning Research into Practice Database, System for Information on Grey Literature in Europe and PsycINFO. The electronic search strategies were developed using those recommended for identifying primary studies of diagnostic tests combined with subject area terms.28 Reports, discussion papers, dissertations and theses, and conference proceedings were searched for grey literature. Reference lists of primary articles and relevant reviews were also searched and personal contact made with experts.

Eligible studies included those that assessed the accuracy of fetal echocardiography against a postnatal reference standard in a low risk population of second trimester women with none of the risk features that typically indicate the need for fetal echocardiography29,30 or in general obstetric populations of second trimester women for which there was no selection as to who should have the scan. Studies were excluded if only high risk pregnant women were included, not all patients received fetal echocardiography, the ultrasound technique was not fully described (including scanning regime, gestation at scanning and equipment used) and there was insufficient data to construct a 2 × 2 table to calculate the accuracy of fetal echocardiography. Along with non-relevant publications, editorials, letters and commentaries, reviews were excluded following extraction of eligible primary studies.

Eligible studies were assessed for their methodological quality using a checklist developed from a literature review.31–33 The criteria included in the checklist addressed bias concerning the selection of women into a study, the appropriate choice and application of the reference standard, and independence (or blinding) in the interpretation of scan results.

From eligible studies, the following data were extracted by the first author who cross-checked with the second author when necessary: quality, study design and study characteristics (e.g. scanning regime, gestation age, and operator skills), and accuracy data to construct 2 × 2 tables.

The results of the studies were synthesised qualitatively. Sensitivity and specificity of fetal echocardiography were also calculated with 95% confidence intervals (CI) using the Exact (Clopper–Pearson) method. A correction factor of 0.5 was added to all four cells when a study had a zero in a cell. The data were not pooled to produce summary estimates of accuracy because there were too few studies with substantial variation in characteristics.

Results

Figure 1 summarises the process of literature identification and selection. In total, five studies were included. Across the five studies there were 60,901 subjects. One study assessed the accuracy of fetal echocardiography among an unselected population and four studies among a low risk population.

Figure 1.

Study selection process for systematic review of fetal echocardiography.

A variety of scanning regimes, operator skills and severity of congenital heart disease were found and these were classified into hierarchies as defined in Table 1. Table 2 shows variation in the characteristics of the eligible studies in terms of, for example, the scanning regime, postnatal reference standard, level of operator skills and prevalence of congenital heart disease per 1000 pregnancies. All the studies were observational in design (four prospective and one retrospective) and scored less than 50% on application of the quality criteria checklist. A major contributor to presence of bias concerned the choice and application of the reference standard and blinding. Ascertainment bias also seemed to be present as prevalence was higher in some low risk populations than in the unselected population, suggesting bias in the selection of women for screening. In addition, three of the studies' postnatal follow up period were at birth. Given that only half of all congenital heart diseases are diagnosed within the first week, these studies may produce biased prenatal detection rates for congenital heart disease.

Table 1.  Hierarchy of scanning regime, level of operator skill and classification of congenital heart disease.
TypeScanning regimeStructures/functions assessed
  • *

    These are pragmatic groups based on those recently defined by Klein et al.35

BasicRoutine fetal echocardiographyfour-chamber view + outflow tracts
 
Basic/ExtendedRoutine followed by extended fetal echocardiography on suspicion of congenital heart diseasefour-chamber view + outflow tracts (+Doppler colour-flow mapping)
 
ExtendedRoutine extended fetal echocardiographyfour-chamber view + outflow tracts + Doppler colour-flow mapping
 
LevelOperator skills 
IExperienced trained specialists (e.g. paediatric cardiologists) 
 
IIExperienced trained non-specialists (e.g. ultrasonographers) 
 
IIIExperienced and inexperienced non-trained non-specialists (e.g. obstetric ultrasonographers, new ultrasonographers) 
 
Group*Classification of congenital heart disease 
MajorMajor structural congenital heart defects that have a high risk of unsuccessful repair (e.g. Tetralogy of Fallot) 
 
MinorMinor structural congenital heart defects, which as a general rule can be satisfactorily repaired (e.g. double-outlet right ventricle) 
 
Non-structural defects/arrhythmiasNon-structural congenital heart disease defects/arrhythmias (e.g. sinus bradycardia) 
Table 2.  Overview of study characteristics.
Author, Year,[Ref]Setting (study design)Gestation age in weeksScanning regimeLevel of operator skillsUltrasound transducer frequencyFollow-up durationStudy populationNo. of casesPrevalence of congenital heart disease per 1000 pregnanciesQuality score (%)
PrenatalPostnatal*
  • *

    Postnatal = postnatal reference standard and when in brackets was only used in suspected cases.

  • TRC = tertiary referral centre.

  • DGH = district general hospital.

  • §

    N-examination = neonatal examination.

Rustico et al., 1995[4] ItalyTRC (Prospective)20–22BasicN/paediatric examination, autopsy (neonatal echo, ECG)III5/3.5 MHz24 monthsLow risk70249.348
Anandakumar et al., 2002[2] SingaporeTRC (Retrospective)21–22Basic/ ExtendedN-examination§II5/3.5 MHz6 monthsUnselected39,8087.637
Hafner et al., 1998[3] AustriaDGH (Prospective)22 and 34Basic/ ExtendedN-examination§ (neonatal echo)IIAt birthLow risk654113.637
Achiron et al., 1992[1] IsraelTRC (Prospective)18–24Basic/ ExtendedN-examination,§ autopsy (neonatal echo)I5/3.5 MHzAt birthLow risk53474.338
Stümpflen et al., 1996[5] AustriaTRC (Prospective)18–28ExtendedN-examination,§ autopsy (diagnostic investigations)I3.5 MHzAt birthLow risk21817.840

Table 3 shows that for both unselected and low risk populations, fetal echocardiography had close to 100% specificity for correctly diagnosing babies without congenital heart disease.Table 4 shows that the sensitivity of fetal echocardiography to correctly diagnose babies for major, minor, non-structural and all defects was higher when performed among the unselected population. Variation in study characteristics and methodological limitations may influence accuracy so these potential confounding factors were considered when discussing this finding and the other results of the review.

Table 3.  Overview of specificity.
Author, Year[Ref]Scanning regimeStudy populationMajor defects % (95% CI)Minor defects % (95% CI)Non-structural defects/arrhythmias (95% CI)All defects % (95% CI)
Rustico et al., 19954BasicLow risk99.9 (99.9–100)99.9 (99.9–100)99.9 (99.8–99.9)
Anandakumar et al., 20022Basic/ExtendedUnselected100 (99.9–100)99.9 (99.9–99.9)99.9 (99.9–99.9)99.9 (99.9–99.9)
Hafner et al., 19983 Low risk99.9 (99.9–100)99.9 (99.9–100)99.9 (99.9–100)99.6 (99.5–99.8)
Achiron et al., 19921 Low risk99.9 (99.9–100)99.9 (99.9–100)99.9 (99.9–100)99.9 (99.9–100)
Stümpflen et al., 19965ExtendedLow risk99.9 (99.8–100)
Table 4.  Overview of sensitivity.
Author, Year[Ref]Scanning regimeStudy populationMajor defects % (95% CI)Minor defects % (95% CI)Non-structural defects/arrhythmias (95% CI)All defects % (95% CI)
Rustico et al., 19954BasicLow risk84.6 (54.6–98.1)23.1 (12.5–36.8)35.4 (23.9–48.2)
Anandakumar et al., 20022Basic/ExtendedUnselected94.0 (84.4–98.5)82.1 (76.5–86.9)95.2 (76.2–99.9)85.4 (80.9–89.2)
Hafner et al., 19983 Low risk87.5 (65.1–97.9)32.4 (21.5–44.8)83.3 (17.7–19.9)46.1 (35.4–57.0)
Achiron et al., 19921 Low risk83.3 (55.6–97.1)50.0 (11.8–88.2)87.5 (28.4–99.9)78.3 (56.3–92.5)
Stümpflen et al., 19965ExtendedLow risk86.1 (61.9–97.6)

Discussion

Fetal echocardiography had specificity close to 100% when diagnosing babies without congenital heart disease among unselected and low risk populations. It was also found to be more sensitive at correctly detecting minor, major, non-structural and all defects when performed among an unselected population. The possibility of ascertainment bias limits the validity of this finding though as three of the four low risk studies had a higher prevalence of congenital heart disease than the unselected population screened by Anandakumar et al.2 This study was further undermined by the choice of postnatal reference standard. Just because there is absence of a murmur at neonatal examination does not mean that congenital heart disease is absent.

The sensitivity of the scanning regime for detecting major defects does not appear to improve with how extensive is the fetal echocardiography examination. Rustico et al.,4 using basic fetal echocardiography, had a similar estimate of sensitivity with overlapping 95% CIs as Achiron et al.1 who used fetal echocardiography routinely and referral for extended fetal echocardiography on suspicion of congenital heart disease. Although there is some evidence to suggest that sensitivity for detecting minor defects and all defects improves with how extensive the fetal echocardiography examination is, this is not consistent across all studies. Anandakumar et al.,2 using basic fetal echocardiography followed by extended fetal echocardiography on suspicion of congenital heart disease, achieved a similar estimate of sensitivity with overlapping 95% CIs for the detection of all defects as Stümpflen et al.5 who routinely used extended fetal echocardiography.

Sensitivity in the detection of congenital heart disease is also not necessarily consistent with an increase in operator skills. Anandakumar et al.2 had lower operator skills than Achiron et al.1 and Stümpflen et al.5 but achieved a similar sensitivity with overlapping 95% CIs for the detection of all defects. Nor does the equipment used appear to explain differences in sensitivity. This is despite evidence that the type of transducer can improve visualisation of the cardiac structures and the accuracy to detect cardiac defects.34

In summary, all eligible studies found fetal echocardiography was close to 100% specific for diagnosing normal fetus among unselected and low risk populations. The sensitivity for correct diagnosis of babies for congenital heart disease defects was higher among the unselected population than among the low risk populations. Indeed, the sensitivities for detecting congenital heart disease among the unselected population was comparable to high risk populations10,23–25; however, the potential for ascertainment bias and the choice of reference standard limits the validity of this finding. Because of the substantial variation in study characteristics, it was also difficult to disentangle which potentially important clinical factors such as scanning regime, operator skill and equipment, could attribute to the higher sensitivities. So in conclusion, there is insufficient evidence to support the routine use of fetal echocardiography among unselected and low risk populations during the second trimester to detect congenital heart disease. Future research should address the methodological limitations of the studies highlighted by this review and consider the need to assess the broader consequences of using fetal echocardiography in these populations such as the effect on health outcomes and associated costs.

Acknowledgements

The authors would like to thank Professor James Weir for his comments. The review was funded from a Medical Research Council MSc studentship in Health Sciences.

Appendices

Appendix A. 

References of studies included

[1] Achiron R, Glaser J, Gelernter I, Hegesh J, Yagel S. Extended fetal echocardiographic examination for detecting cardiac malformations in low risk pregnancies. BMJ 1992;304:671–674.

[2] Anandakumar C, Nuruddin M, Wong YC, Chia D. Routine screening with fetal echocardiography for prenatal diagnosis of congenital heart disease. Ultrasound Rev Obstet Gynecol 2002;2:1–55.

[3] Hafner E, Scholler J, Schuchter K, Sterniste W, Philipp K. Detection of fetal congenital heart disease in a low-risk population. Prenatal Diagn 1998;18:808–815.

[4] Rustico MA, Benettoni A, D'Ottavio G, Maieron A, Fischer-Tamaro I, Conoscenti G. Fetal heart screening in low-risk pregnancies. Ultrasound Obstet Gynecol 1995;6:313–319.

[5] Stümpflen I, Stümpflen A, Wimmer M, Bernaschek G. Effect of detailed fetal echocardiography as part of routine prenatal ultrasonographic screening on detection of congenital heart disease. Lancet 1996;348:854–857.

Appendix B. 

References of studies excluded

High risk patients only

[1] Berghella V, Pagotto L, Kaufman M, Huhta JC, Wapner RJ. Accuracy of prenatal diagnosis of congenital heart defects. Fetal Diagn Ther 2001;16:407–412.

[2] Marasini M, Rimini A, Conti F, Pongiglione G. Functional aspects of congenital heart disease in utero. J Perinat Med 1994;22:51–53.

[3] Perolo A, Prandstraller D, Ghi T, Gargiulo G, Leone O, Bovicelli L. Diagnosis and management of fetal cardiac anomalies: 10 years of experience at a single institution. Ultrasound Obstet Gynecol 2001;18:615–618.

[4] Strauss A, Toth B, Schwab B, Fuchshuber S, Schulze A, Netz H. Prenatal diagnosis of congenital heart disease and neonatal outcome—a six years experience. Eur J Med Res 2001;6:66–70.

[5] Tometzki AJ, Suda K, Kohl T, Kovalchin JP, Silverman NH. Accuracy of prenatal echocardiographic diagnosis and prognosis of fetuses with conotruncal anomalies. J Am Coll Cardiol 1999;33:1696–1701.

[6] Hsieh CC, Kuo DM, Chiu TH, Hsieh TT. Prenatal diagnosis of major congenital cardiovascular malformations. Gynecol Obstet Invest 1996;42:84–87.

Not all patients received fetal echocardiography

[7] Klein SK, Cans C, Robert E, Jouk PS. Efficacy of routine fetal ultrasound screening for congenital heart disease in Isere County, France. Prenat Diagn 1999;19:318–322.

[8] Stoll C, Alembik Y, Dott B, Roth PM, Degeeter B. Evaluation of prenatal diagnosis of congenital heart disease. Prenatal Diagn 1993;13:453–461.

[9] Wyllie J, Wren C, Hunter S. Screening for fetal cardiac malformations. Br Heart J 1994;71:20–27.

Gestation not reported

[10] Ott WJ. The accuracy of antenatal fetal echocardiography screening in high- and low-risk patients. Am J Obstet Gynecol 1995;172:1741–1747.

Scanning not described in enough detail

[11] Rustico MA, Benettoni A, D'Ottavio G, Bogatti P, Fontana A, Pecile V. Fetal echocardiography: the role of the screening procedure. Eur J Obstet Gynecol Reprod Biol 1990;36:19–25.

[12] Stoll C, Alembik Y, Dott B, Meyer MJ, Pennerath A, Peter MO. Evaluation of prenatal diagnosis of congenital heart disease. Prenat Diagn 1998;18:801–807.

Insufficient data reported to construct 2 × 2 table

[13] Bromley B, Estroff JA, Sanders SP, Parad R, Roberts D, Frigoletto FD. Fetal echocardiography—accuracy and limitations in a population at high and low-risk for heart-defects. Am J Obstet Gynecol 1992;166:1473–1481.

[14] Hunter S, Heads A, Wyllie J, Robson S. Prenatal diagnosis of congenital heart disease in the northern region of England: benefits of a training programme for obstetric ultrasonographers. Heart 2000;84:294–298.

Ancillary