From the Times of India
1. What study design should we use to identify the cause(s)? Can you explain why other epidemiologic designs aren't suitable?
2. What are the first few decisions epidemiologists should make as they design this study?
1. I would use a case-control study choosing age and sex matched individuals from the same area as my controls. An interventional trial in this setting is unethical as you would either 1) expose individuals to an agent that would cause harm or 2) withhold treatment for suffering individuals. Cohort studies would not fit as you cannot assume the population will be exposed again in the future and should, in fact, use the data you have currently to prevent this from happening.
2. I think the biggest decision is to pick appropriate controls that will accurately represent the prevalence of risk factors in the population from which controls were drawn.
Hi everybody...
Another interesting read about a current panic in India...
1) Since this is a disease unknown previously with a limited number of cases, the most appropriate study design to reach the cause or exposure would case- control study. Other designs are not suitable. For a cohort study, we need to start with an exposure- here the cause or exposure is unknown and so a cohort study is not practical. An interventional study would mean exposing healthy ppl to a possible cause like organochlorines and then look for development of the disease- which is of course unethical. Another option would be to try an antidote to a possible exposure and see the response in cases- but the suspected substance here is organochlorines- that don't have an antidote and is only managed by symptomatic treatment. So this option is also not viable. I am not sure if an ecological study can be done here (comparing this particular town with other neighboring towns). I was wondering what is happening different in this particular town when compared to the neighboring areas/towns. A group level analysis might help with that.
2) Here the most important decision would be to select the appropriate controls. The disease has been seen in one particular town only. So the controls must be from this same town, at least age matched, and maybe matched according to the locality where they live as well. Another decision to be made is what exposures to look at, how to define these exposures, and how to collect information about these possible exposures. For example, this article tells that they looked at water and milk samples to look for any contaminants. Similarly they will need to define what to look for.
Fathima
Another interesting read about a current panic in India...
1) Since this is a disease unknown previously with a limited number of cases, the most appropriate study design to reach the cause or exposure would case- control study. Other designs are not suitable. For a cohort study, we need to start with an exposure- here the cause or exposure is unknown and so a cohort study is not practical. An interventional study would mean exposing healthy ppl to a possible cause like organochlorines and then look for development of the disease- which is of course unethical. Another option would be to try an antidote to a possible exposure and see the response in cases- but the suspected substance here is organochlorines- that don't have an antidote and is only managed by symptomatic treatment. So this option is also not viable. I am not sure if an ecological study can be done here (comparing this particular town with other neighboring towns). I was wondering what is happening different in this particular town when compared to the neighboring areas/towns. A group level analysis might help with that.
2) Here the most important decision would be to select the appropriate controls. The disease has been seen in one particular town only. So the controls must be from this same town, at least age matched, and maybe matched according to the locality where they live as well. Another decision to be made is what exposures to look at, how to define these exposures, and how to collect information about these possible exposures. For example, this article tells that they looked at water and milk samples to look for any contaminants. Similarly they will need to define what to look for.
Fathima
Hey folks, totally agreed on use of a case-control study and the considerations from John and Fathima on this.
As well as selection of suitable controls, I think identification of relevant risk factors may be a key challenge. It might be useful to conduct an unstructured qualitative interviews with cases in order to bottom-up construct hypotheses for potential causal factors. Cases could be interviewed (noting need for parental approval likely required, given age of patients) with thematic coding of line-by-line from responses. This could help identify any risk factors that should be measured in cases and controls. It could be that the biochemistry and other research also helps create hypotheses!
As well as selection of suitable controls, I think identification of relevant risk factors may be a key challenge. It might be useful to conduct an unstructured qualitative interviews with cases in order to bottom-up construct hypotheses for potential causal factors. Cases could be interviewed (noting need for parental approval likely required, given age of patients) with thematic coding of line-by-line from responses. This could help identify any risk factors that should be measured in cases and controls. It could be that the biochemistry and other research also helps create hypotheses!
Hi everyone!
1. What if we used 2 type of studies? let say a cross sectional study first, and then a case control study.
First, we design a cross sectional study, we define cases from hospital records, then we identify cases in the community. From this data, we can define potential exposures based on characteristics from cases: food, work, activities, places of frequentation. Then we can design a case control study as was mentioned by John, Fathima and Bethany.
2. I agree in all mentioned by my fellow colleagues, I would only add we would like to perform a quick intervention in order to stop as early as possible the exposure to the causal agent.
Cheers,
Nico
1. What if we used 2 type of studies? let say a cross sectional study first, and then a case control study.
First, we design a cross sectional study, we define cases from hospital records, then we identify cases in the community. From this data, we can define potential exposures based on characteristics from cases: food, work, activities, places of frequentation. Then we can design a case control study as was mentioned by John, Fathima and Bethany.
2. I agree in all mentioned by my fellow colleagues, I would only add we would like to perform a quick intervention in order to stop as early as possible the exposure to the causal agent.
Cheers,
Nico
That's a very interesting thought Nico... And frankly speaking, i think that is what will practically happen as well... thank you for pointing it out... Even to think about a case-control, we would require a cross-sectional idea about the possible exposures in the area .. makes sense...!
Fathima
Fathima
I think we would need to do an outbreak investigation. As it looks like a completely new disease , we would have to define what we can consider as a suspected case.
Then we can find out the time, place and person characteristics to find out if they all had some kind of common exposure which can link them.
We can generate hypothesis by this kind of cross sectional study and maybe later confirm them by case control study.
Doing a case control study in middle of an outbreak I guess would be difficult as we wouldn't be able to choose controls. We wouldn't have enough information to match cases and controls.
Interventional study as pointed out earlier in the discussion is unethical and not feasible.
Then we can find out the time, place and person characteristics to find out if they all had some kind of common exposure which can link them.
We can generate hypothesis by this kind of cross sectional study and maybe later confirm them by case control study.
Doing a case control study in middle of an outbreak I guess would be difficult as we wouldn't be able to choose controls. We wouldn't have enough information to match cases and controls.
Interventional study as pointed out earlier in the discussion is unethical and not feasible.
Hi all
if were knew a suspected causative factor or several of themm, could a historical cohort be done if the town is not too big to make this unfeasible?
if were knew a suspected causative factor or several of themm, could a historical cohort be done if the town is not too big to make this unfeasible?
otherwise a case control, but I agree with the limitation above, it could be tricky to select controls that have also been exposed but have a different outcome as a result..if we dont have a clear exposure in mind yet. Including such controls would underestimate or even hide an association
but if our range of potential causative factors is limited and their effects are well known then it could be easier to exclude controls manifesting conditions linked to the exposure .
Olga