Randomized Controlled Trial

We talked about a random sample last week.  Today we are going to talk about randomized controlled trials. The use of the term randomized in this context does not refer to the sampling method.  Randomized controlled trials use convenience sampling, meaning they recruit whoever is available and meets the enrollment criteria.  Because of this, randomized controlled trials are still subject to sampling bias.

But what randomized controlled trials provide is study groups who should not differ on the characteristics that might confound the outcome.  For example, women who give birth at a birth center may think differently about birth than women who give birth at a hospital.  

The groups may have chosen to prepare for labor differently, and may have different plans for how to manage any pain in labor.  Because of this, comparing the proportions of women who use pharmacologic pain relief may tell you more about the differences in the women than the differences in practice at the two birth sites. If you want to know how effective birth centers are at reducing pharmacologic pain relief, you need to randomize women to either hospital or birth center for their birth and then watch to see what happens.

When you randomize the women, it means that you control what intervention the women receive rather than the women choosing it. This is what makes this type of study an experiment rather than an observation, and the reason it provides stronger support for understanding problems.

Observational vs Experimental Research

To better understand the difference, think of your child’s science fair project.  We can pretend your child is interested in plants, so she chooses a project on how light affects plant growth.

Your child could measure the size of different plants in your garden and compare the heights based on how much time each plant spends in full sun each day. This would be an observational study, because she is observing what naturally happens. The drawback of this method is that the plants in the sun and the plants in the shade are likely going to be different plants. This means the differences she finds may be due to variation in plants, not variation in light.

Your child could buy three similar sized houseplants, maybe African Violets (my mother’s favorite), and place them strategically so one plant gets very little light, one plant gets light almost 24 hours per day and one plant is the “control” and gets the average amount of light near a window. This is an experiment because your child is determining how much light each plant will receive and then comparing how the difference in light affects growth.

The control

Part of the strength of the randomized control trial is the use of a control group.  The control group is a group of participants similar to those involved in the experiment who do not receive the intervention.  After the study group receives the intervention the groups are compared to see if there is any difference.

The control is important because sometimes it is difficult to know what would happen if you didn’t do the experiment.  For example, if you were to give antibiotics to a study group with a cold you might think the antibiotics were responsible for the study groups’ improved health.  But with a control group you would see that both those who took antibiotics and those who did not were healthy again in about the same time so the antibiotics were not effective.

Problems with Randomized Controlled Trials

Randomized Controlled Trials are expensive, and they take a long time to do well.  Because of this, researchers tend not to use them until there is significant evidence to suggest an intervention will be successful. This means these types of studies come later than other studies, and there will be fewer of them.  But they are worth the wait. The randomized controlled trial provides strong support for the usefulness of an intervention (do antibiotics actually help you get over a cold) and can help to determine what size effect you can expect if you use the intervention (how much faster you will get over a cold with antibiotics).

An additional problem for pregnancy and birth research is that women tend to have very strong opinions about what they want for pregnancy and labor care. This makes it difficult (if not impossible) to randomize women to groups for some research questions. So while the randomized controlled trial would provide strong support for different interventions and comfort measures, the best research we can get is still observations.

A Literature Example

I chose this article as an example for a randomized controlled trial because it is open access, you do not need a journal subscription to read the full article. The joural also includes an editors summary to help you understand the research question and results:  Effect on postpartum hemorrhage of prophylactic oxytocin by injection by community health officers in Ghana

Go ahead and have a look. There is a lot of information in this paper, so take your time.  Pay particular attention to Figure 2, the CONSORT flow diagram.  We’ll talk about that next week.

The Birth Worker Survey

I tried to have an example of a randomized controlled trial for you from the birth worker survey, but I wrote the survey program incorrectly.  I chose a great method for randomization — using birthdays split into even and odd days.  This method gave two groups with equal numbers and good distribution of characteristics except the number of college graduates. Check out this table to see how the groups compared.

Group 1

Group 2



38 (SD 8.9)

34 (SD 7.6)


Income >$50K

9 (60%)

7 (44%)


Completed College


6 (38%)



11 (73%)

13 (81%)


Given Birth

14 (93%)

14 (88%)


So what went wrong?

The question behind the experiment was this, Do the stories birth workers are exposed to affect their recommendations for comfort measures? The idea was to have one group read a birth story where the woman praised a comfort measure most birth workers wouldn’t choose to recommend, then see if that group was more likely to rate the method as useful in labor, or at least be ambivalent about it.

While the initial division into groups worked properly, I contaminated my groups. Only one group was supposed to have read the birth story, but because I put the sections in the wrong order everyone read the story and the groups were equal in the number who felt lying in bed was not helpful and those who were ambivalent. I’m sorry about that.

So instead what happened was that only one group rated a description of a common comfort technique, abdominal breathing.  What is interesting is that those who did read the description of abdominal breathing were more likely to rate it as helpful in labor (87% compared to 69%), but the result was not significant.

Next time we will look at the CONSORT document and other tools to help evaluate research.