Objective: To assess the suitability of either the log-linear or reciprocal-linear regression procedure for maternal weight correction of biochemical marker MoMs in the first trimester.
Methods: Data from two prospective first-trimester OSCAR screening programmes including 32,010 women with first-trimester maternal serum-free beta-hCG and PAPP-A measured by the Kryptor analyser was analysed by regression analysis to provide parameters for the log-linear and reciprocal-linear MoM correction procedures. Assessment was made by goodness of fit to the data. The impact on detection rate and false-positive rate of the different correction procedures was assessed using statistical modelling with biochemical markers alone.
Results: Both log-linear and reciprocal-linear correction were shown to fit the data well. For free beta-hCG, the log-linear procedure was marginally superior to the reciprocal-linear procedure (r2=0.986 v 0.980), whilst for PAPP-A the reciprocal-linear procedure was marginally better (r2=0.991 v 0.985). Log-linear correction reduced the variance for both markers more than did the reciprocal-linear procedure. For free beta-hCG, the sd was reduced from 0.2675 to 0.2605 and for PAPP-A, it was reduced from 0.2545 to 0.2336. Correcting for maternal weight was shown to reduce the population false-positive rate from 7.0 to 6.5%, whilst maintaining the same detection rate at a risk cut-off of 1 in a 100. At individual levels, a two-fold variation in risk was demonstrated depending upon the individual's weight.
Conclusions: To provide accurate individual patient-specific risks for trisomy 21, maternal weight must be taken into account and should be a mandatory data item for screening programmes. Maternal weight correction in the first trimester using free beta-hCG and PAPP-A can be best achieved using the log-linear procedure.
Copyright 2003 John Wiley & Sons, Ltd.