SIMULTANEOUS DETERMINATION OF DRUGS IN CONCENTRATION RATIOS ABOVE 40 1 BY APPLICATION OF MULTIVARIATE CALIBRATION TO ABSORBENCY AND DERIVATIVE SPECTROPHOTOMETRIC SIGNALS/

The performance of two multivariate calibration methods, multiple line ar regression (MULTI3) and partial least-square regression (PLS-2), fo r the resolution of the ternary mixtures of acetylsalicylic acid-caffe ine-codeine and acetaminophen-caffeine-codeine is compared. The method ologies were checked by applying them to the analysis of two sets of l aboratory-prepared mixtures over the concentration ranges 13.0-19.0, 1 .00-3.00 and 0.20-1.00 mu g . ml(-1) for acetylsalicylic acid-caffeine -codeine and 12.0-22.0, 0.40-2.00 and 0.20-1.00 mu g . mg(-1) for acet amimophen-caffeine-codeine, respectively. While the results provided b y MULTI3 were unacceptable in the majority of the cases, those obtaine d by PLS-2 were quite good with considerably diminished errors, as a r esult of calibration and/or checking with MULTI3, taking no account of potential interactions between analytes. It was shown that it is poss ible by using PLS-2 to resolve complex mixtures of analytes in a highl y disparate proportion, even when they have overlapping signals. The p roposed method was successfully demonstrated for pharmaceutical tablet s.