Abstract
The assay on anhydrous basis is a mathematically derived value from an experimental results of assay and water content tests. The results of assay and water content tests are determined, separately, on as-is basis. The industry-accepted formula for assay on anhydrous basis = (assay on as-is basisx100)/(100-%water). Statistically, the two variables involved in accepted formula are assay on as-is basis and water to obtain assay on anhydrous basis. The experimental errors associated with these two variables propagate in assay on anhydrous basis. The error propagates either in constructive or destructive mode. The constructive mode of error propagation is combination of positive error of assay on as-is basis and positive error of water or negative error of assay on as-is basis and negative error of water. The constructive mode of error propagation has more impact on assay on anhydrous basis values and its confidence interval. The destructive mode of error propagation is combination of a positive error of assay on as-is basis and a negative error of water or vice versa. The destructive mode of error propagation has lesser impact on assay on anhydrous basis values and its confidence interval in comparison to the constructive mode of error propagation. In accepted formula said above, the constructive or destructive error propagation causes unrealistic drift of assay on anhydrous basis towards either lower or higher side of content limit of substance. The risk of rejection of pharmaceutical use substance is higher based on assay test results that results are calculated from industry-accepted formula. The purpose of the study is to propose an alternative formula to overcome limitations of accepted formula and justify the propagation of errors in realistic way. We have given three examples of pharmaceutical use substances to emphasise the above proposition. The proposed formula for assay on anhydrous basis= (assay on as-is basisxPhi)/(Phi-%water) in which Phi is sum of experimental results of assay and water content tests experimentally determined, separately, on as-is basis.