case control and cohort

Hello everyone. Hope this finds you all well. Today I'm going to talk about two very important study designs and these are both observational study designs but they are widely used in different epidemiological investigations. I'm going to talk about case control study and cohort study designs. You might have seen this graph, this figure before. I showed this when I was talking, I was giving an overview of different epidemiological study designs in EpiMinutes 1 session, in the very first session of EpiMinutes. And so far we have discussed about case report. case series and cross-sectional study designs we have talked about correlational study designs and case report case series and cross-sectional study designs they are all descriptive study designs on the other hand correlational studies can only focus on groups not individuals now in the past we have discussed that in case report and case series we cannot test a hypothesis we cannot look for associations between an exposure and outcome because they don't have any comparison groups in cross-sectional studies we lack temporality because we take snapshots we don't know that's why we don't know whether exposure happens before the outcome and that's why we cannot prove any association and in correlational studies we work on groups of people we look we work on group level data and because of that we cannot apply the findings from our correlational studies on individuals now in both cohort and case control studies we actually deal with comparison groups we have comparison groups and that's why we can look for associations between an exposure and an outcome and that's why they are known as they are called analytic studies even though they are observational studies the investigators they just observe they do not intervene now in theory we can Apply either a case control study or a cohort study design to test a hypothesis. In practice, we need to choose one because each of the designs have their unique advantages and disadvantages. So what is the key difference between these two study designs? The key difference lies in the process of subject selection or how we divide our study participants into different groups. Let's start with case control study, then it will become more clear to us. So in a case control study, we identify a group of individuals who have the outcome of interest and we call them as cases excuse me and we also identify a group of individuals who do not have the outcome of interest and we call them controls and then we look we compare the proportion with the exposure of interest between among the cases and controls between the cases and controls This is a very nice visual presentation of case control study design. It is taken from Lancet has a very good series on different study designs. So this is taken from Lancet publication on study designs. They have a nice series on that. And if you look at the yellow box on the top right end, you can see that We are identifying a sample of cases. and we are at the present time so cases have the outcome of interest and then we go back we look back to the past and we try to identify among those cases who had exposure in the past who had the exposure of interest in the past in the second box the purple box you can see that we identify a sample of controls and also we look back and we try to identify among the controls who had exposure in the past and you also you also notice that directionality of the time which is very important so the time is moving forward but we are actually going backwards in a case control study design and both the cases and our controls are coming from population with outcome and population without outcome respectively. This is another nice presentation, a simple presentation of case control study design and you can see that among the cases a certain proportion were exposed in the past, among the controls a certain proportions were exposed in the past and we make comparison between these two proportions. of exposures in the past in many different ways. We'll discuss them in future when we'll talk about measures of associations. And also notice the directionality of the time and as investigators where we are, when the study is starting, when the onset of the study is happening. So the directionality of the time is going forward. On the other hand, the direction of our inquiry is actually going backwards to the past. This is a great example and a very famous example of case control study design and in this study the exposure was smoking and the outcome was lung cancer. It was published in British Medical Journal in 1950 and this report this paper was written by Dr. Richard Dahl and Sir Bradford Hill. And you might remember Richard Dahl because in the very last session on ecological study we talked about a paper on meat consumption and colon cancer which was published in 1975 and Richard Dahl was one of the investigators for that study. and Sir Bradford Hill was a famous statistician and when we'll talk about causality you'll see that we will actually talk about Hill's criteria because he set a series of criteria to establish causality. Now in this study they made a simple comparison They identified a group of lung cancer patients and they also collected data in terms of the smoking history among those cancer patients. And then they also identified a comparable group of participants without lung cancer. and then they collected information on their smoking history. And in the table below, they nicely presented them. And then they did chi-square test and then they identified p-value and definitely they found very significant association between smoking and lung cancer. So this was a very early stage example of case control design and you'll see when we'll be talking about measures of association that we can actually come up with some other measures of associations in case control design. Now, can you tell me what is this? Any idea? Yes, they are a group of soldiers. and they're known as cohort which is a very common term in military and the cohort study this terminology came from this because in a cohort study we identify a group of people who march forward in time having an exposure and or not and then we follow them we observe them closely to see whether they also develop the outcome of interest in time so in a cohort study We divide the participants differently. We identify a group of individuals who are exposed to a particular factor of interest and we identify another group of individuals who are not exposed to a particular factor or exposure of interest. So those who are exposed and those who are not exposed. We identify two groups of people. First we identify a cohort and then we divide them into two groups as exposed and unexposed and then we see whether they develop the outcome of interest so we follow them for a period of time and how long you are going to follow them it depends on the nature of the disease how common the disease is So this is a nice presentation, graphical presentation of a cohort study design. And you can see that we identify with a group of people first, a cohort. And at that time point, none of them were having the outcome of interest. And then we divide them as exposed and unexposed. And then we follow them to see whether. Among the exposed, anyone is developing the outcome of interest. Among the unexposed, anyone is developing the outcome of interest or the incidence among the exposed and incidence among the unexposed. And look at the directionality of the time and the directionality of inquiry. They are going towards the same direction. In a case control study, time was moving forward but our direction of inquiry was going backward. But here, Both time and inquiry have the same directionality. Now cohort studies can be of two types. They can be prospective cohort study or they can be retrospective cohort study. The prospective cohort study is also known as concurrent cohort study design. And in a prospective cohort study, The investigation begins at a point in time when the exposure has occurred but the disease has not occurred yet. In a retrospective cohort study, the investigation begins at a point in time after both the exposure and disease have already occurred. Let's have a look at this graph. First, Let's have a look at the top purple line and look at now. So at present only exposure has occurred and now we are following the exposed and unexposed and the entire cohort into the future to see how many among the exposed develop the disease of interest and how many among the unexposed develop the disease of interest. So this is a prospect. cohort study design or the concurrent cohort study design the purple line on the top right the red line look at now now both exposure and outcome have occurred we are ignoring outcome we are going back to the past we are going to the time point to the beginning when only exposure occurred None of the participants, none of the members of the cohort had outcome. And then we divide them as exposed and unexposed. And then we come back to the present to see what proportion of the exposed now is currently having outcome of interest. What proportion among the unexposed currently is having the outcome of interest. So. this becomes a retrospective cohort design. So if you compare between the two In a prospective cohort design, now we are starting the study. None of the study participants have developed the outcome of interest. We divide them as exposed and unexposed and we follow them into the future and we'd like and we'll see how many among the exposed develop the outcome of interest and how many among the unexposed develop the outcome of interest. In a retrospective cohort study, Now, outcomes have also occurred and of course exposure happened in the past. We are ignoring the outcome status. We are going back to the past. This is how it is different than a case control study design. If it was a case control study design, we divide them as cases and controls, but we are ignoring that because we are designing a cohort study. So we need to start with exposed and unexposed. So we go back to the past when no one had the outcome. We divide the participants of the cohort as exposed and unexposed and then we come back to the present and then see how many those who were exposed in the past are now having the outcome. What proportion among the unexposed in the past are now having the outcome of interest. Then we can compare incidents between the two groups that we cannot do in a case control study. and also if you want to continue into the future in your retrospective cohort study design then you can continue like the green line at the bottom then it becomes an ambidirectional study or retro prospective study examples of a prospective cohort study design we want to evaluate the possible association between exclusive breastfeeding and diarrhea in infants during the first six months of life. We identify a birth cohort, a group of children who are just born, follow them into the future and collect information on both exposure and outcome on a regular interval. And then we compare proportion of individuals who develop diarrhea among those who are exclusively breastfed and those who are not. Retrospective cohort study example, we form a cohort consisting of x-ray technicians and other technicians by reviewing the personal records of hospital. So we are actually going back to the past 20, 30, 40 or more years. And then how many among the x-ray technicians and how many among the. other technicians developed cancer over the period of time till now and have died. And we can use this data to compare the mortality experience of X-ray technicians with that of other technicians. This is a very nice graphical presentation of case control study and prospective and retrospective. course study designs. This is taken from the famous Epidemiology in Medicine book written by Charles Henneken and Julie Bering. Julie was actually our teacher when I was doing my master's at Harvard many years ago. Interestingly, she didn't teach her book. She was, she taught us a course, but the book that we followed was Rothman's Epidemiology. But this book is actually also very good, especially if you are interested in clinical epidemiology, you'll find this book very useful. Now, in this figure on top, this is an example of a case control study. And then we have prospective cohort and then we have retrospective cohort study designs. Now look at the man or woman standing where that person is standing. On top, in a case control study design, the person is standing at the very end when both exposure and outcome have occurred. So the person or the investigator or you, when you are starting the study, both exposure and outcome have occurred. You divide the participants as cases and controls, having the disease or not, and then you go further back. for their exposure history. The second one prospective cohort study design. The investigator is standing in the middle. Exposure has occurred. Outcome hasn't occurred yet. You divide the participants as exposed and unexposed. You follow them into the future for outcome. The third one retrospective cohort study design. The investigator is actually at the same point time point like case control study but he is or she is ignoring the disease status. The investigator is going farther back and dividing the participants as exposed and unexposed and then coming back to the present for disease. I hope this graph clarifies the different study designs well. Now which study design are you going to choose? Case control or cohort study design? It will depend on the nature of the disease under investigation, the type of exposure, and the available resources. If you don't have much resources, you would like to do a study quickly without spending much money. And if you don't have much time, you would like to do a case-control study because that will not require follow-up into the future. or a retrospective course study that will not require time and if the disease is for example very rare you would like to do a case control study otherwise you need to gather a large group of people to identify one case if you wanted to do a cohort study on the other hand if the exposure is rare what would you do or if the outcome is common like diarrhea you can go for a cohort study because you won't have to wait long and you will get many cases so it's up to you you need to think about the nature of the disease how common is that how long the follow-up period will need to be how much resources you have how much time you have hope you have found this session useful And I hope to come back to you and discuss more about epidemiology in future. And I'd like to thank you for watching. Bye.

I'm going to talk about case control study and cohort study designs. You might have seen this graph, this figure before. I showed this when I was talking, I was giving an overview of different epidemiological study designs in EpiMinutes 1 session, in the very first session of EpiMinutes. And so far we have discussed about case report. case series and cross-sectional study designs we have talked about correlational study designs and case report case series and cross-sectional study designs they are all descriptive study designs on the other hand correlational studies can only focus on groups not individuals now in the past we have discussed that in case report and case series we cannot test a hypothesis we cannot look for associations between an exposure and outcome because they don't have any comparison groups in cross-sectional studies we lack temporality because we take snapshots we don't know that's why we don't know whether exposure happens before the outcome and that's why we cannot prove any association and in correlational studies we work on groups of people we look we work on group level data and because of that we cannot apply the findings from our correlational studies on individuals now in both cohort and case control studies we actually deal with comparison groups we have comparison groups and that's why we can look for associations between an exposure and an outcome and that's why they are known as they are called analytic studies even though they are observational studies the investigators they just observe they do not intervene now in theory we can Apply either a case control study or a cohort study design to test a hypothesis.

In practice, we need to choose one because each of the designs have their unique advantages and disadvantages. So what is the key difference between these two study designs? The key difference lies in the process of subject selection or how we divide our study participants into different groups. Let's start with case control study, then it will become more clear to us. So in a case control study, we identify a group of individuals who have the outcome of interest and we call them as cases excuse me and we also identify a group of individuals who do not have the outcome of interest and we call them controls and then we look we compare the proportion with the exposure of interest between among the cases and controls between the cases and controls This is a very nice visual presentation of case control study design.

It is taken from Lancet has a very good series on different study designs. So this is taken from Lancet publication on study designs. They have a nice series on that. And if you look at the yellow box on the top right end, you can see that We are identifying a sample of cases. and we are at the present time so cases have the outcome of interest and then we go back we look back to the past and we try to identify among those cases who had exposure in the past who had the exposure of interest in the past in the second box the purple box you can see that we identify a sample of controls and also we look back and we try to identify among the controls who had exposure in the past and you also you also notice that directionality of the time which is very important so the time is moving forward but we are actually going backwards in a case control study design and both the cases and our controls are coming from population with outcome and population without outcome respectively.

This is another nice presentation, a simple presentation of case control study design and you can see that among the cases a certain proportion were exposed in the past, among the controls a certain proportions were exposed in the past and we make comparison between these two proportions. of exposures in the past in many different ways. We'll discuss them in future when we'll talk about measures of associations.

And also notice the directionality of the time and as investigators where we are, when the study is starting, when the onset of the study is happening. So the directionality of the time is going forward. On the other hand, the direction of our inquiry is actually going backwards to the past.

This is a great example and a very famous example of case control study design and in this study the exposure was smoking and the outcome was lung cancer. It was published in British Medical Journal in 1950 and this report this paper was written by Dr. Richard Dahl and Sir Bradford Hill. And you might remember Richard Dahl because in the very last session on ecological study we talked about a paper on meat consumption and colon cancer which was published in 1975 and Richard Dahl was one of the investigators for that study. and Sir Bradford Hill was a famous statistician and when we'll talk about causality you'll see that we will actually talk about Hill's criteria because he set a series of criteria to establish causality.

Now in this study they made a simple comparison They identified a group of lung cancer patients and they also collected data in terms of the smoking history among those cancer patients. And then they also identified a comparable group of participants without lung cancer. and then they collected information on their smoking history.

And in the table below, they nicely presented them. And then they did chi-square test and then they identified p-value and definitely they found very significant association between smoking and lung cancer. So this was a very early stage example of case control design and you'll see when we'll be talking about measures of association that we can actually come up with some other measures of associations in case control design.

Now, can you tell me what is this? Any idea? Yes, they are a group of soldiers.

and they're known as cohort which is a very common term in military and the cohort study this terminology came from this because in a cohort study we identify a group of people who march forward in time having an exposure and or not and then we follow them we observe them closely to see whether they also develop the outcome of interest in time so in a cohort study We divide the participants differently. We identify a group of individuals who are exposed to a particular factor of interest and we identify another group of individuals who are not exposed to a particular factor or exposure of interest. So those who are exposed and those who are not exposed.

We identify two groups of people. First we identify a cohort and then we divide them into two groups as exposed and unexposed and then we see whether they develop the outcome of interest so we follow them for a period of time and how long you are going to follow them it depends on the nature of the disease how common the disease is So this is a nice presentation, graphical presentation of a cohort study design. And you can see that we identify with a group of people first, a cohort. And at that time point, none of them were having the outcome of interest. And then we divide them as exposed and unexposed.

And then we follow them to see whether. Among the exposed, anyone is developing the outcome of interest. Among the unexposed, anyone is developing the outcome of interest or the incidence among the exposed and incidence among the unexposed. And look at the directionality of the time and the directionality of inquiry. They are going towards the same direction.

In a case control study, time was moving forward but our direction of inquiry was going backward. But here, Both time and inquiry have the same directionality. Now cohort studies can be of two types. They can be prospective cohort study or they can be retrospective cohort study. The prospective cohort study is also known as concurrent cohort study design.

And in a prospective cohort study, The investigation begins at a point in time when the exposure has occurred but the disease has not occurred yet. In a retrospective cohort study, the investigation begins at a point in time after both the exposure and disease have already occurred. Let's have a look at this graph. First, Let's have a look at the top purple line and look at now. So at present only exposure has occurred and now we are following the exposed and unexposed and the entire cohort into the future to see how many among the exposed develop the disease of interest and how many among the unexposed develop the disease of interest.

So this is a prospect. cohort study design or the concurrent cohort study design the purple line on the top right the red line look at now now both exposure and outcome have occurred we are ignoring outcome we are going back to the past we are going to the time point to the beginning when only exposure occurred None of the participants, none of the members of the cohort had outcome. And then we divide them as exposed and unexposed. And then we come back to the present to see what proportion of the exposed now is currently having outcome of interest.

What proportion among the unexposed currently is having the outcome of interest. So. this becomes a retrospective cohort design. So if you compare between the two In a prospective cohort design, now we are starting the study. None of the study participants have developed the outcome of interest.

We divide them as exposed and unexposed and we follow them into the future and we'd like and we'll see how many among the exposed develop the outcome of interest and how many among the unexposed develop the outcome of interest. In a retrospective cohort study, Now, outcomes have also occurred and of course exposure happened in the past. We are ignoring the outcome status.

We are going back to the past. This is how it is different than a case control study design. If it was a case control study design, we divide them as cases and controls, but we are ignoring that because we are designing a cohort study. So we need to start with exposed and unexposed. So we go back to the past when no one had the outcome.

We divide the participants of the cohort as exposed and unexposed and then we come back to the present and then see how many those who were exposed in the past are now having the outcome. What proportion among the unexposed in the past are now having the outcome of interest. Then we can compare incidents between the two groups that we cannot do in a case control study. and also if you want to continue into the future in your retrospective cohort study design then you can continue like the green line at the bottom then it becomes an ambidirectional study or retro prospective study examples of a prospective cohort study design we want to evaluate the possible association between exclusive breastfeeding and diarrhea in infants during the first six months of life. We identify a birth cohort, a group of children who are just born, follow them into the future and collect information on both exposure and outcome on a regular interval.

And then we compare proportion of individuals who develop diarrhea among those who are exclusively breastfed and those who are not. Retrospective cohort study example, we form a cohort consisting of x-ray technicians and other technicians by reviewing the personal records of hospital. So we are actually going back to the past 20, 30, 40 or more years. And then how many among the x-ray technicians and how many among the. other technicians developed cancer over the period of time till now and have died.

And we can use this data to compare the mortality experience of X-ray technicians with that of other technicians. This is a very nice graphical presentation of case control study and prospective and retrospective. course study designs. This is taken from the famous Epidemiology in Medicine book written by Charles Henneken and Julie Bering. Julie was actually our teacher when I was doing my master's at Harvard many years ago.

Interestingly, she didn't teach her book. She was, she taught us a course, but the book that we followed was Rothman's Epidemiology. But this book is actually also very good, especially if you are interested in clinical epidemiology, you'll find this book very useful.

Now, in this figure on top, this is an example of a case control study. And then we have prospective cohort and then we have retrospective cohort study designs. Now look at the man or woman standing where that person is standing.

On top, in a case control study design, the person is standing at the very end when both exposure and outcome have occurred. So the person or the investigator or you, when you are starting the study, both exposure and outcome have occurred. You divide the participants as cases and controls, having the disease or not, and then you go further back. for their exposure history.

The second one prospective cohort study design. The investigator is standing in the middle. Exposure has occurred.

Outcome hasn't occurred yet. You divide the participants as exposed and unexposed. You follow them into the future for outcome. The third one retrospective cohort study design.

The investigator is actually at the same point time point like case control study but he is or she is ignoring the disease status. The investigator is going farther back and dividing the participants as exposed and unexposed and then coming back to the present for disease. I hope this graph clarifies the different study designs well.

Now which study design are you going to choose? Case control or cohort study design? It will depend on the nature of the disease under investigation, the type of exposure, and the available resources.

If you don't have much resources, you would like to do a study quickly without spending much money. And if you don't have much time, you would like to do a case-control study because that will not require follow-up into the future. or a retrospective course study that will not require time and if the disease is for example very rare you would like to do a case control study otherwise you need to gather a large group of people to identify one case if you wanted to do a cohort study on the other hand if the exposure is rare what would you do or if the outcome is common like diarrhea you can go for a cohort study because you won't have to wait long and you will get many cases so it's up to you you need to think about the nature of the disease how common is that how long the follow-up period will need to be how much resources you have how much time you have hope you have found this session useful And I hope to come back to you and discuss more about epidemiology in future. And I'd like to thank you for watching.

Bye.

Transcript for:case control and cohort

Transcript for:
case control and cohort