Disease Detectives B/C

[Flavorflav]

Hey does anyone have the formula for Sensitivity and Specificity?

http://en.wikipedia.org/wiki/Sensitivit ... pecificity

[ohafer]

Can anyone tell me how to calculate infectious rate or tell me where I can learn (give me a link or something)?

[Flavorflav]

Are there any "giveaways" for different study designs?
I'm mainly looking for obvious characteristics of:

Ecological
Case-control
Cohort
Randomized controlled Trial

Thanks!

Ecological analysis compares two or more populations, generally based on place of residence - for example, residents of town A or town B. It is similar to a cohort study, but a cohort is more likely to be (for example) the 2011 graduating class of the High School in town A vs. some other similar group. Cohort is a stronger design, since the cohort is chosen because of some hypothesized common exposure (chemical contamination of the high school, for example) whereas in the ecological analysis you have very little information about the actual exposure of town residents (people could live there without being exposed to the chemicals in the high school - dropouts, private or home schooled students, etc.) The case-control is easy because you have a group of cases and a group of controls, and the RCT is easy because the experimenter is manipulating something, usually a medicine, although on occasion you will run into an experiment in which subjects were intentionally exposed to an infectious agent, such as the series of experiments from the 50's and 60's that demonstrated that exposure to cold temperatures did not increase the risk of contracting a cold.

Can anyone tell me how to calculate infectious rate or tell me where I can learn (give me a link or something)?

There is no such thing as infectious rate - a disease is either infectious or it isn't. Do you mean infectivity? http://en.wikipedia.org/wiki/Infectivity
Attack rate? http://en.wikipedia.org/wiki/Attack_rate
Or do you mean virulence? http://en.wikipedia.org/wiki/Virulence

[rkoopma2]

They seem similar, but a cohort study starts with healthy individuals and assesses response to exposure (or lack thereof), while a case control study starts with individuals with the disorder, and collects historical data.

While not inaccurate that is a bit misleading, since a retrospective cohort also looks at historical data. The key distinction is that a cohort study follows a defined group over time, either really (as in the prospective cohort) or virtually (in the retrospective). A case-control is more of a snapshot, comparing cases to controls for some factor of interest. There is usually only one defined group in a cohort study, since you are generally comparing + and - for some factor within the cohort, while a case-control study always has at least two groups - the cases and the controls.

true, but the main distinction is that cohort studies define groups based on exposure status and track whether they develop the disease, while a case-control study defines groups based on illness and then determines exposures.

[rkoopma2]

is matching the date/place of a historical outbreak with the agent/disease causing it really disease detectives. It doesn't seem like it. not that i'm complaining, beat the second place team by 23 points at regionals

[The Eviscerator]

Could someone explain what the odds ration actually means? I know how to calculate it and all, but I want to know what it numerically represents.
Like how a relative risk of 3.8 for eggs and salmonella would mean that someone would be 3.8 times more likely to get salmonella if they ate the eggs.

[kjhsscioly]

it is the increased quantity of risk that one would be in if they were exposed vs. non exposed. For your example, it would mean that people who were exposed to the eggs were 3.8 times more likely to contract salmonella than those people in the same conditions who did not eat eggs. A low relative risk, such as lower than one would not indicate a correlation; a high one would.

[Flavorflav]

They are the same thing if the sample is large enough. With a smaller sample, odds ratio tends to be higher than RR.

[The Eviscerator]

it is the increased quantity of risk that one would be in if they were exposed vs. non exposed. For your example, it would mean that people who were exposed to the eggs were 3.8 times more likely to contract salmonella than those people in the same conditions who did not eat eggs. A low relative risk, such as lower than one would not indicate a correlation; a high one would.

uhhh... I was asking what the odds ratio means. I already know relative risk.

Edit: Why would they be the same thing if odds ratio is used for case control studies and relative risk is used for cohort studies?

[gleekymel0n]

Any suggestions on what to put on our notes? So far we've got: vocab, diseases, formulas, study designs. Missing anything else??