| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1 Genetics, North York General Hospital, Toronto, M2K 1E1 Canada
aGenetics, North York General Hospital, 4001 Leslie St., Toronto, Ontario, M2K 1E1 Canada; fax 416-756-6727, e-mail thuang{at}nygh.on.ca)
Second-trimester maternal serum screening (MSS) for Down syndrome has been widely used in routine prenatal care in developed countries. The screening combines maternal age-specific risk of Down syndrome with risk estimation obtained by measuring maternal serum markers to assign women an expected risk of having a term Down syndrome pregnancy. Diagnostic tests were offered to women whose risk exceeded the risk cutoff determined by the screening program. The commonly used triple test, which involves the use of maternal age, serum 
-fetoprotein, unconjugated estriol, and human chorionic gonadotropin, was expected to have a Down syndrome detection rate of 60–65% and false-positive rate of 5% (1). Although the expected screening performance has been achieved in many screening programs, the accuracy of individual risk calculated by a relatively complex computation based on a statistical model was not immediately obvious. Good agreement between the expected risk of Down syndrome and observed prevalence has been reported previously in several screening programs (2)(3)(4)(5). We evaluated the accuracy of expected risk of Down syndrome in a large provincial, multiple test center, MSS program in Ontario, Canada.
MSS has been coordinated at the provincial level in Ontario since 1993. Triple maker screening (-fetoprotein, unconjugated estriol, and ß-human chorionic gonadotropin) was carried out in seven regional laboratory centers. Information including screen utilization, results, follow-up data, and the pregnancy outcomes of all women screened in the seven centers was collected in the Ontario MSS database. Data on outcomes for all pregnancies screened were obtained through the Canadian Institute of Health Information, which records every hospital admission in Canada. Where necessary, information was verified through provincial genetic-counseling centers and cytogenetic laboratories. Using this protocol, we obtained 94.4% of outcomes.
The study was based on 311 256 women screened in the Ontario MSS program between October 1993 and September 1998. Of the 311 256 women screened, a Down syndrome risk level was recorded for 301 700, and 284 804 (94.4%) of them had outcome data from Canadian Institute of Health Information, including 506 pregnancies associated with Down syndrome. The expected risks of a term Down syndrome pregnancy were calculated with AFP Expert (Benetech). The risk cutoff used in the Ontario MSS program was 1 in 385.
Using a technique described by Wald et al. (2), we ranked the women screened according to their individual expected risk of Down syndrome. They were divided into 10 groups, each group containing 44–59 cases of Down syndrome pregnancies. Two factors were considered when grouping the cases: (a) that each risk group covered an appropriate risk range; and (b) that there was a similar number of cases in each group. The mean expected term risks of an affected pregnancy were calculated for each group. The risks were then compared with the observed risks (prevalence) of that particular group (2). Because it was estimated that 23% of Down syndrome pregnancies will abort spontaneously after 16 weeks of gestation, cases with positive screening results (risk, 
1 in 385 in our program) and diagnosed prenatally were multiplied by 77% to allow for the spontaneous fetal losses (6).
Table 1
compares the expected risk of Down syndrome with its observed prevalence; it gives the risk category, the mean expected risk, number of Down syndrome cases, adjusted number of Down syndrome cases, and observed birth prevalence of Down syndrome for each group of women. The mean expected risks were close to the observed prevalence across all the risk groups, particularly for women with very high expected risks (women in risk groups 1 in 8 or greater, 1 in 9 to 1 in 25, and 1 in 26 to 1 in 45).
|
The logarithmic transformed mean expected risks of Down syndrome are plotted against the logarithmic transformed observed prevalence in Fig. 1
. The plot shows that the mean expected risk of Down syndrome was close to the observed prevalence (R2 = 0.9901).
|
Good agreement between the expected and observed risks of Down syndrome has been reported in several studies. Wald and colleagues (2)(4) described a technique that can be used to validate the expected risk of Down syndrome. Using this method, they compared expected risk of Down syndrome with the prevalence observed in their screening program. The studies consisted of approximately 120 000 women screened, including 153 cases of Down syndrome screened with the triple test and 86 cases of Down syndrome screened with the quadruple test. The results showed that the estimated risks were accurate across the entire range of risks (2)(4). Similar results were reported by Canick and Rish (3), who detected in their program 66 cases of Down syndrome among the 49 139 women screened with the triple test. The agreement between expected and observed risks of Down syndrome was also assessed by Onda et al. (5) based on 9350 Japanese women screened using triple test and 24 Down syndrome cases.
To our knowledge, our data set is the largest for assessing agreement between expected and observed risks of Down syndrome. Our results showed that the expected risks of Down syndrome assigned to individual women were close to the observed risks across all the risk ranges. The expected risks and observed risks for women in high risk groups (risk 1 in 45) were almost identical.
We examined the completeness of the ascertainment of Down syndrome in our study population by comparing the expected number of Down syndrome cases with the number identified through the screening program. The expected number of Down syndrome births was estimated by applying the age-specific Down syndrome risk to the age distribution of women screened with an available outcome. In our study population, we would have expected 444 Down syndrome births in the absence of prenatal diagnosis and selective termination. After adjusting for spontaneous fetal losses, we would have expected 424 term Down syndrome pregnancies. The rate of ascertainment was consistent with that reported by Canick and Rish (3) in a similar, but small-scale, study. We have also estimated the completeness of this ascertainment by adding the actual observed number of Down syndrome births to the number terminated or lost spontaneously, multiplying by 0.77 to adjust for the spontaneous fetal losses. Using this adjustment, we identified 439 cases of term Down syndrome, a number very close to the expected 444 cases.
In conclusion, the expected risk assigned to the individual woman in the Down syndrome serum screening program is accurate, reflecting the term risk of having a fetus with Down syndrome.
Acknowledgments
We thank all members of the Ontario MSS consultative committee and participating MSS centers for their contributions to the Ontario MSS program. We also thank the women of Ontario for supporting the MSS program.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
