This course presents critical concepts and practical methods to support planning, collection, storage, and dissemination of data in clinical research. Understanding and implementing solid data management principles is critical for any scientific domain. Regardless of your current (or anticipated) role in the research enterprise, a strong working knowledge and skill set in data management principles and practice will increase your productivity and improve your science. Our goal is to use these modules to help you learn and practice this skill set. This course assumes very little current knowledge of technology other than how to operate a web browser. We will focus on practical lessons, short quizzes, and hands-on exercises as we explore together best practices for data management.
Baseline Data and Demographics
Loading...
From the course by Vanderbilt University
Data Management for Clinical Research
380 ratings
From the lesson
Planning a Data Strategy for a Prospective Study
This module reviews the process of planning data elements for a real-world research study.
Meet the Instructors
Stephany Duda, PhDAssociate Professor
Department of Biomedical Informatics
Paul Harris, PhDProfessor
Biomedical Informatics and Biomedical Engineering
So, coming back to this study of the pain measurement
and the, the sample use case of around morphine and Marinol.
Now that we've defined the study, and we've sort of laid out
how the procedures will work and what the What, what the population
will look like, and how we'll kind of lay them out
on the, on a calendar from a, from a study by it.
Sorry.
Participant, by participant basis. And what we'll be doing there?
I think now, the next thing we want to do is
look at the, look at brainstorming data collection.
You know, let's again go back and see, see what
we would collect, how we would, we would collect it.
After that, we'll go into actually
implementing this in the electronic data capture.
But before we start thinking about electronic case report
forms it's always very, very wise to step back and
think about, you know, all of the different pieces of
what you're trying to measure, what you're trying to study.
So, so that at the end of the day You've got something
that, that that, that regardless of how the, the data come out
you, you can either prove or disprove the hypothesis and you've got
some scientific benefit that can be
published and, and there again benefit society.
So, as we go through this, we'll, we'll, we'll kind of go back to a few
of the, the important concepts, or the
best practices that we've talked about early, earlier.
[COUGH]
One of the, one of the very, most important
of those concepts is again, sort of take a step backwards.
Think about things before we start doing them.
So, it's very important that we, that we take
the time right now as we're considering this study.
rather than hurry it along because if
we hurry it along we'll, we'll undoubtedly forget
something that, that may in fact be
the difference between having a, a good scientific
conclusion and, and maybe having something that's inconclusive.
Again taking that concept one step further.
It's an ethics thing.
So we are no matter how safe we define a study,
no matter how much societal benefit there is to that study, and,
and that's what we're, that's our end objectives in any research study
that we are conducting, there is always going to be some risk.
Or some inconvenience to the individuals that are working with us on studies.
And so I always teach in data management that, that doing this, planning
things up front it's one of the most important things that we do.
If we don't, and we don't get as much scientific benefit out of
that even potential risk, or discomfort, or
inconvenience to our subjects we're really doing
a disservice to everyone.
So, another good practice that we'll think about as we're
going through as being exact and discrete when defining our variables.
We know that at the end of this study, we want to be able to analyze
this data, and we want to be able to sort of come up with an answer.
Of whether there is a synergistic effect between morphine and marinol.
if we don't think hard about it, if we don't come up
with things that are quantifiable in, in a way that is structured and
measurable, then you know all we'll have at the end of this study is a hunch.
Or, or, or something that we're not quite sure of
[COUGH]
We will, we'll look, especially at the beginning,
of how we'll kind of start brainstorming this.
We'll look at the confounding factors.
You know, making sure that at the end of the day, when we
get ready to study this, and if, if we see variations, that we've measured.
Things that, up front may be confounding factors that, that could potentially, play
a role in, separating our responses, even among healthy volunteers.
So, so first let's think about that.
So, demographics you know, I, I've worked
many, many studies over many, many years and
whether you call these demographics or participant
data, or, or, you know, somewhere in between.
There's always sort of this collection of one
time data about an individual that, that you're studying.
So, so if we look at the, you know, maybe, maybe taking the first one of
those might be name.
You know we need to identify that patient for later.
And, and again, here we could digress a little bit and we could talk about
the fact that some studies are, are
completely deidentified in terms of the data management.
They, they don't know and they don't need to keep
track of who, who an individual is after a visit.
So maybe it's completely deidentified.
in this case, we'll just make an assumption that we do need identifiers.
And so,
the first thing that we might we might consider is a name.
So, you know, name is pretty easy.
Mine is Paul Harris.
We could, we could have a, a, if we were doing this on paper, we could have
a put your name here line or if we
were doing it electronically, we could put your name.
You know, in this box, and I might type
Harris comma Paul, and I might type Paul Harris.
I might type P.
Harris.
So, so to get rid of the ambiguity there, it's probably a
good idea, even thing about things like the
discreetness factor when we, when we get into names.
And so we'll collect that as a first name and a last name.
information about our participants like their, their sex.
So, so here you know, we, we know that potentially by going back to the
literature we know that there might be differences in how males and
females characterize and, and deal with pain.
And so given that that may be a confounding factor in
the future, it's probably a good idea to collect that data.
And so we, we, learned in one of our previous
lectures about best practices that if we're going to collect data,
it's a good idea to, to, in, in a categorical
way like this, it's a good idea to code them.
that way we don't have the situation where an
individual types in male Or M and female or F.
You know we have everything structured.
We'll sort of list in one place that, that male
equals one, female equals two, and maybe transgender equals three.
That way we, we've got a structured way of sort of collecting it.
But also it's going to help us a lot downstream when we
get ready to analyze the data, to have these in structured format.
Hair color, so there's,
there's recent studies that show that at
least individuals with red hair may may process
and deal with pain in a different way than individuals that do not have red hair.
So, it's a confounding factor, you know, maybe when we
get ready to analyze the data We see some grouping in
the answers, and if we, if we go back and
we stratify by hair color, maybe that would explain some of
the variation.
If that's possibly the case, then we'd better collect the data.
And so maybe we would again go back and code black hair as one,
dark brown as two, light brown as three, blonde as four, and red as five.
And I didn't make these up.
I went to the, to the, to the
literature, saw how other people had collected these types
of data after I looked at how you
know, hair color could possibly be a confounding factor.
And by the way, this is one of those examples where
I'm not exactly sure, but the study in question, this morphine study.
Was done at least around ten years ago.
And so this is an area where, you know?
It could very well be that this hair color finding happened after the fact.
And so, it just sort of drives home the point that thinking about things.
And thinking about things that, that, that actually might play, play a role.
I-, it's important before we start the study.
even considering
the fact that there might be things that are unknown now.
That, that might become known later.
And we talked about that in a previous lecture
maybe about collecting DNA for the same types of reasons.
Handedness, so just, just doing a quick review of
the literature shows that at least in some cases,
in some circumstances, people that are left-handed might process
and deal with pain different than people that are
right handed.
So, not a bad idea to collect, in part of our
demographics about each person, whether they're left handed or right handed.
And again, we'll code those, one is left, two is right.
body mass or you know, obesity, there's some recent literature that, that talk
about the fact, talks to the point that You know individuals at, high,
high obesity levels might have, might process pain in
a different way than individuals that, that, that are leaner.
So here we get into one of those questions,
well how are we going to measure that one?
Should measure, Just measure weight, or should we measure BMI.
I think it's a good idea to, to measure BMI which
is one of those characteristic measurements of, of sort of obesity,
and, and, body mass.
But, but, lets not collect it and store it as BMI, we talked in a, in
a previous lecture about the fact that we
won't collapse our variables until we need to.
So why not just collect it, since BMI is made up of weight and height?
We'll just collect those as two different measurements.
we'll collect weight, and we'll collect height, and we can
always, we can always calculate the BMI from those measurements.
This sort of brings up a question of, well, when do we collect those data?
in a non-pediatric study where we're dealing with adults, and for
a fairly short amount of time, you know, one, two years.
This study is a really small one.
We'll probably have a patient enrolled and finished within a month
of, of, of, of engaging in a relationship with that particular volunteer.
But, in general, you know?
In, in cases like that one.
Probably, the individual's height is not going to change during that study.
If it's pediatric study, maybe so.
in, in which case, we'd want to measure height at every visit.
The weight, you know, for a month study.
It's probably not going to fluctuate a great deal.
But, you know, it wouldn't hurt, if, if, if it's convenient, or if it's possible.
It might be that we'll collect the height,
on the demographics, on the screening visit.
We'll calculate, we'd, we'd collect the weight there as well.
But as we'll go along in the study,
even over a month time period, maybe we'll collect
the weight each time they come in for a
visit, because we do, we all Fluctuate in weight.
And so, so getting that at the visit level rather
than at the one-time study entry level might make sense.
Other factors that we might consider. smoking status.
You' know there's
some information in the literature that talks
about smoking being a confounding factor in Individual
sensitivity to pain, anxiety and depression, how athletic
an individual is, stress level on the day
of the study, you know other things again
we'll definitely want to collect in that demographics
or screening visit would be whether they're pregnant
or not, whether they're a current drug user.
And again we're going to collect those and we're going to use those
as inclusion, exclusion criteria so that you
know, if an individual says that they are
a current drug user or if our
testing indicates that they are currently using drugs.
Then we would, we would remove them from the study, but we would
still, more than likely, collect the data about them and maintain the data
about them so that we would know that, at the end of the
study, that if we had people that didn't go forward that we knew why.
And that we could convince ourselves that we weren't biased
in treating in any particular patient or patient cohort.
Sorry, volunteer cohort.
Okay age. Age is a thing that we might consider.
There is information in the literature that says
people in different age brackets process pain differently.
So here, again, it's kind of like that BMI one.
Do we, do we collect the data do we
collect age and store, you know, how old you are?
What if you,
what if your birthday is tomorrow?
So I'm 49 and, and tomorrow I'm going to be
50 or I'm 70 and tomorrow I'm going to be 71.
It, it's a good idea because again we can calculate age.
If we know the date of your visit and we know the date of birth.
And we can always calculate the age and you know, it's, it's just a good idea to
sort of think this in and make the, make
the measurements as, as discrete as, as we can.
So, what I would recommend there would be
we collect the date of birth on the initial
visit and then each particular visit would also going to
be storing that the date of that visits, so
that we can calculate the age across the entire study.
Ethnicity, so, so here's where you, you
know, we get into collecting information on ethnicity
and race, which again might very well
be confounding factors, in, in this particular study.
Here's where you know, we sort of have to look at local context.
So this study was conducted in the United States.
in the United States we have some fairly structured and, standardized
ways that we collect data around ethnicity as well as race.
And those are part of information that our National Institutes of Health, put out
and sort of have guidance they have around gathering and reporting this information.
So for this
particular study.
We'll, we'll go with ethnicity as either
Hispanic or Latino, or not Hispanic or Latino.
And then the race will be cap, captured in these particular categories.
Again, we al, we always have to consider local context.
If we're doing a study in Puerto Rico.
Or if we're doing a study in Japan or France.
the-, these racial, these racial and ethnicity categories
maybe they don't make sense for that particular population.
So don't take any of this information as a this is exatly how you should do a study.
But look at it as more of a guidance and way to think through things.
Okay.
So we're at the end of this video segment and we'll pick
up again in the next one with collecting data at the visit level.
Explore our Catalog
Join for free and get personalized recommendations, updates and offers.
Coursera provides universal access to the world’s best education,
partnering with top universities and organizations to offer courses online.
© 2018 Coursera Inc. All rights reserved.
