OBJECTIVE: To investigate the application of Multiple Limit Regression (MR), that is alternative to regression on order statistics, in imputing nondetacts below multiple detect limits. METHODS: The Multiple limits regression method was used to dealing with the nondetects in cadmium residue in Fish food in 2004 from China markets; The nondetects are imputed with the predicted values based on MR; The results of estimates are compared with those from simple substitution methods. The analysis was performed using self compiled SAS macro code. RESULTS: The mean estimated by MR methods is 0.01509 mg/kg, the means from substitution methods with 0, 1/2LOD and LOD are 0.014759, 0.015270, 0.015781 respectively. CONCLUSION: MR is a worthy recommendatory method in handling nondetects with multiple limits of detect for its rationality and can be completed conveniently by compiled SAS macro code.
OBJECTIVE: To investigate the application of Multiple Limit Regression (MR), that is alternative to regression on order statistics, in imputing nondetacts below multiple detect limits. METHODS: The Multiple limits regression method was used to dealing with the nondetects in cadmium residue in Fish food in 2004 from China markets; The nondetects are imputed with the predicted values based on MR; The results of estimates are compared with those from simple substitution methods. The analysis was performed using self compiled SAS macro code. RESULTS: The mean estimated by MR methods is 0.01509 mg/kg, the means from substitution methods with 0, 1/2LOD and LOD are 0.014759, 0.015270, 0.015781 respectively. CONCLUSION: MR is a worthy recommendatory method in handling nondetects with multiple limits of detect for its rationality and can be completed conveniently by compiled SAS macro code.
Authors: Mahan Shahrivari; Elizabeth Wise; Micheline Resende; Jonathan J Shuster; Jingnan Zhang; Roberto Bolli; John P Cooke; Joshua M Hare; Timothy D Henry; Aisha Khan; Doris A Taylor; Jay H Traverse; Phillip C Yang; Carl J Pepine; Christopher R Cogle Journal: Circ Res Date: 2017-05-10 Impact factor: 17.367