Data monitoring

Item 28b: Explanation of any interim analyses and stopping guidelines, including who will have access to these interim results and make the final decision to terminate the trial.

Example

​

“There will be one formal interim analysis to test the primary hypothesis. The decision boundaries at the interim analysis are calculated for either stopping for futility or stopping for efficacy using the O’Brien-Fleming error spending function [References]. Since we adopt the non-binding futility boundary [References], if it is decided the study will continue due to other considerations even if we cross the futility boundary at the interim look, there will be no inflation of type I error. …  In our design setting, even if the efficacy boundary is crossed at the interim look, the study may continue and there will be no inflation of type I error, as efficacy is already established at the interim. The interim analysis will occur when 528 events (50% information) are collected.

If the interim result meets the stopping criterion for futility, that is, the one-sided p-value for Wald’s test on the log(hazard ratio) between SR [spatial repellent] and placebo at the interim is > 0.3450, the study may stop for futility. .... If the one-sided p-value < 0.00882, then the study can stop for efficacy; otherwise, the study will proceed. However, if it is decided the study will continue due to other considerations even if we cross the efficacy boundary at the interim look, there will be no inflation of type I error, as efficacy is already established at the interim.

If the one-sided p-value at the interim is > 0.00882 but < 0.3450, then the study will continue. At the final analysis, if the one-sided p-value from Wald’s test on the log(HR) between SR and placebo < 0.04668, we will reject the null hypothesis, claiming SR reduces the malaria hazard rate compared to placebo in Kenya at the significance level of 5% ….

Interim analysis data will be available to the DSMB [data and safety monitoring board], Funder, Sponsor, and SCJ [SC Johnson] along with the study oversight contractor, fhiClinical, and any ad hoc experts deemed appropriate. The DSMB has the ability to recommend stopping the trial based on safety concerns, but do not have the responsibility of stopping the trial due to their assessment of efficacy or futility. The responsibility to stop the trial is held by the Sponsor ” [457].

​

Explanation

​

Interim analyses can be conducted to formally monitor the accumulating data in clinical trials. They are generally performed in trials that have a Data Monitoring Committee (DMC), longer duration of recruitment, and potentially serious outcomes.

The results of these analyses, along with non-statistical criteria, can be part of a stopping guideline that helps inform whether the trial should be continued, modified, or halted earlier than intended for benefit, harm, or futility [458]. Criteria for stopping for harm are often different from those for benefit and may, or may not, employ a formal statistical criterion [459]. Stopping for futility occurs in instances where, if the study were to continue, it is unlikely that an important effect would be seen (i.e., low chance of rejecting null hypothesis). Multiple analyses of the accumulating data increase the risk of a false positive (type I error), and various statistical strategies have been developed to compensate for this inflated risk [460]. Aside from informing stopping guidelines, pre-specified interim analyses can be used for other trial adaptations such as sample size re-estimation, alteration of the proportion of participants allocated to each study group, and changes to eligibility criteria [461].

Interim analyses were described in a third of 894 trial protocols approved from 2000–2003 in Switzerland, Germany, and Canada [455]. Such analyses were also reported in 71% (106/150) of cancer trial protocols with time-to-event outcomes in Italy from 2000–2005; [462] among 86 protocols with plans for efficacy interim analyses, 100% reported the timing of the analyses, 90% specified the overall reason for stopping (e.g., superiority, futility), and 94% detailed the statistical approach [462].

A complete description of any interim analysis plan, even if it is only to be performed at the request of an oversight body (e.g., the data monitoring board), should be provided in the protocol – including the statistical methods, who will perform the analyses, and when they will be conducted (timing and indications). If applicable, details should also be provided about the decision criteria – statistical or other – that will be adopted to judge the interim results as part of a guideline for early stopping or other adaptations.

In addition, it is important to state who will see the outcome data while the trial is ongoing, whether these individuals will remain blinded (masked) to study groups, and how the integrity of the trial implementation will be protected (e.g., maintaining blinding) when any adaptations to the trial are made. A third of protocols for industry-initiated randomised trials receiving Danish ethics approval in 1994–1995 stated that the sponsor had access to accumulating trial data, which can introduce potential bias due to conflicts of interests [38]. Finally, the protocol should specify who has the ultimate authority to stop or modify the trial – e.g., the principal investigator, trial steering committee, or sponsor.

Summary of key elements to address

Interim analyses:

• When they will be conducted (timing and indications), and by whom

• Statistical methods

• Who will have access to interim results, and whether they will be blinded

Stopping guidelines:

• Any criteria (statistical or non-statistical) that will be used to inform decisions about early stopping or other adaptations (e.g., sample size re-estimation)

• Who will make the decision to continue, stop, or modify the trial

​

​