16/04/2026
Common Pitfalls Orthodontic Publication – Part 2
In my last post, I discussed some pitfalls related to systematic reviews and RCTs, including sample size calculation and a brief introduction to shortcomings in randomization. In this post, I will elaborate more on randomization.
2️⃣ Randomized controlled trials:
🎲 Randomization:
The purpose of randomization is to create similar groups and isolate the treatment effect from any other effect. This way, we can say the observed effect is due to the intervention.
🔶 Some studies perform simple randomization which is nice but they provide very balanced groups in small samples. This raises some concern, why?
For example, if we flip a coin 30 times (representing our sample size) to assign participants into two groups, the probability of getting exactly 15 participants in each group is relatively low (around 14%). In practice, simple randomization can easily lead to unequal group sizes, particularly in smaller trials.
🔶 For this reason, researchers often use block randomization. In block randomization, participants are divided into blocks, and each block has an equal number of participants assigned to each group (color denotes group).
Now, can we get very similar groups?
Not really. Block randomization gives us equal group sizes, but not necessarily similar groups in terms of important covariates.
🔶 To improve this, we use stratification. Stratification means dividing the sample based on important covariates, then randomizing within each group. For example, splitting participants by s*x (male/female) or age groups, then randomizing inside each subgroup.
🔳 Allocation concealment:
Randomization should be hidden from anyone involved in enrolling participants (allocation concealment).
⛔️The question now can we start randomization before collecting the whole sample❓
3️⃣ Blinding⏳
