Second trimester serum tests for Down's Syndrome screening



Down's syndrome occurs when a person has three copies of chromosome 21 ‐ or the specific area of chromosome 21 implicated in causing Down's syndrome ‐ rather than two. It is the commonest congenital cause of mental retardation. Noninvasive screening based on biochemical analysis of maternal serum or urine, or fetal ultrasound measurements, allows estimates of the risk of a pregnancy being affected and provides information to guide decisions about definitive testing.  


To estimate and compare the accuracy of second trimester serum markers for the detection of Down’s syndrome.

Search methods

We carried out a sensitive and comprehensive literature search of MEDLINE (1980 to May 2007), EMBASE (1980 to 18 May 2007), BIOSIS via EDINA (1985 to 18 May 2007), CINAHL via OVID (1982 to 18 May 2007), The Database of Abstracts of Reviews of Effectiveness (The Cochrane Library 2007, Issue 1), MEDION (May 2007), The Database of Systematic Reviews and Meta‐Analyses in Laboratory Medicine (May 2007), The National Research Register (May 2007), Health Services Research Projects in Progress database (May 2007). We studied reference lists and published review articles.  

Selection criteria

Studies evaluating tests of maternal serum in women at 14‐24 weeks of gestation for Down's syndrome, compared with a reference standard, either chromosomal verification or macroscopic postnatal inspection.

Data collection and analysis

Data were extracted as test positive/test negative results for Down's and non‐Down's pregnancies allowing estimation of detection rates (sensitivity) and false positive rates (1‐specificity). We performed quality assessment according to QUADAS criteria. We used hierarchical summary ROC meta‐analytical methods to analyse test performance and compare test accuracy. Analysis of studies allowing direct comparison between tests was undertaken. We investigated the impact of maternal age on test performance in subgroup analyses.

Main results

Fifty‐nine studies involving 341,261 pregnancies (including 1,994 with Down's syndrome) were included. Studies were generally high quality, although differential verification was common with invasive testing of only high‐risk pregnancies. Seventeen studies made direct comparisons between tests. Fifty‐four test combinations were evaluated formed from combinations of 12 different tests and maternal age; alpha‐fetoprotein (AFP), unconjugated oestriol (uE3), total human chorionic gonadotrophin (hCG), free beta human chorionic gonadotrophin (βhCG), free alpha human chorionic gonadotrophin (αhCG), Inhibin A, SP2, CA125, troponin, pregnancy‐associated plasma protein A (PAPP‐A), placental growth factor (PGF) and proform of eosinophil major basic protein (ProMBP).

Meta‐analysis of 12 best performing or frequently evaluated test combinations showed double and triple tests (involving AFP, uE3, total hCG, free βhCG) significantly outperform individual markers, detecting six to seven out of every 10 Down's syndrome pregnancies at a 5% false positive rate. Tests additionally involving inhibin performed best (eight out of every 10 Down's syndrome pregnancies) but were not shown to be significantly better than standard triple tests in direct comparisons. Significantly lower sensitivity occurred in women over the age of 35 years. Women who miscarried in the over 35 group were more likely to have been offered an invasive test to verify a negative screening results, whereas those under 35 were usually not offered invasive testing for a negative screening result. Pregnancy loss in women under 35 therefore leads to under ascertainment of screening results, potentially missing a proportion of affected pregnancies and affecting the accuracy of the sensitivity.

Authors' conclusions

Tests involving two or more markers in combination with maternal age are significantly more sensitive than those involving one marker. The value of combining four or more tests or including inhibin have not been proven to show statistically significant improvement. Further study is required to investigate reduced test performance in women aged over 35 and the impact of differential pregnancy loss on study findings.


S Kate Alldred, Jonathan J Deeks, Boliang Guo, James P Neilson, Zarko Alfirevic


Plain language summary

[Summary title]

[Summary text]


S Kate Alldred, Jonathan J Deeks, Boliang Guo, James P Neilson, Zarko Alfirevic

Reviewer's Conclusions

Authors' conclusions 

Implications for practice 

The evidence for tests involving inhibin as a marker suggests a superiority that is not found to be statistically significant, and based on small populations of women. We would not recommend that these tests should be introduced into wider clinical practice without careful consideration of cost.

The review has shown that tests involving two or three markers in combination with maternal age are significantly better than those involving one marker. We would therefore recommend that one marker tests are not used for Down's syndrome screening. The choice of multiple markers will depend on the availability of certain assays in local laboratories. There was no test combination shown to be superior to others therefore, we cannot recommend a specific test combination.

The performance of tests at earlier gestations will be the subject of a separate Cochrane review and the alternative first trimester, cross‐trimester, ultrasound and combinations of serum and ultrasound should also be considered when making policy decisions.

Implications for research 

Further evaluation of inhibin‐based test combinations are required to determine whether their apparent advantages are not chance findings. Further study of the attenuated performance of test combinations in women over 35 is required, as this age group has the highest incidence of Down's syndrome and has the greatest requirement for tests with high detection rates.

Future studies should ensure that adequate sample sizes are recruited, and take opportunities to make comparisons of test performance testing several alternative test combinations on the same serum samples. Such direct comparison removes issues of confounding when making test comparisons, and allows a clear focus on testing the incremental benefit of increasingly complex and expensive testing strategies. The reporting of studies of test accuracy can be improved and more closely adhere to the STARD reporting standards. Three key aspects of this are 1) formally testing the statistical significance of differences in test performance in direct comparisons and estimating incremental changes in detection rates (together with confidence intervals), 2) clearly reporting the number of mothers studied and their results, and 3) reporting the numbers of women who are lost to follow‐up. Many authors reported results of extrapolating findings to age‐standardised national cohorts to demonstrate the performance of the test, and failed to report the actual numbers studied and evaluated.

For the purposes of meta‐analysis and to allow for comparisons to be made between different tests and combinations, we would recommend the publication of consensus standard algorithms for estimating risk, and reporting of test performance at a standard set of thresholds. This would be difficult to achieve and implement, but an attempt at consensus should be made.

Get full text at The Cochrane Library