This paper describes a variety of test designs and statistical technics for use with the mouse hemidiaphragm method for assaying insulin.
Designs for both qualitative tests and quantitative assays are presented together with examples. The only object of the qualitative test is to detect the presence of insulin in a sample, and this is done by demonstrating a significantly higher level of glycogen in hemidiaphragms incubated with the sample than in those incubated with insulin-free medium. Procedures for analysis of variance and for making tests of significance are given.
The aim of the quantitative assay is to estimate the potency in milliunits per milliliter and to assign confidence limits. For this purpose a test with standard insulin and the unknowns, each at two dose-levels, is utilized. The two-dose design is applicable to the assay of insulin either with or without anti-insulin serum. As previously reported, the use of this serum greatly increases the precision (λ-value) of the assay by increasing the slope of the doseresponse curve without simultaneously increasing the standard deviation. The antiserum may also be used to “bracket” the potency of samples within known limits, prior to carrying out a two-level quantitative assay.
Methods are described for ihe analysis of variance of the two-dose design and for the calculation of relative potency and confidence limits. Procedures are given for assessing the validity of the assays and also for combining the results of several independent tests.