Accuracy of Alpha in Antenatal Screening

An audit of a large routine Down’s syndrome screening programme was carried out to: (i) Demonstrate the value of the Validation Plot facility in αlpha for determining the accuracy of the Down’s syndrome risks calculated by the αlpha software and (ii) Determine the accuracy of αlpha in estimating Down’s syndrome screening performance.

The audit was carried out on data from 292,633 pregnancies screened as part of the routine screening service provided by the Wolfson Institute of Preventive Medicine Screening Service in London, England.  The pregnancies were screened using the first trimester Combined test (NT, PAPP-A and free ß hCG together with maternal age).  The risk of a pregnancy being affected with Down’s syndrome was calculated by the αlpha software using the statistical parameters and methods described in the Serum, Urine and Ultrasound Screening Study (SURUSS) study (references 1 & 2).  Risks greater than or equal to 1 in 150 at term were screen positive for Down’s syndrome.

i) Use of Validation Plots to determine accuracy of risk estimation

839 of the 292,633 pregnancies screened were affected with Down’s syndrome.  Each affected screened pregnancy was identified using the αlpha software database and pregnancy outcome recorded.

The data in αlpha were used to derive the Validation Plot (reference 3) shown in the figure below in which the observed prevalence of Down's syndrome is plotted against the median risk in six approximately equal groups of affected pregnancies ranked by the reported risk.  The diagonal line represents perfect agreement between the reported risk of Down's syndrome and the prevalence of Down's syndrome in the absence of screening.   The plotted points are all close to the diagonal line showing that the reported Down’s syndrome risks are accurate.

ii) Accuracy of Alpha’s reporting of screening performance

αlpha was used to calculate the expected screening performance for the first trimester Combined test at a risk cut-off of 1 in 150 at term with the age distribution in the screened population using the methods described in the SURUSS study.  The table below shows that the expected screening performance is remarkably close to that observed in the audit.

Screening Performance Detection rate False Positive Rate Odds of being
affected given a
positive result (OAPR)
Expected 81.6% 3.0% 1:14
Observed 81.8% (95% CI 79.0 - 84.3%) 2.71% (95% CI 2.65 - 2.77%) 1:11.5

Conclusion

This illustrates the accuracy of the αlpha software for antenatal screening in a large screening programme and the value of using Validation Plots to audit screening programmes.

References

1. First and second trimester antenatal screening for Down’s syndrome: the results of the Serum, Urine and Ultrasound Screening Study (SURUSS) (2003)
2. Improvements in antenatal screening for Down’s syndrome (2013)
3. Validation plots in antenatal screening for Down’s syndrome (2006)