[Clinical Trial] Superiority Trial of Medical Device Clinical Trial Design

The purpose of the clinical superiority trial of medical devices is to confirm that the effectiveness/safety of the experimental device is better than the control device/standard treatment method/placebo control, and the difference is greater than the pre-set superiority value, that is, the difference between the test device and the control device has the minimum value of clinical practical significance.

Due to the different characteristics of the test device, control device and main evaluation indicators, some superiority trials did not consider the superiority threshold. If so, the applicant needs to state the reason for not considering the superiority threshold [1].

In the superiority trial, the full analysis set should be used as the main analysis set because it contains non-compliant patient and may underestimate the effectiveness. The analysis results based on the full analysis set are conservative. Per protocol set shows the effectiveness of the test device in accordance with the prescribed program. Compared with the post-marketing effectiveness, there may be an overestimation of the effectiveness [1]

1. Scope of application

For a newly developed test device, it usually has some advantages. Generally, it is necessary to conduct a superiority trial with a placebo control or standard device to compare its true effectiveness and safety to judge its interest risk.

2. Superiority threshold

In general, take ∆ = 0 (absolute number index) or 1 (relative number index); in special cases, you need to determine another value [2].

3. Hypothesis test and interval estimation

For different types of metrics and indicators, the expressions of the original hypothesis (H0) and alternative hypothesis (H1) of the superiority trial are different, as shown in the following table. Among them, Δ is the non-inferiority margin, absolute measurement indicators include mean difference and rate difference, and relative measurement indicators include rate ratio, risk ratio, odds ratio, etc. The higher-better indicator is an indicator whose larger value indicates better effectiveness and the lower-better indicator is an indicator whose smaller value indicates better effectiveness.

Table Null hypothesis (H0) and alternative hypothesis (H1)

Indicator type	higher-better indicator	lower-better indicator
Absolute measure (mean difference / rate difference)	H0：T- C ≤∆（Δ>0） H1：T- C>Δ（Δ>0）	H0：T-C≥-Δ（Δ>0） H1：T-C<-Δ（Δ>0）
Relative measure （RR/HR/OR）	H0：T/ C≤∆（Δ>1） H1：T/ C>∆（Δ>1）	H0：T/ C≥1Δ（Δ>1） H1：T/ C<1Δ（Δ>1）

Hypothesis test: single test α=0.025, if P>α, H0 is not rejected, and it is not yet proved that the test device is superior to the control group; if P≤α, H0 is rejected, and the test group is superior to the control group.

Interval estimation: The higher-better indicator (T-C), that is, the lower limit of the (1-2α)% two-sided confidence interval is greater than Δ (superior effect limit value), can make clinical superiority conclusions, and the lower-better indicator (T-C) that is, the upper limit of the (1- 2α)% bilateral confidence interval is greater than Δ (superior effect limit), and a clinical superiority conclusion can be made.

4. Conversion of superiority to non-inferiority

If the results of the superiority trial indicate that there is no significant difference between the groups, it is feasible to switch the objective of the experiment from superiority to non-inferiority. The premise is a pre-determined non-inferiority threshold (limited to widely accepted thresholds); the results of the full analysis set and the matching analysis set should be similar, showing the confidence interval and P value of the non-inferior results. At the same time, the trial is designed and implemented according to the strict requirements of the non-inferiority trial; the sensitivity of the trial is high enough to ensure that the actual difference can be detected. There is direct or indirect evidence that the control group has its due effectiveness.

Reference:

[1] Guidelines for the design of clinical trials of medical devices, 2018

[2] Qin H., & Ming Z. (2007). Understanding of non-inferiority, equivalence and superiority design in statistical hypothesis testing of clinical trials. Chinese Journal of Clinical Pharmacology [J], 23(1):63-67.