Disease Detective B/C

[butter side up]

Does anyone have any tips on extracting important data from the walls of text the test tends to have quickly and accurately? Also, I frequently encounter not even being able to find information for a 2x2 table, so can anyone give me hints for when the numbers aren't obvious? Thanks in advance.

Unfortunately, those walls of text are just a fact of life. I would suggest doing a preliminary skim to look for relavant details [try reading the pertaining questions first so you have an idea of what you are looking for] and underlining them. [Be sure to ask the supervisor before you mark the test.] Often this will allow you to find what you need fairly quickly. However, if you have time, read everything. I have encountered tests where we were unsure about the type of study it was, and then found a line at the bottom of the page reading "from a cohort study preformed by...", so details are often helpful. As far as the 2x2 goes, the data is usually there, but is not always given as raw numbers. You may have to do some subtraction or addition to find all of them.
Hope this helps!

[liutony66]

I have three main questions:

Will properties/symptoms/other information about various diseases be present on the test?

How do we know if an outbreak is "major" or not? The wikipedia list of epidemics is, well, quite gargantuan.

And finally, will the tests depend a lot upon the speed you are able to do the problems?

Thanks a lot!

[Flavorflav]

Can someone explain the difference between Cohort and Case Control? For instance, when one would be used as opposed to the other? I know one compares the exposed and non-exposed vs those with the disease and not - however, what is the real difference between Odds ratio and Relative Risk?

While they are used in two different scenarios, it's important to understand that they are basically the same thing, but from different "points of view." With odds ratio, you are comparing the odds of getting infected from the exposure, and with relative risk you are comparing the risk of being infected from exposure to the risk of being infected without exposure. They are basically trying to find the same thing. Besides knowing which type of study to use them in, you should also think which is necessary. In a case control, you are working backwards from the cases. You have determined (a) source, and you are trying to find the significance of that source in causing disease. With a cohort, you are working forward from the source, and to determine whether disease followed a certain path, you must find the significance of disease onset from that path.

In addition, I'd suggest also looking into things like absolute risk, and other measures of risk, such as excess risk, and risk difference.

Excess Risk: [(a+c)/a]-[(a+c)/b]
Risk Difference: [a/(a+c)]-[b/(b+d)]

Besides risk, and at a high school level, mathematics becomes extremely hard, and you'll be asked to both figure out and calculate numerous measures. Here are some I've commonly seen:

Cumulative Incidence Ratio: Number of new cases in a period of time / total population at risk (average population if changing w/time)
Point Prevalence: Cases in population / total population
Period Prevalence: Number of new cases in a certain time period / average population at the time
Case Fatality: Deaths / Total Population with disease [it's expressed per period of time]
Mortality Rate: Deaths / Total Population [it's expressed per period of time]
Rate Ratio: Incidence rate of a certain population / incidence rate of another population

And most of those should be expressed by per 10,000 or 100,000 people (whichever is specified or more appropriate)

If I wrote any of them wrong, please correct me, these are from the top of my head, and if there are others, please post them.

Thanks

A few corrections: technically, in a case-control you are calculating the odds of having been exposed for "infected" (cases) and controls, not the odds of getting infected for exposed people. That is actually the basic difference between the two designs, as you point out in the next several sentences..

I have not seen excess risk calculated that way. It is usually the simple difference between risk in group 1 and risk in group 2 - or, in terms of the 2x2, (a/a+b)-(c/c+d)*100%. I suspect you mean this when you say risk difference, but you seem to be lettering your 2x2 in a nonstandard way - generally, it reads across a-b- a+b, then c-d-c+d on the next line. Your risk difference only makes sense if you are reading down, instead.
Period prevalence refers to existing cases, not new cases.
Case Fatality Rate includes only Cause Specific Deaths.


Also, a general note: relative risk (which is almost always the same thing as rate ratio) is not always exposed v. unexposed. Very often, it is simply exposed v. the general population. That is less likely for foodborne, but it used to trip people up all the time in pop. growth.

[Epidemiologist]

Can someone explain the difference between Cohort and Case Control? For instance, when one would be used as opposed to the other? I know one compares the exposed and non-exposed vs those with the disease and not - however, what is the real difference between Odds ratio and Relative Risk?

While they are used in two different scenarios, it's important to understand that they are basically the same thing, but from different "points of view." With odds ratio, you are comparing the odds of getting infected from the exposure, and with relative risk you are comparing the risk of being infected from exposure to the risk of being infected without exposure. They are basically trying to find the same thing. Besides knowing which type of study to use them in, you should also think which is necessary. In a case control, you are working backwards from the cases. You have determined (a) source, and you are trying to find the significance of that source in causing disease. With a cohort, you are working forward from the source, and to determine whether disease followed a certain path, you must find the significance of disease onset from that path.

In addition, I'd suggest also looking into things like absolute risk, and other measures of risk, such as excess risk, and risk difference.

Excess Risk: [(a+c)/a]-[(a+c)/b]
Risk Difference: [a/(a+c)]-[b/(b+d)]

Besides risk, and at a high school level, mathematics becomes extremely hard, and you'll be asked to both figure out and calculate numerous measures. Here are some I've commonly seen:

Cumulative Incidence Ratio: Number of new cases in a period of time / total population at risk (average population if changing w/time)
Point Prevalence: Cases in population / total population
Period Prevalence: Number of new cases in a certain time period / average population at the time
Case Fatality: Deaths / Total Population with disease [it's expressed per period of time]
Mortality Rate: Deaths / Total Population [it's expressed per period of time]
Rate Ratio: Incidence rate of a certain population / incidence rate of another population

And most of those should be expressed by per 10,000 or 100,000 people (whichever is specified or more appropriate)

If I wrote any of them wrong, please correct me, these are from the top of my head, and if there are others, please post them.

Thanks

A few corrections: technically, in a case-control you are calculating the odds of having been exposed for "infected" (cases) and controls, not the odds of getting infected for exposed people. That is actually the basic difference between the two designs, as you point out in the next several sentences..

I have not seen excess risk calculated that way. It is usually the simple difference between risk in group 1 and risk in group 2 - or, in terms of the 2x2, (a/a+b)-(c/c+d)*100%. I suspect you mean this when you say risk difference, but you seem to be lettering your 2x2 in a nonstandard way - generally, it reads across a-b- a+b, then c-d-c+d on the next line. Your risk difference only makes sense if you are reading down, instead.
Period prevalence refers to existing cases, not new cases.
Case Fatality Rate includes only Cause Specific Deaths.


Also, a general note: relative risk (which is almost always the same thing as rate ratio) is not always exposed v. unexposed. Very often, it is simply exposed v. the general population. That is less likely for foodborne, but it used to trip people up all the time in pop. growth.

Yes, you're right for most of that. I typed that on my phone, thanks for the corrections.

[SONinjaGuy]

Does anyone have any tips on extracting important data from the walls of text the test tends to have quickly and accurately? Also, I frequently encounter not even being able to find information for a 2x2 table, so can anyone give me hints for when the numbers aren't obvious? Thanks in advance.

Unfortunately, those walls of text are just a fact of life. I would suggest doing a preliminary skim to look for relavant details [try reading the pertaining questions first so you have an idea of what you are looking for] and underlining them. [Be sure to ask the supervisor before you mark the test.] Often this will allow you to find what you need fairly quickly. However, if you have time, read everything. I have encountered tests where we were unsure about the type of study it was, and then found a line at the bottom of the page reading "from a cohort study preformed by...", so details are often helpful. As far as the 2x2 goes, the data is usually there, but is not always given as raw numbers. You may have to do some subtraction or addition to find all of them.
Hope this helps!

Yes this was helpful, or at least as helpful as anyone really could have been. I talked to my partner and this is basically what he said. We pretty much just need to work on speed.

[butter side up]

I have three main questions:

Will properties/symptoms/other information about various diseases be present on the test?

How do we know if an outbreak is "major" or not? The wikipedia list of epidemics is, well, quite gargantuan.

And finally, will the tests depend a lot upon the speed you are able to do the problems?

Thanks a lot!

1. They are not really common, but I find that they do show up with some regularity. I would devote maybe 1/2 to 1/4 of a page to a chart with common food-bourne illnesses and other important ones (E. coli 0157:H7, salmonella,cholera, influenza, etc...) and their agents (this is really important- it shows up a lot- bacteria/virus/prion/etc). A summary of their symptoms, treatments, and likely sources is also helpful, as are some quick notes, like important outbreaks.
2. A "major outbreak" depends on the context, but I've never seen a question on if an outbreak qualifies as major. I suppose it would just be dependant on the mortality/ morbity rate. It might be referring to the famous historical out breaks, like John Snow's cholera outbreak.
3. You don't need to rush, but you should be able to do the problems pretty quickly. Most of these tests are reasonable lengths, but it isn't one of those that you have to speed-write nonstop. I suppose at some point you will be going too slow just based on common sense, but it isn't usually too big of a deal. Most people will finish the tests in time.

[Sherry Berry]

Question:

I've never done this event before, and I was wondering if they ask you about basic epidemiology questions at competition? Or is the whole test JUST about Food Borne Illnesses?

[akatsuki9]

From my experience, they'll ask you epidemiology questions as well.

[Sherry Berry]

Thanks.

Now I need to modify my cheat sheet...

[danny9696]

Can someone check the validity of the problem situation (outbreak of Norovirus) in this test? http://scioly.org/w/images/f/f2/Dis.pdf
The answer key says it uses a case-control test, however (http://scioly.org/w/images/b/b3/Dis-sol.pdf) all of the details of the outbreak seem to indicate that it should use a cohort test.

Thanks!