Should maternal alpha-fetoprotein estimation be carried out in centers where ultrasound screening is routine? A sensitivity analysis approach.

We describe a case of small, open spina bifida that was detected on the basis of maternal serum alpha-fetoprotein levels but consistently missed on careful ultrasound examination. This prompted us to reexamine the policy (adopted by an increasing number of departments) of relying exclusively on ultrasound examination for the detection of spinal lesions. In this article we therefore analyze the number of additional cases of spina bifida that could be detected by offering maternal serum alpha-fetoprotein screening in addition to a routine ultrasound examination. Our calculations are made on the basis of published assessments of the performance of these tests and on the assumptions about test independence. We test our conclusions against a wide variety of test performance values in a detailed sensitivity analysis. The final decision to carry out maternal serum alpha-fetoprotein testing is a value judgment, but this decision should be based on realistic numeric estimates of the potential benefits and hazards of this procedure.