Treatment effects/differences design

The significance level relates to the risk of designating a chance occurrence as statistically significant. Usually a 5% level is utilized for testing treatment effects. If a p-value of 0.04 is reported for a treatment effect, this means that there is only a 4% chance that the difference in response between the active and control treatments occurred due to chance. Keep in mind, however, that if many tests are run in a trial, it is entirely possible that one or two might be significant due to chance. As an extreme example, consider a study in which 100 statistical tests are run. We would expect five of those tests to show significance with a p-value of 0.05 or less due to chance. Therefore, it is essential to specify the main tests to be run in the protocol. Any tests that are conducted after the trial has been completed should be clearly labeled as post hoc exploratory analyses. 

Figure 2 shows the marker by treatment interaction design discussed by Sargent et al. (18) and by Pusztai and Hess (19). Both marker positive and marker negative patients are randomized to the experimental treatment or control. The analysis plan either calls for separate evaluation of the treatment difference in the two-marker strata or for testing the hypothesis that the treatment effect is the same in both marker strata. 

Analysis of variance appropriate for a crossover design on the pharmacokinetic parameters using the general linear models procedures of SAS or an equivalent program should be performed, with examination of period, sequence and treatment effects. The 90% confidence intervals for the estimates of the difference between the test and reference least squares means for the pharmacokinetic parameters (AUCo-t, AUCo-inf, Cmax should be calculated, using the two one-sided t-test procedure). 

Evolutionary designs were devised by Dixon and Armitage. Although the statistical analysis is rather different, they have the same objective, which is to detect a treatment effect at the earliest moment possible, using the fewest possible patients, while retaining statistical robustness. Both types are suited for exploratory clinical research and diseases which are rare. 

This test has very low power, however, for three reasons. (1) It is a between-patient test for a trial which has been designed to use within-patient differences to detect treatment effects. (2) The carry-over effect where it occurs is likely to be somewhat smaller than the pure effect of treatment. (3) The carry-over is in any case only manifested in the second period. Therefore, although it is necessary to use the totals to compare sequences to account for other effects that might bias the test of carry-over, the direct information for carry-over comes only from the second period and the effect of this is diluted. In short, although a test of carry-over is available it is too weak to be of much use. 

Random-effect model. A term which is used in at least two rather different senses by statisticians in the context of drug development. (1) A model for which more than one term is assumed random but the treatment effect is assumed fixed. (All statistical models, including so-called fixed ones have at least one error term which is random.) (2) A model in which the treatment effect itself is assumed to vary randomly from unit to unit. For balanced designs, random-effect models of the first sort can lead to identical inferences to fixed-effect models. Even for balanced designs, random-effect models of the second sort will not. 

The third advantage of quantitative real-time RT-PCR is that it allows one to monitor the treatment effect in individual patients as a potential surrogate marker after adjuvant chemotherapy (K3) or immunotherapy (S4) against micrometastatic diseases. For the purpose of monitoring, a relative quantification method which is designed to determine exact, PCR efficiency-corrected mRNA concentration, normalized to a calibrator, might be desirable to overcome the inter-assay variation from run to run (SI3). This may make it possible to directly compare the mRNA values at different time points. 

With this design all the measurements using 0.1 M perchloric acid as the quinine solvent occur (by chance) on the first two days, whereas those using 0.5 M perchloric acid happen to be made on the last two days. If it seemed that there was a difference between the effects of these two acid levels, it would not be possible to tell whether this difference was genuine or was caused by the effect of using the two treatments on different pairs of days. A better design is one in which each treatment is used once on each day, with the order of the treatments randomized on each day. For example ... 

Consider now a different scenario. Imagine a different trial of a similar design in which the treatment effect point estimate was also 8.00 mmHg, but the two-sided 95 % confidence interval has a lower limit of 2.5 mmHg and, therefore, an upper limit of 13.5 mmHg. The result would be written as follows ... 

---