Behavioral Science Flashcards
Case Control Study
- Retrospective observational study
- Looks at patients who have been diagnosed with a disease and then looks back at a risk factor that may have cause it.
- Most likely is an odds ratio (AD/BC) ratio diseased with/without risk factor over ratio not diseased with/not diseased without
- Patients who have COPD are selected and then asked about their smoking history
Cohort
- Finds two groups of people with certain characteristics or features and compares their liklihood of getting a specific disease
- Risk Ratio (A/A+B)/(C/C+D)
- Takes patients who smoke and look back to see if they have an increased risk of having COPD
Cross Sectional
- Looks at a population at A SINGLE POINT IN TIME to get disease prevalence
- Most associated with prevalence
- Can’t establish causality
Stages of Clinical Trial
1-small group of health volunteers
2 - small group of diseased
3 - large group of diseased
4 - post market surveilance
Sensitivity
TP/TP+FN
-Used to rule out a disease (picks up the most true positives)
Specificity
TN/TN+FP
-Used to rule in a disease (Picks out all the people who tested positive who actually have the disease)
PPV
TP/TP+FP
-The liklihood that given a positive test, the patient will actually have the disease
NPV
TN/TN+FN
-The liklihood that given a negative test result the patient will really not have the disease
Incidence
- The number of patients diagnosed with a disease over a specific time / The susceptible population
- Patients who are currently diagnosed or who are dead do not count as the susceptible population
Prevalene
-The proportion of people who currently have the disease over the total population
Odds ratio
- Descibes the ratio of disesed exposed to the risk factor over the non diseased exposed to the risk factor
- (a/c)/(b/d) = AD/BC
Relative Risk
- Describes the liklihood the role that a risk factor may play in the disease
- (a/a+b)/(c/c+d)
Attributable Risk
(a/a+b)-(c/c+d)
- When it is used to show how a treatment improves survival it is called number needed to treat
- When it is used to describe how a risk factor might increase the liklhood of disease is it number needed to harm
Selection Bias
-Not selecting right patients, berkson is a part
Berkson’s Bias
-Some patients are lost to follow up and thus interfere with interpretation
Lead Time
-Earlier Diagnosis does not mean improved survival
Hawthorne
-Patients will change their behavior when they know they are being studied.
Standard Error of Mean
SD/(sqrt(n))
-Used to calculate confidence intervals
Positive Skew
Mean is greater than the median.
-The tail will be in the positive direction
Negative Skew
Mean is less than the median
-Tail will be in negative direction
Standard Deviation and SEM
- Standard deviation is not the same as SEM
- SEM is used to report confidence intervals and is standard deviation over the sqrt(n)
Type I Error
- Alpha or false positive rates
- The liklihood that the study reject the null hypothesis, or comes to a positive conclusion when in fact one does not exist.
- Measured by alpha which is often a p value (SEM) or CI which is SEM
Type 2 error
- False negative rate. If the study is too small, it is possible that a true conclusion will be missed
- Measured with beta
- Power is 1-beta. Higher the number, the higher the power, the lower the beta.
T Test
-Used to compare two quantatitve variables
ANOVA
-Used to compare more than 2 quantatative variables
Chi Squared
-Used to compare 2 categorical variables