Scribd
Upload
12 views5 pages

GHR Study Design Part2

Uploaded by

paul junior yebga
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
12 views5 pages

GHR Study Design Part2

Uploaded by

paul junior yebga
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

FUNDAMENTALS OF GLOBAL HEALTH RESEARCH: PLANNING, IMPLEMENTATION, & DISSEMINATION

Optimizing Study Designs Part 2


Lecturer: Dr. Grace John-Stewart

In part two of the study design lectures, I'll talk about randomized clinical trials.

RCTs
Now, let's talk about randomized clinical trials or RCTs. With these kinds of studies, you
randomize two similar groups of people, give one group an intervention and the other
no intervention and compare the outcomes between the two groups.

Recall from the last lecture some of the questions I mentioned around early HIV
research. How effective are condoms? Will certain vaccines work? How effective is post
exposure prophylaxis? Does circumcision or STD treatment prevent HIV? A lot of these
questions could be looked at with RCTs except for the question about condoms. Once
you have evidence that something is sexually transmitted it's not ethical to tell someone
not to use some sort of protection so in that case you couldn't do a randomized trial. It's
really interesting that we tell everybody to use condoms to prevent sexual transmission
of HIV but there's no RCT that basically tells you that condoms prevent HIV; however,
there are different types of evidence that show you that condoms are effective.

The biggest advantage of an RCT is that you've got what we call temporality. You also
have causality because when you randomly pick a group of people you don't have a lot
of hidden biases. The purpose of randomization is to make these two groups as similar
as possible without bias that's why RCTs are very strong study designs.

Recall the ecologic and cross-sectional studies on circumcision that I shared with you
earlier. A lot of people even after these studies didn't believe that circumcision really
had an effect on HIV transmission. Policy wouldn't change unless there was an RCT
because people felt there was still residual confounding. They felt that circumcised men
who participated even longitudinally had different sexual behavior than uncircumcised
men and so it wasn't the circumcision that was protective. So, a randomized trial was
conducted where men were randomly allocated into two groups: circumcised or not
circumcised. The researchers then followed both groups to see who got HIV. What's
incredible about these studies is that 20 years ago people believed that men would
never agree to be randomly allocated to be circumcised, but they did, and three clinical
trials were conducted; all of which showed that there was a marked benefit of
circumcision in decreasing HIV incidence.

So, when you do a clinical trial of circumcision and you find that there's lower incidence
of HIV, would you change your policy to recommend circumcision at this point? One of
the interesting things with these studies is that one study was found to be significant

Fundamentals of Global Health Research online course


University of Washington 1
while the other two were still going on. The question was, should they stop the other
two or continue? Since they knew the findings from the first RCT study, people felt that
participants needed to be informed that there was a benefit to circumcision. Once
they're fully informed, they have a choice. However, the decision was to keep the
studies going after informing participants to add to the weight of the evidence for
policy.

This is an interesting example. Experts debate as to whether one study is enough to


change policy or if you need two or three studies. Is it ethical to continue another study
once you have evidence? Most of the time if there's very strong evidence, people will
stop and say we have enough evidence, we don't need to keep going with another
study.

Factorial Clinical Trial


So, the next kind of study I want to touch on is factorial design. This type of study has a
lot of efficiency because you have four arms. You can look at intervention A and
intervention B, a third arm where you give both intervention A and B, and a final arm
with neither intervention. So, you can see if either one of these interventions A or B
works. You can also see if combined they have synergies or antagonism or if there's no
difference between them. Here’s a factorial study where there are four arms: vitamin A,
multivitamins combined, and placebo. In this study, the researchers were able to look at
the effects of all of these different arms.

Community RT
Another kind of randomized trial is a community randomized trial. In this type of RCT,
researchers take similar pairs of cities or regions and then randomly allocate one of the
pair to one of the two interventions.

Fundamentals of Global Health Research online course


University of Washington 2
Let's look at an example. In this community randomized trial, they had six paired
communities with 1,000 adults in each community. In one group of communities, they
did aggressive STD interventions and in the other they didn't. They found that HIV
incidence was much lower in the communities where they randomized to the STD
intervention. This was a very expensive study with a lot of logistics but in this study, they
found STD treatment was beneficial.

Here's another community randomized trial, which was done in Uganda. The
researchers had ten community clusters. One community cluster received home-based
mass antibiotics and the other received vitamins. In this community randomized trial,
there was no difference in HIV incidence between the clusters. As you might expect
these community randomized trials are expensive and sometimes have unclear data. In
this way, they are like the women's health study we talked about in lecture 1; however,
there are a lot of nested studies that occur within these community studies and we can
get a lot of good products from these studies that make them not a waste of time and
money. But it's good to contrast these large studies with the efficiency of a case control
study.

Advantages and Disadvantages of Randomized Trials


Let's look at some of the advantages of randomized trials. They control for bias and
confounding. Like other observational cohorts you can look at multiple outcomes in an
RCT. However, RCTs do have some disadvantages. They are expensive and time-
consuming. The participants need to adhere to the intervention. You need what is called
ethical equipoise, which means balanced uncertainty. Participants who are willing to be
in the study need to be willing to be randomly allocated to either of these arms and not
feel that one is better. Ethical equipoise will be covered in the module on ethics.

There are different reasons that RCTs are built. Sometimes they're built for proof of
concept that an intervention works and if you really want to understand if the
intervention works you want to have a motivated group of people with very high
retention just to see that in the best possible circumstance this intervention works. So,
you will be conservative to get the most highly adherent group of people who will show
up and be retained in your study to prove that it works. But then in the future you need
to balance being generalizable to the larger population that may or may not be exactly
who you had in your randomized trial. If this intervention works the first time, we need
to ask how generalizable it is. For example, there have been a lot of studies on heart
disease in men; however, some of these findings weren't generalizable to women. So,
now there are heart disease studies with women.

Generalizing to populations who are less likely to come back might hurt the validity of
your study so there's this tension. And you can't do a randomized trial in every single
country and region of the world with every subgroup of people. You usually go where
the disease burden needs the intervention and then try and understand if it works
there.

Fundamentals of Global Health Research online course


University of Washington 3
Meta-analysis
Like ecologic studies, meta-analysis is another study design that I highly recommend you
explore in your career. With meta-analysis you do a comprehensive review of the
literature to systematically understand the weight of the evidence. You would pull your
results for consistency and look at the overall association. I like to do these studies when
I have a lot of time as developing new study aims because then I know what the
evidence looks like before a new study is conducted. Meta-analyses are very
inexpensive and take advantage of many different studies in the past to help you
understand the issue.

Summary
So, to summarize, it's important to keep in mind when you're designing your study what
your question is and then to ask yourself how can I answer it as efficiently and as
strategically designed as possible.

So, to summarize, it's important to first define your question, understand what your
question is and then design a study as efficiently and as strategically as possible. So, for
studies that have a rare outcome and a long duration you will consider doing a case
control design. For studies where you really want to understand the effects of an
exposure on a multitude of outcomes, a cohort study would be a natural pick. There are
simple designs that we cover that include case reports, case series, surveys, cross-
sectional designs that all take very little resources and time but are very important to
generate hypotheses.

It's also very valuable to look at the literature and synthesize those in a meta-analysis to
understand what the burden of evidence is before you design a study, and this also
takes very little budget and can use resources available to you at the library. Similarly,
an ecologic study while descriptive is quite important to generate a hypothesis.

Finally, for policy and for treatment, we often rely on randomized controlled trials. And
randomized controlled clinical trials help get around some of the issues of potential bias
and confounding and help us have similar groups that receive or don't receive an
intervention so that we can be sure that the intervention is responsible for the effect.
And so randomized trials tend to be what is the most valuable type of study to move a
new intervention into widespread use. For both randomized trials and prospective
cohorts, it's extremely important that people who are enrolled follow up and are
followed to the end of the study for us to have a valid outcome or a valid result.

So, I think knowing this whole arsenal of different types of study designs can help you in
your career as you design different studies to build the right study for the right question
and not all of them have to be cumbersome or take a lot of time and can still get at the
answer to your question.

Fundamentals of Global Health Research online course


University of Washington 4
Fundamentals of Global Health Research online course
University of Washington 5

You might also like