Interim analyses

This section contains the following:

Introduction
Things to consider
Illustrative examples - WHO pre-eclampsia trial
Additional resources
Further reading

Introduction

Any analysis used to compare efficacy or safety issues between the treatment arms prior to formal completion of a trial is called an interim analysis. All interim analyses should be carefully planned in advance and described in the protocol. The optimal timing and frequency of analyses will vary between trials but it is generally six monthly or annually, or after a given number of outcome ‘events’. In most cases, the interim analysis should be performed by the trial statistician only and no other members of the trial research group should have access to the results. An independent Data Monitoring Committee should be set up to view the results.

The format of the interim analyses should be agreed in advance with the Data Monitoring Committee (see scheduling and organising the DMC meetings). It will often follow the structure described in the Dummy Tables with the addition of further details on adverse or safety issues. Statistical testing is usually only performed on the primary outcome(s), but again this is at the discretion of the Data Monitoring Committee.

Formal statistical stopping guidelines can be applied to interim analyses, but these are only one of a number of considerations that a Data Monitoring Committee to take into account. Others include the balance of risk and benefit and the consistency with external evidence. Details of the statistical methods and assumed boundaries for stopping, if any, should have been detailed in the protocol.

Things to consider

The timing and frequency of any planned interim analyses should be agreed with the Data Monitoring Committee.
The trial statistician will perform the analysis and should be the only member of the trial research group who has access to the results.
The structure of the interim analyses should be agreed at the start of the trial with the Data Monitoring Committee.
If formal statistical stopping procedures are to be applied, details of the methods and boundaries for stopping will be in the protocol and the trial standard operating procedures manual.

Illustrative examples - WHO pre-eclampsia trial
An interim analysis was suggested to the DMC to be conducted after the completion of the first 4500 subjects. If the difference between the proportions of occurrence of pre-eclampsia in the two groups was significant in a two-sided test at the level alpha=0.01 then the trial would have been stopped prematurely. The adjustment for repeated testing at the final analysis will then be negligible, leaving a significance level of practically alpha=0.05. (WHO Multicentre Randomized Trial of Calcium Supplementation for the Prevention of Pre-eclamsia - go to protocol - go to publication)

Illustrative examples - WHO pre-eclampsia trial

An interim analysis was suggested to the DMC to be conducted after the completion of the first 4500 subjects. If the difference between the proportions of occurrence of pre-eclampsia in the two groups was significant in a two-sided test at the level alpha=0.01 then the trial would have been stopped prematurely. The adjustment for repeated testing at the final analysis will then be negligible, leaving a significance level of practically alpha=0.05. (WHO Multicentre Randomized Trial of Calcium Supplementation for the Prevention of Pre-eclamsia - go to protocol - go to publication)

Additional resources

Analysis and interpretation checklist

This checklist was developed by Dave Sackett, who prepared it for the 3rd edition of Clinical Epidemiology; A Basic Science for Answering Questions about Health Care, published by Lippincott, Williams & Wilkins in 2005.

ICH Harmonised Tripartite Guideline: Statistical Principles for Clinical Trials

This document provides guidance for the design, conduct, analysis, and evaluation of clinical trials of an intervention in the context of its overall clinical development.