[The result-sequence method (sequence analysis) in leptospira research. 3. Communication: The bilateral sequential test (de Boer, Armitage) for testing the difference of mean values of two binomial distributions (author's transl)].

The influence of the liver-activated and non-activated cytostatic drug Cyclophosphamid on the respiration of Leptospira biflexa semaranga Veldrat S 173 was tested in three different concentrations by group-sequential testing of two relative frequencies in a two-sided test-reading. The conditioned probability theta = 0.82 results from thetan = 0.40 and thetaa = 0.75. On account of a practicable sample size, we choose theta' = 0.80 (activated form superior) and theta'' = 0.20 (non-activated form superior). The test-adjusted level of significance is alpha = 0.05 less than beta = 0.10. The expected values for the number of discordant pairs are Etheta' = Etheta'' = 13, Etheta = 1/1 = 14 and Ethetamax = 16. The acceptance inspection performed by control chart and tabulated reference figures results - in comparison with conventional procedures - in savings of discordant pairs of 65 per cent (concentration 10(-4) g/ml) in a decision in favour of activated form, of 65 per cent (concentration 10(-8) g/ml) in an equal efficiency of both states of Cyclophosphamid and of 10 per cent (concentration 10(-11) g/ml) in a non-significant difference. Taking into consideration all unrestrictedly selected pairs, the duration of the experiment takes only 4.25 months instead of 6.75 if the statistical data analysis is not performed conventionally, but group-sequentially instead.