Making the case for objective performance metrics in newborn screening by tandem mass spectrometry

Abstract

The expansion of newborn screening programs to include multiplex testing by tandem mass spectrometry requires understanding and close monitoring of performance metrics. This is not done consistently because of lack of defined targets, and interlaboratory comparison is almost nonexistent. Between July 2004 and April 2006 (N=176,185 cases), the overall performance metrics of the Minnesota program, limited to MS/MS testing, were as follows: detection rate 1:1,816, positive predictive value 37% (54% in 2006 till date), and false positive rate 0.09%. The repeat rate and the proportion of cases with abnormal findings actually been reported are new metrics proposed here as an objective mean to express the overall noise in a program, where noise is defined as the total number of abnormal results obtained using a given set of cut-off values. On the basis of our experience, we propose the following targets as evidence of adequate analytical and postanalytical performance: detection rate 1:3,000 or higher, positive predictive value>20%, and false positive rate<0.3%.