Clinical and statistical methodology for cooperative clinical assays of analgesics

This paper describes a method for carrying out a cooperative clinical bioassay of analgesics. The method was developed in five Veterans Administration Hospitals. The patients were largely older men with acute postoperative pain. Procedures for collecting and recording data and for automatic data processing are discussed. Uniformity trial results with morphine are reported. These data indicate the magnitude of hospital differences, patient differences, differences due to order of administration, and differences due to morphine dose. The inherent variation in the assay is reported for various completely randomized complete‐block, crossover designs. The number of patients required for specified reliability of potency estimates is presented. Statistical techniques for analysis, particularly the pooling of results from hospitals, are discussed.