Department of Medical Biology, Amsterdam University Medical Centers, University of Amsterdam, Amsterdam, Netherlands.
Amsterdam Reproduction and Development Research Institute, Amsterdam, Netherlands.
Cochrane Database Syst Rev. 2024 May 9;5(5):CD014715. doi: 10.1002/14651858.CD014715.pub2.
Prenatal ultrasound is widely used to screen for structural anomalies before birth. While this is traditionally done in the second trimester, there is an increasing use of first-trimester ultrasound for early detection of lethal and certain severe structural anomalies.
To evaluate the diagnostic accuracy of ultrasound in detecting fetal structural anomalies before 14 and 24 weeks' gestation in low-risk and unselected pregnant women and to compare the current two main prenatal screening approaches: a single second-trimester scan (single-stage screening) and a first- and second-trimester scan combined (two-stage screening) in terms of anomaly detection before 24 weeks' gestation.
We searched MEDLINE, EMBASE, Science Citation Index Expanded (Web of Science), Social Sciences Citation Index (Web of Science), Arts & Humanities Citation Index and Emerging Sources Citation Index (Web of Science) from 1 January 1997 to 22 July 2022. We limited our search to studies published after 1997 and excluded animal studies, reviews and case reports. No further restrictions were applied. We also screened reference lists and citing articles of each of the included studies.
Studies were eligible if they included low-risk or unselected pregnant women undergoing a first- and/or second-trimester fetal anomaly scan, conducted at 11 to 14 or 18 to 24 weeks' gestation, respectively. The reference standard was detection of anomalies at birth or postmortem.
Two review authors independently undertook study selection, quality assessment (QUADAS-2), data extraction and evaluation of the certainty of evidence (GRADE approach). We used univariate random-effects logistic regression models for the meta-analysis of sensitivity and specificity.
Eighty-seven studies covering 7,057,859 fetuses (including 25,202 with structural anomalies) were included. No study was deemed low risk across all QUADAS-2 domains. Main methodological concerns included risk of bias in the reference standard domain and risk of partial verification. Applicability concerns were common in studies evaluating first-trimester scans and two-stage screening in terms of patient selection due to frequent recruitment from single tertiary centres without exclusion of referrals. We reported ultrasound accuracy for fetal structural anomalies overall, by severity, affected organ system and for 46 specific anomalies. Detection rates varied widely across categories, with the highest estimates of sensitivity for thoracic and abdominal wall anomalies and the lowest for gastrointestinal anomalies across all tests. The summary sensitivity of a first-trimester scan was 37.5% for detection of structural anomalies overall (95% confidence interval (CI) 31.1 to 44.3; low-certainty evidence) and 91.3% for lethal anomalies (95% CI 83.9 to 95.5; moderate-certainty evidence), with an overall specificity of 99.9% (95% CI 99.9 to 100; low-certainty evidence). Two-stage screening had a combined sensitivity of 83.8% (95% CI 74.7 to 90.1; low-certainty evidence), while single-stage screening had a sensitivity of 50.5% (95% CI 38.5 to 62.4; very low-certainty evidence). The specificity of two-stage screening was 99.9% (95% CI 99.7 to 100; low-certainty evidence) and for single-stage screening, it was 99.8% (95% CI 99.2 to 100; moderate-certainty evidence). Indirect comparisons suggested superiority of two-stage screening across all analyses regarding sensitivity, with no significant difference in specificity. However, the certainty of the evidence is very low due to the absence of direct comparisons.
AUTHORS' CONCLUSIONS: A first-trimester scan has the potential to detect lethal and certain severe anomalies with high accuracy before 14 weeks' gestation, despite its limited overall sensitivity. Conversely, two-stage screening shows high accuracy in detecting most fetal structural anomalies before 24 weeks' gestation with high sensitivity and specificity. In a hypothetical cohort of 100,000 fetuses, the first-trimester scan is expected to correctly identify 113 out of 124 fetuses with lethal anomalies (91.3%) and 665 out of 1776 fetuses with any anomaly (37.5%). However, 79 false-positive diagnoses are anticipated among 98,224 fetuses (0.08%). Two-stage screening is expected to correctly identify 1448 out of 1776 cases of structural anomalies overall (83.8%), with 118 false positives (0.1%). In contrast, single-stage screening is expected to correctly identify 896 out of 1776 cases before 24 weeks' gestation (50.5%), with 205 false-positive diagnoses (0.2%). This represents a difference of 592 fewer correct identifications and 88 more false positives compared to two-stage screening. However, it is crucial to acknowledge the uncertainty surrounding the additional benefits of two-stage versus single-stage screening, as there are no studies directly comparing them. Moreover, the evidence supporting the accuracy of first-trimester ultrasound and two-stage screening approaches primarily originates from studies conducted in single tertiary care facilities, which restricts the generalisability of the results of this meta-analysis to the broader population.
产前超声广泛用于在出生前筛查结构异常。虽然传统上是在妊娠中期进行，但越来越多地在妊娠早期使用超声来早期发现致命和某些严重的结构异常。
评估低危和非选择性孕妇在 14 周和 24 周之前使用超声检测胎儿结构异常的诊断准确性，并比较当前两种主要的产前筛查方法：单独的妊娠中期扫描（单阶段筛查）和妊娠早期和中期扫描联合（两阶段筛查）在 24 周之前检测异常的能力。
我们检索了 MEDLINE、EMBASE、科学引文索引扩展版（Web of Science）、社会科学引文索引（Web of Science）、艺术与人文引文索引和新兴来源引文索引（Web of Science），检索时间从 1997 年 1 月 1 日至 2022 年 7 月 22 日。我们将研究限制在 1997 年后发表的研究，并排除了动物研究、综述和病例报告。没有进一步的限制。我们还筛选了每一项纳入研究的参考文献列表和引用文献。
如果研究包括低危或非选择性孕妇在妊娠 11 至 14 周或 18 至 24 周时分别进行第一和/或第二阶段胎儿异常扫描，且参考标准为出生时或死后检测到异常，则符合纳入标准。
两位综述作者独立进行了研究选择、质量评估（QUADAS-2）、数据提取和证据确定性评估（GRADE 方法）。我们使用单变量随机效应逻辑回归模型对敏感性和特异性进行荟萃分析。
共纳入了 87 项研究，涉及 7057859 名胎儿（包括 25202 名有结构异常的胎儿）。没有一项研究在 QUADAS-2 的所有领域都被认为是低风险的。主要的方法学问题包括参考标准领域的偏倚风险和部分验证的偏倚风险。在评估第一阶段扫描和两阶段筛查的研究中，由于经常从没有转诊的单一三级中心招募患者，且没有排除转诊患者，因此存在患者选择方面的适用性问题。我们报告了胎儿结构异常的超声准确性，包括严重程度、受影响的器官系统以及 46 种特定异常。在所有检查中，各分类的检测率差异很大，在所有检查中，胸腹壁异常的敏感性估计最高，而胃肠道异常的敏感性估计最低。第一阶段扫描检测结构异常的综合敏感性为 37.5%（95%置信区间[CI]31.1 至 44.3；低确定性证据），致命异常的敏感性为 91.3%（95%CI83.9 至 95.5；中等确定性证据），总特异性为 99.9%（95%CI99.9 至 100；低确定性证据）。两阶段筛查的综合敏感性为 83.8%（95%CI74.7 至 90.1；低确定性证据），而单阶段筛查的敏感性为 50.5%（95%CI38.5 至 62.4；极低确定性证据）。两阶段筛查的特异性为 99.9%（95%CI99.7 至 100；低确定性证据），单阶段筛查的特异性为 99.8%（95%CI99.2 至 100；中等确定性证据）。间接比较表明，两阶段筛查在所有分析中均具有更高的敏感性优势，特异性无显著差异。然而，由于缺乏直接比较，证据的确定性非常低。
第一阶段扫描在妊娠 14 周之前具有检测致死性和某些严重异常的高准确性，尽管其总体敏感性有限。相反，两阶段筛查在妊娠 24 周之前具有检测大多数胎儿结构异常的高准确性，具有高敏感性和特异性。在一个假设的 10 万例胎儿队列中，第一阶段扫描预计将正确识别 124 例致命性异常中的 113 例（91.3%）和 1776 例任何异常中的 665 例（37.5%）。然而，预计在 98224 例胎儿中会出现 79 例假阳性诊断（0.08%）。两阶段筛查预计将正确识别出所有结构异常中的 1448 例（83.8%），假阳性为 118 例（0.1%）。相比之下，单阶段筛查预计将在 24 周之前正确识别出 1776 例病例中的 896 例（50.5%），假阳性为 205 例（0.2%）。与两阶段筛查相比，这代表着正确识别出的病例减少了 592 例，假阳性诊断增加了 88 例。然而，必须认识到两阶段筛查与单阶段筛查相比的额外获益存在不确定性，因为目前没有直接比较它们的研究。此外，支持第一阶段超声和两阶段筛查方法准确性的证据主要来自于在单一三级保健机构进行的研究，这限制了本荟萃分析结果在更广泛人群中的普遍性。
