The parallel line bioassay is a classical statistical technique used in pharmacology to estimate the relative potency of a test compound relative to a standard drug. Unfortunately, this complex analysis is not available on commercial statistical analysis software. Therefore, shortcut methods, that are not based on a valid statistical process are used to obtain relative potency estimates. A user-friendly statistical analysis program, entitled BIOASSAY, using the parallel line bioassay being as core procedure, is being developed and written in C++ for use with DOS systems. The program is designed to make more efficient use of balanced experimental designs, such as those used for human and primate pharmacological studies, but it may be used with in vitro studies as well. Also included are other statistical tests which are useful in analyzing dose-response data: linear regression, Dunnett's test, missing values calculations, Scheffe-Box test for homogeneity of variance, t-tests and a data transformation routine. Data may be imported from LOTUS files, and there are three ways to output the final analyses. The unique statistical advantages of the program include: the calculation of relative potency estimates and confidence limits based on linear, parallel portions of two dose-response curves for equi-effective drug effects; partitioning out a Between Subjects variance component with the parallel line, linear regression and Dunnett's tests; calculating missing values for randomized block experimental designs and making the appropriate corrections in the ANOVA; and having the capability of making inverse predictions with confidence limits from linear dose-response curves (i.e. estimating the dose of a drug required to produce a given effect). The scope of the project encompasses program development, validation studies and preparation of an instruction manual. At this time, programming is complete, validation is in progress, and a draft of the instruction manual has been prepared.
