Summary of findings 5. Comparative accuracy of single‐ and two‐stage screening approaches in detecting fetal structural anomalies before 24 weeks' gestation.

Comparative accuracy of single‐ and two‐stage screening approaches in detecting fetal structural anomalies before 24 weeks' gestation

First + second‐trimester scan (two‐stage screening approach)

Single second‐trimester scan (single‐stage screening approach)

Comparison

Sensitivity

Specificity

Studies

Fetuses (cases)

Sensitivity (95% CI)

Specificity (95% CI)

Studies

Fetuses (cases)

Sensitivity (95% CI)

Specificity (95% CI)

P value

Certainty (GRADE)

P value

Certainty (GRADE)

I. Overall accuracy of ultrasound in detecting fetal structural anomalies

Fetuses affected

18

181,614 (3051)

83.8% (74.7 to 90.1)

99.9 (99.7 to 100)

9

158,767 (3168)

50.5 (38.5 to 62.4)

99.8 (99.2 to 100)

< 0.001

Very lowa,b,c

0.350

Lowc,e

Anomalies*

17

229,274 (4975)

84.5 (75.6 to 90.5)

‐

3

17,845 (340)

52.2 (33.2 to 70.7)

‐

0.012

Very lowa,b,c

0.692

Lowc,e

Implications**

In a hypothetical cohort of 100,000 fetuses, 1776 are expected to have one or more structural anomalies. A two‐stage screening approach will miss 288 cases and 118 fetuses will receive a false‐positive diagnosis. A single‐stage screening approach will miss 880 cases and 206 fetuses will receive a false‐positive diagnosis.

II. Subgroup analysis of ultrasound accuracy in detecting anomalies that are less susceptible to under‐reporting due to incomplete postnatal identification (i.e. anomalies that are externally visible, symptomatic at birth or considered to be lethal)

Fetuses affected

18

181,614 (1534)

86.7 (78.8 to 92.0)

100 (99.9 to 100)

7

58,014 (559)

59.2 (45.7 to 71.5)

99.9 (99.6 to 100)

< 0.001

Very lowb,c,d

0.170

Lowc,e

Implications**

In a hypothetical cohort of 100,000 fetuses, 864 are expected to have a structural anomaly that is externally visible, symptomatic at birth or lethal. This subgroup of anomalies is assumed to be less susceptible to under‐reporting due to incomplete postnatal identification, allowing for better comparison of the screening approaches as the estimates of test accuracy are assumed to be impacted less by differences in postnatal identification rates between studies. A two‐stage screening approach will miss 115 cases of these anomalies and 30 fetuses will receive a false‐positive diagnosis. A single‐stage screening approach will miss 353 cases and 69 fetuses will receive a false‐positive diagnosis.

CI: confidence interval *No specificity was determined as the number of true‐negative test results could not be determined by the number of anomalies. **To allow accurate comparison of the effectiveness of single‐ and two‐stage screening approaches, all calculations are based on the median prevalence of structural anomalies reported across studies that evaluated a first + second‐trimester scan combined (Table 2).

GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

aContributing studies of both groups had design limitations (‐1), including risk of bias in the reference standard domain (‐0.75) and risk of bias in the flow and timing domain (‐0.25). bUnexplained heterogeneity of results among studies included in the single second‐trimester scan group (‐1). cIndirectness in terms of the comparison between single‐ and two‐stage screening strategies (‐1). dWide 95% CI, deviating more than 10% from the summary estimate of the single second‐trimester scan studies (‐1). eContributing studies had design limitations, including risk of partial verification bias of pregnancies due to incomplete verification of ultrasound findings in pregnancies with an adverse outcome (‐1).