A new two-stage vaccine trial design _1258

e707004304

A new two-stage design of vaccine trials

Domain criteria and sample size determination formula: optimal experimental design is to meet the following two types of errors (3), (4) and (2) under the conditions of the ESS. Minimum: P (X, l ≥ c, l, Xr2 ≥ cr2IP, 0) ≤,, (3) P (】 ≤ cc1 '≤ c port IPl0) ≤, (4) where and are the two first errors pre-given level of significance. If there are multiple designs make ESS0 minimum, select the test power P (Xd ≥ cr1'Xr2 ≥ cr2, l ≤ cc1'X mouth ≤ c port IPr1'Pc1'0) greatest. If you want to consider the minimax design, that is to meet the two types of errors (3), (4) and (2) under conditions of minimum total sample size. 2 vaccine trial design examples in this section the method is applied to design of vaccine trials. Phase Ⅱ trials for a vaccine,[link widoczny dla zalogowanych], we ask that the response rate of the vaccine at least 30%, and the toxicity ratio of no more than 10%. On the other hand, if the vaccine response rate is less than or equal to 10%, or toxicity ratio greater than or equal to 30%, we are instantly out of such vaccines. To this end, construct the following hypothesis test:: p, 0 ≤ O. 1 or Pl0> 10.3H: p1> 10.3 and ≤ O. Pc】 1. If the required probability of committing Type I error is less than 0.05, the probability of committing Type II error is less than 0.20, then the traditional two-stage optimal design is this: start l7 patients vaccinated, if at most 2 effect is significant, or 4 had significant side effects, then immediately terminate the test; Otherwise, no amount of treatment of 23 patients, if the effective two-stage total of 8, and up to 7 people have side effects, then we have good reason to reject the null hypothesis, that the vaccine is worth further study. Sometimes a patient may find 4O Ⅱ trial, the design can not be achieved then the idea of the first section, the main control response rate of Type I error (ie, required = 0.05), and appropriate to relax Type I error toxicity limits, such as setting = 0.3, other conditions remain unchanged. This corresponds to the optimal two-stage design is as follows: Test started l3 patients were vaccinated, up to a patient if the effect is significant, or 4 had significant side effects, then the test immediately terminated; Otherwise, no amount of injection l4 patients, if the two phase of a total of 6 people effectively, and up to 7 people have side effects, it is legitimate to reject the null hypothesis that the vaccine should be entered Ⅲ trial. Compared to traditional two-stage optimal design, the proposed two-stage design of the optimal number of patients needed was reduced from the original 4O to 27, 32.5% reduction in the sample. Table 2 lists several other conditions and optimal two-stage minimax design, for reference.