#####
GW CREOG Video Project: Learn about CREOG Statistics by Dr. Julia Whitley

This video is going to be a review of common statistics topics for the kriya exam first we’re going to go over different characteristics of screening tests including sensitivity specificity and positive and negative predictive value so first to talk about sensitivity we have our two by two table there sensitivity is the proportion of true positive tests out of all

The patients who have a condition so in other words it’s the ability of a test to yield a positive result for a patient who actually has the disease which is very important for screening tests because you want your screening test to be highly sensitive to identify as many patients with the disease as possible to calculate this you’re going to divide the number of

True positives out of all the patients who have the disease or the true positives plus the false negatives or a over a plus c another way to think about this is that the false negative rate is going to be one minus the sensitivity specificity is the percentage of true negatives out of all patients who do not have the disease or in other words it’s the ability of

The test to obtain normal results for a patient who does not have a disease to calculate this you’re going to divide the number of true negatives out of all the patients who do not have a disease or false positives plus true negatives so d over b plus d and similarly the false positive rate is going to be one minus the specificity the positive predictive value and

The negative predictive value are dependent on the prevalence of the disease and the population that you’re studying and the positive predictive value tells you out of all the positive results that you get how many are actually true positives so to calculate this you divide the number of true positives out of all the positive results so true and false positives

And in other words a over a plus b the negative predictive value similarly is dependent on the prevalence of the disease in the population that you’re studying and can be calculated by dividing the number of true negatives out of all the patients who tested negative so d over c plus d likelihood ratios are another tool to understand the results of the test that you

Ordered because it tells you how much the use of that test will alter the probability of the patient having the disease that you’re testing for so for example the positive likelihood ratio is the probability that a positive test would be expected in a patient with the disease divided by the probability that a positive test would be expected in a patient without a

Disease so you can think of this as the sensitivity over the one minus specificity or the true positive rate over the false positive rate and a higher likelihood ratio or a likelihood ratio that’s greater than one is going to increase the likelihood of your patient having the disease that you’re testing for the negative likelihood ratio is the probability of the

Patient testing negative who has the disease divided by the probability of a patient testing negative who does not have the disease you can also think about this as the false negative rate over the true negative rate or one minus sensitivity over the specificity now i’m going to talk about different study designs that you may be asked about on the creogs the

First two that we’re going to talk about are observational studies and these are case control and cohort studies are the most commonly tested observational studies so the case control study is where you’re choosing patients with a disease comparing them to patients without that disease and looking backwards to look at their exposures and you’re this is going

To yield an odds ratio or the odds of disease in the patients or the odds of exposure in the patients who went on to develop the disease divided by the odds of that exposure in patients without the disease which boils down to a times d over b times c a cohort study is similar but you’re looking forwards so you’re choosing patients with risk factors and then

Following them to see if they develop the disease in question and then you’re going to calculate a relative risk or the risk of disease in patients who are exposed divided by the risk of disease in patients without exposure or a over a plus b divided by c over c plus d the what sort of the as the gold standard for a lot of research questions is the clinical

Trial specifically a randomized control trial this is where you’re comparing the placebo group compared to the intervention group and this may or may not be blinded but these are more time intensive and difficult to conduct there are different forms of bias that you may be tested on in the creogs most these are the most common types of bias that we are asked about

Here so one of these is selection bias and this is when the participants in your study are seem to be biased in some way and that they are not representative of the population that you’re intending to study so for example if you meant to study a population of all 20 year olds about sleep deprivation but for whatever reason most of the participants in your study

Are your co-residents that’s going to yield different results because of different sleep patterns in residence compared to the general population recall or survivorship bias is when you’re asking is when you’re studying patients who developed a particular disease for example and you’re asking them to recall their exposures and patients who recently developed

A disease may be more likely to report certain exposures because they may for some reason be attributing their disease to specific exposures so for example asking women whose children are diagnosed with neurodevelopmental delay about any medications that they took during pregnancy and they may just looking through anything that they might have taken and then

You may attribute an outcome to tylenol that may or may not be accurate observation bias or the hawthorne effect is when study participants behave differently when they’re being observed in a study and that may influence the results of the study and now i’m going to end with talking about type 1 and type 2 error so there’s a quick and easy way that someone taught

Me to remember this that has stuck with me but type 1 error is what i think of this is a false positive error so you found a difference in your study but actually there’s no difference and an easy way to remember this is if you draw the numeral one you can easily turn that into a p and that stands for positive so this is a false positive error and similarly

Type two error you can think of this as a type or as i had negative error false negative error so you did not find a difference in your study but actually there is a difference and if you look at the roman numerals too you can easily connect those to make an n and you can think of that as a false negative error

Transcribed from video

CREOG Statistics By EDU Chief GW