Disease Detectives B/C

cosine
Member
Member
Posts: 2
Joined: December 18th, 2012, 4:38 pm
Division: C
State: CA
Has thanked: 0
Been thanked: 0

Re: Disease Detectives B/C

Post by cosine » March 8th, 2014, 5:31 pm

Is odds ratio the only thing used for case-control studies?

And the equation is ad/bc, correct? What if the denominator is 0? Is infinity a valid odds ratio? For example, take the two following tables:

Exposed to substance 1 Case patients Controls Total
Yes a = 30 b = 20 50
No c = 0 d = 5 5

Odds ratio: 150/0

Exposed to substance 2 Case patients Controls Total
Yes a = 30 b = 10 40
No c = 0 d = 15 15
Odds ratio: 450/0

In both cases, they are infinity. It does look to me that substance 2 is the more likely cause though.Am I interpreting these wrong?

Flavorflav
Member
Member
Posts: 1376
Joined: February 5th, 2006, 7:06 am
Division: Grad
State: NY
Has thanked: 0
Been thanked: 1 time

Re: Disease Detectives B/C

Post by Flavorflav » March 8th, 2014, 7:00 pm

There are several methods for that. One is to add 0.5 to every cell, another is to add it to only the cell with the 0. Either way you aren't doing a pure Odds Ratio, but a Modified Maximum Likelihood Estimator of the OR.

ETA: you shouldn't see that situation arise in competition, though.

User avatar
hmath729
Member
Member
Posts: 102
Joined: February 21st, 2014, 4:27 pm
Division: C
State: CA
Location: Helix Heights
Has thanked: 0
Been thanked: 0

Re: Disease Detectives B/C

Post by hmath729 » March 9th, 2014, 5:35 am

We had regionals a few weeks ago and we got that exact problem - the denominator was zero. I don't get it. :roll:
Go HVS Dolphins!
"You don't beat the tests. The tests beat you!"
Medals (10 total) Inv./Reg./State
2016
Anat. -/4th/-
BPL 1st!/1st!/-
DD. -(9th)/1st!/-
2015
Anat. -(8th)/4th/-
BPL 2nd!/5th/-
DD -(9th)/4th/-
2014
Anat. -(8th)/5th/-
DD -(9th)/5th/-
Heredity -(7th)/-(7th)/-

hscmom
Coach
Coach
Posts: 242
Joined: March 3rd, 2010, 8:41 pm
Division: Grad
State: CO
Location: Colorado
Has thanked: 0
Been thanked: 0

Re: Disease Detectives B/C

Post by hscmom » March 9th, 2014, 1:06 pm

Flavorflav wrote:There are several methods for that. One is to add 0.5 to every cell, another is to add it to only the cell with the 0. Either way you aren't doing a pure Odds Ratio, but a Modified Maximum Likelihood Estimator of the OR.

ETA: you shouldn't see that situation arise in competition, though.
Our teammate (who's done this a bit longer than we have) saw it in competition (and kudos to him for knowing what to do -- he was probably 11 or 12 at the time!).
Homeschool Science Colorado since August 2008

Flavorflav
Member
Member
Posts: 1376
Joined: February 5th, 2006, 7:06 am
Division: Grad
State: NY
Has thanked: 0
Been thanked: 1 time

Re: Disease Detectives B/C

Post by Flavorflav » March 13th, 2014, 5:00 pm

Experts disagree on what to do in that situation, so I don't think it is a very good question for students.

GrayEpi
Member
Member
Posts: 25
Joined: June 21st, 2013, 8:50 pm
Division: Grad
State: GA
Has thanked: 0
Been thanked: 0

Re: Disease Detectives B/C

Post by GrayEpi » March 19th, 2014, 7:11 pm

Case-control studies and odds ratios - While there may be measures of association other than odds ratios used for case control studies, I don't know what they might be. There are a number of ways to handle 0 cell tables - I was taught to add 0.5 or 1 to each cell. I don't know that we have ever had a 0 cell table in a national exam. I think if one were to show up we would either give some basic instruction on how to deal with it, provide a reference beforehand or be flexible in how we graded it. While it is pretty safe to say if it is a case-control study, the measure of association will be an odds ratio, the reverse is not the case. One often sees odds ratios used as measures of association for cohort studies. This is because the methods used in cohort studies to control for more than one or two variables (we just started touching on that a bit in the new rules with adjusted rates and stratified analysis) start to fall apart (think 0's in cells) with traditional relative risk types of analyses. Logistic regression and other multivariable techniques can be used in cohort studies but they give estimates of odds ratios. If you see case-control study - think odds ratio. If you see relative risk - think cohort type study. If you see odds ratio - it could be either. In that case check on how the study was set up. If they started with a group of folks who had the outcome of interest and then selected a group who did not have the outcome of interest - it is a case-control study. If they started with a group who had an exposure of interest and a group who did not have the exposure of interest or a mixed group - some of whom had the exposure and some who did not, it is a cohort or cross-sectional study.

There is a fundamental difference between the Disease Detectives event and an exam. I taught night classes for many years and nothing would have made me happier than to have every student in my class get a 100 on one of my exams (never happened). However, it would be a disaster for a Disease Detectives event if even two teams were to get 100s. Just as no one expects an Olympic weight lifter to stack on and lift every plate on the platform, we don't expect a team to answer every question perfectly. We expect competitors to do their best, to take the event seriously and play fair. You do that and I promise you will have the respect of the event supervisors.

Gearbox
Member
Member
Posts: 40
Joined: March 20th, 2012, 10:47 am
Division: C
State: PA
Location: Shady Side Academy
Has thanked: 0
Been thanked: 0

Re: Disease Detectives B/C

Post by Gearbox » March 20th, 2014, 6:57 am

Quick question, I have been doing this event for 4 years now but still can't quite get a grasp on different types of analytic studies. What are the differences and how do they help. Thanks
2015 Events

Protein Modeling
Cell Biology
Disease Detectives
Chem Lab

Image

BipolarEconomist
Member
Member
Posts: 6
Joined: January 18th, 2014, 7:20 pm
Division: C
State: CT
Has thanked: 0
Been thanked: 0

Re: Disease Detectives B/C

Post by BipolarEconomist » March 20th, 2014, 9:54 pm

Hello!
With states soon approaching, I have some last minute questions with which I'm a bit obscure.

First, an someone clarify what exactly the epidemiological triad is?
I thought it was person/place/time, but there's now something called a chain of transmission/infection triad, and that's getting seriously confusing.

Also, what exactly is the method for finding the incubation period of a disease by looking at the EPI curve diagram? I know that intervals should be spaced at 1/4 of average incubation period, but then how do some of the answers end up being a range such as 6-12 hours instead of a definite number?

Finally, how do you distinguish between the three types of EPI curves? I know point source has only one peak and a very steep downward slope, continuous common source is a bit more gradual, and propagated (progressive source) has regular intervals, and I try to look at those, but THEY ALL LOOK THE SAME TO ME! Any specific advice regarding x/y-axes or better ways of looking at graph shape to distinguish types? Thanks.

Also, what are you guys going to put on your double-sided paper of information that you're allowed to bring? I'm thinking of including Hill's 9 Criteria for Causation and perhaps the 10 Steps to an Outbreak Investigation along with some common diseases in case the test asks about pairing diseases to causes.

By the way, how do you guys recommend studying for questions where it gives you a bunch of decently significant diseases in the past and asks you to classify them as bacterium, prions, viruses, parasites, etc.? Is there a simple resource with some classification charts?

Thanks! All help is greatly appreciated!!
2016: Fossils, Chem Lab, Astronomy

jwalkotten
Member
Member
Posts: 14
Joined: February 4th, 2014, 5:20 pm
Division: C
State: MI
Has thanked: 0
Been thanked: 0

Re: Disease Detectives B/C

Post by jwalkotten » March 21st, 2014, 3:33 pm

BipolarEconomist wrote:Hello!
With states soon approaching, I have some last minute questions with which I'm a bit obscure.

First, an someone clarify what exactly the epidemiological triad is?
I thought it was person/place/time, but there's now something called a chain of transmission/infection triad, and that's getting seriously confusing.

Also, what exactly is the method for finding the incubation period of a disease by looking at the EPI curve diagram? I know that intervals should be spaced at 1/4 of average incubation period, but then how do some of the answers end up being a range such as 6-12 hours instead of a definite number?

Finally, how do you distinguish between the three types of EPI curves? I know point source has only one peak and a very steep downward slope, continuous common source is a bit more gradual, and propagated (progressive source) has regular intervals, and I try to look at those, but THEY ALL LOOK THE SAME TO ME! Any specific advice regarding x/y-axes or better ways of looking at graph shape to distinguish types? Thanks.

Also, what are you guys going to put on your double-sided paper of information that you're allowed to bring? I'm thinking of including Hill's 9 Criteria for Causation and perhaps the 10 Steps to an Outbreak Investigation along with some common diseases in case the test asks about pairing diseases to causes.

By the way, how do you guys recommend studying for questions where it gives you a bunch of decently significant diseases in the past and asks you to classify them as bacterium, prions, viruses, parasites, etc.? Is there a simple resource with some classification charts?

Thanks! All help is greatly appreciated!!
Your other triad is agent/host/environment
Your incubation is the time period from the initial exposure to your first case as your minimum, to the time from initial exposure to last case. it's a range
for epi curves...its basically those shapes. maybe just look at some more examples
we put on those things, diseases and their causes, formulas, any helpful notes. we use a small font and cram :)
for those...I guess your disease detectives team has built up a list over the years. any time you get a new one on a test, add it to the list. any time you get a new one on a practice test or case study, add it to the list. by now we have a list of about 75. so basically just keep a growing file of any diseases, then classify those, and put them on your note sheet (using alphabetical order can help if you have a lot)
hope this helps!
2013 Regionals - 1st Overall, 1st Exp. Design, 3rd Anatomy, 5th Disease Detectives
2013 State - 4th Overall, 3rd Exp. Design, 3rd Disease Detectives, 5th Shock Value, 7th Anatomy

GrayEpi
Member
Member
Posts: 25
Joined: June 21st, 2013, 8:50 pm
Division: Grad
State: GA
Has thanked: 0
Been thanked: 0

Re: Disease Detectives B/C

Post by GrayEpi » March 28th, 2014, 8:48 pm

jwalkotten - Right on with the second triad. We try to tie the person-place-time with descriptive epi while agent-host-environment links with chain of transmission.

Using the interval from first to last case on an epi curve to estimate incubation period works for point source outbreaks with no secondary cases and time reflects onset rather than report date. The idea is that everybody on the curve was exposed at the same time.

For continuing source outbreaks - the duration of exposure messes things up - the onset of the first and last cases reflect the start and end of the exposure within the limits of the minimum and maximum incubation periods (the first person sick may not have been the first exposed if they had a very short incubation period).

Epi curves for propagated outbreaks can also give estimates in the early stages of the outbreak. These curves often have a saw-tooth appearance (each peak is usually a bit higher than the preceding ones as things get worse) the distances between the peaks reflecting waves of transmission can give a rough estimate of the incubation period. However incubation periods have a mean and variance and as the outbreak progresses and more cases occur with each generation - these eventually blur into one humongous mess as the longest incubations from two waves before get sick about the same time as the shortest incubation from the preceding waves.

Hope this all makes sense - it really does work.

Sorry about the huge lists of diseases and causes. That is really not what this event was meant to be about. Understanding the difference between the main groups of pathogens and how they come into play is an integral part of epi - for example bacteria will grow in potato salad - viruses will not (KNOW WHY?). In working up a foodborne outbreak - storage and handling is especially important in bacterial outbreaks as they can let small inocula grow - in viral outbreaks - they are somewhat less important as what you put in is all you are ever going to get out. Knowing that transmissible mink encephalopathy is a prion disease or that hand-foot-and mouth disease in children is caused by an enterovirus while foot and mouth disease in animals is caused by an aphthovirus is a bit like trivial pursuit.

Locked

Return to “2014 Study Events”

Who is online

Users browsing this forum: No registered users and 1 guest