Randomization is the process of allocating people to treatment groups, using an element of chance to determine the assignment.
Randomization is meant to reduce the chance of systematic bias that can be encountered if investigators, study staff, or volunteers are involved in making a decision about which treatment a volunteer receives. In the context of clinical trials, randomization is a method based on chance alone, by which eligible and consented volunteers are assigned to one of the study interventions.
Randomization keeps treatment allocation free from selection bias.
The primary reason for randomization in a clinical trial is to produce groups of patients who differ only in their treatment assignment. If these groups were to differ in some other way, then any future comparison might be biased.
The randomization procedure should be detailed in the protocol. Ideally, it should be determined by the investigator and the biostatistician, by considering the following:
- Open trial vs. blinded trial
- Single blinded or double blinded
- Number of volunteers to be randomised
- Number of study sites
- Number of treatment groups or arms
- Treatment allocation ratio
- Stratification factors, if any.
The randomization list and the method of implementation should be generated by biostatistician. The SOP should be detailed and should include the following:
- Brief description of the randomization process
- A record of variables used in the procedure
- Details of the person generating the randomization list
- Software used to generate the list and perform the randomization (if applicable), and validation details of the system.
- Approach used for allocation concealment (example: password protected electronic format)
- Details of the secure location where the randomization list is stored
- Details of the personnel who will have access to the randomization list. Please note, this list should not be accessible by the investigator and the trials team until database is locked and the code is officially broken at the end of the trial. For emergencies (example: adverse events or serious adverse events), the list can be provided to the investigator in a sealed envelope.
- Details of the person (should not be part of the study) who will securely hold the treatment codes.
- If the study is blinded, the details on how the IP will be packaged, coded and labelled to protect the blinding – these must be provided to the IP manufacturer.
- Details of any additional documents that have to be completed (due to randomization).
- Details on how the pharmacy will be informed of the randomised treatment code allocation. Example: via fax / email / other media.
The protocol should also define the level of blinding in the trial, [examples are: unblinded (also called open), single-blind, or double-blinded]; also how the blinding will be implemented (example: use of identical placebo).
The code breaking procedure must be thoroughly documented in the protocol, including the following:
- Circumstances under which the blinding can be broken. Example:
- Medical emergency where the knowledge of the blinded treatment is essential.
- Occurrence of an adverse event
- Someone in the volunteer’s household accidentally takes the IP.
- In the event of an SUSAR (Suspected Unexpected Serious Adverse Event) that might require expedited reporting.
- If requested by the DSMB.
- Details of the format of the code break (example: tear off labels / scratch cards / 24-hour telephone number / IVRS [Interactive Voice Response System] / IWRS [Interactive Web Response System], etc.).
- Investigator can break the treatment blinding when a volunteer’s safety is at risk. The code break process, process to be followed for notification, documentation to be recorded should be established at the time of trial planning (before a code is broken).
- Secure location of code break envelopes.
- Details of circumstances where a volunteer can continue to participate in the study following unblinding.
Resources and further reading
- Good Clinical Practices for Clinical Research in India, Central Drugs Standard Control Organization, Ministry of Health and Family Welfare, 2001, available online (last accessed on 26.02.2019).
- ICH Harmonised Tripartite Guideline, Statistical Principles for Clinical Trials, E9, current step 4 version, dated 5th February 1998, available online (last accessed on 26.02.2019).
- Clinical Development Services Agency SOP for Randomization, Blinding and Code Breaking Procedures, version 1.0, 2015.
- Online resource: Research Randomizer (free) – https://www.randomizer.org/