Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
127 Cards in this Set
- Front
- Back
Define Epidemiology
|
Study of
1) disease patterns 2) determinants of disease patterns in populations |
|
Epidemic
|
Study of disease in excess of expected rate: 1 yr
|
|
Endemic
|
Study of disease in excess of expected rate: >1 yr
***in LOCALIZED AREA |
|
Pandemic
|
Study of disease in excess of expected rate: 1 yr
***in at least 3 countries |
|
Rate=
|
Actual cases/ Potential cases
|
|
Population changes
|
-depend on census count
-change from year to year (not month to month) |
|
Prevalence=
|
All Diseases/ Population
*for any given time and location |
|
Incidence=
|
New Diseases/ Population
*for any given time and location |
|
Attack Rate=
|
Cases of disease/ Total number of people exposed to supposed risk factor
|
|
Screening Tests should be highly...
|
Sensitive
|
|
Sensitivity is used to rule in/out healthy/diseased
|
Sensitivity used to rule IN diseased
|
|
Specificity is used to rule in/out healthy/diseased
|
Specificity is used to rule IN healthy
|
|
Sensitivity=
|
+ test results with dz/ total dz'd
|
|
Specificity=
|
- test results with NON-dz/ total NON-dz'd
|
|
PPV=
|
+ test results with dz/ total + results
|
|
NPV=
|
- test results with NON-dz/ total - results
|
|
When to think about answering via PPV, NPV (3)
|
1) ask for PPV, NPV
2) they give test results -if pos, do PPV -if neg, do NPV 3) they ask: how much more likely that I have/ dont have disease with a pos/neg test |
|
When you compare same test bw 2 different populations, PPV/NPV will change based upon....
|
PREVALENCE!
|
|
When they give SEN/SPEC and ask whats PPV/NPV, what do you do?
|
1) make table
2) do total population as 1,000 (bottom corner outside table) 3) do dz'd population OUT OF 100 (3rd column below dz's) |
|
When they asked, "what study used", what's the first thing you look at?
|
1) did they use a control group
2) are they insinuating comparison |
|
When they asked, "what study used" and they
1) did NOT use control group or are NOT insinuation comparison, ....now what do you look at? |
are they talking about 1 or more pts
|
|
When they asked, "what study used" and
1) did NOT use control group or are NOT insinuation comparison, 2) only talking about 1 pt ...now what? |
you know they used CASE REPORT
|
|
When they asked, "what study used" and
1) did NOT use control group or are NOT insinuation comparison, 2) only talking about MORE than 1 pt ...now what? |
you know they used CASE-SERIES
|
|
When they asked, "what study used" and they
1) DID use control group or ARE insinuating comparison, ....now what do you look at? |
is it
1) retrospective 2) one point in time 3) prospective |
|
When they asked, "what study used" and they
1) DID use control group or ARE insinuating comparison 2) timing is retrospective ... now what? |
you know they used CASE-CONTROL
|
|
When they asked, "what study used" and they
1) DID use control group or ARE insinuating comparison 2) timing is "one point in time" ... now what? |
you know they used CROSS-SECTIONAL
|
|
When they asked, "what study used" and they
1) DID use control group or ARE insinuating comparison 2) timing is prospective ... now what? |
you know they used COHORT
|
|
Case report, Case series, Cross sectional, Case control, Cohort
are examples of what kinds of studies? |
Observational studies
|
|
There are 2 general categories of studies
|
1) Observational studies (non-expiremental studies)
2) Expiremental studies |
|
What are the Observational studies?
|
Case report, Case series, Cross sectional, Case control, Cohort
|
|
What are the Expiremental studies?
|
Clinical trials, Double-blind, Community trials, Crossover trials
|
|
Define Case report
|
single patient no control group
|
|
Define Case series
|
Several patients, no control group
|
|
Define cross sectional
|
one point in time
|
|
Define Case control
|
retrospective
|
|
Define Cohort
|
prospective
|
|
Cross sectional
-clues -formula to use |
clues: incidence, prevelance
formula: Chi-square |
|
Case control
-clues -formula to use |
clues: many RF, 1 dz
formula: Odds ratio |
|
Cohort
-clues -formula to use |
clues: 1 RF, many dz's
formula: RR, AR |
|
How many steps to approving a drug are there?
List them? |
6 steps (2 pre-phases, 4 phases)
1) INDA: app to FDA to exp on animals 2) NDA: app to FDA to exp on humans 3) Phase1: random people, pharmacokinetics 4) Phase2: people who have disease only 5) Phase3: double-blind 6) Phase4: marketing study |
|
In phase 4 (marketing study), whats being evaluated?
|
S/E profile (if it does not match what you gave FDA, pull it)
***this is once its already being sold in stores |
|
2 steps to solving odds ratio
1) make table- what should it look like? 2) use formula- what is it? |
table: make A your disease with RF,
then just cris-cross to find everything else formula: AD/BC |
|
How is Odds Ration different from RR and AR?
|
Odds ratio compares NUMBERS
RR, AR compare INCIDENCE (thats why odds ratio can only calculate retrospective studes, and RR, AR can calculate prospective studies) |
|
Odds ratio
-verbalized -formula |
getting dz with exposure/ getting disease w/o exposure
(NUMBER +dz with +RF/ NUMBER +dz with -RF) x (NUMBER -dz with -RF/ NUMBER -dz with +RF) (AD/BC) |
|
Risk Reduction (RR)
-verbalized -formula |
incidence rate of exposed group/ incidence rate of non-exposed group
(+dz with +RF/ all +RF) / (+dz with -RF/ all -RF) |
|
Relative Risk Reduction (RRR)
-formula |
RRR= 1-RR
|
|
How do you know when they want you to use RRR or just RR?
|
If RF inc chances of dz: RR
If RF dec chances of dz: RRR |
|
Attributable Risk (AR)
-verbalized -formula |
incidence rate of exposed group- incidence rate of NONexposed group
(+dz with +RF/ all +RF) - (+dz with -RF/ all -RF) |
|
Forumla for Number Needed to Harm (NNH) |
1/AR |
|
Absolute Risk Reduction (ARR) -verbalized -formula |
??? |
|
Number needed to treat formula (NNT)
|
1/ ARR
|
|
Precision is synonymous with
|
-Reliable
-Consistent |
|
Accuracy is synonymous with
|
Validity
|
|
Definition of valid/ invalid
|
valid= nl lab value
invalid= abnl lab value |
|
Other words for selection bias
|
-Sampling bias
-Berkson's bias -Nonrespondent bias |
|
How do we solve selection bias
|
Random, Independent samples
|
|
Whats the word for subjects who want to impress those testing them?
What kind of bias is this? |
Hawthorne effect
-measurement bias -admission rate bias |
|
How do we solve Hawthorne effect (measurement bias)?
|
Placebo group (control group)
|
|
Whats another word for interviewer bias?
|
Pygmalion effect
|
|
How do we solve Interviewer bias (Pygmalion effect)?
|
double-blind-study
|
|
Whats Lead-time bias?
|
early detection confused with increased survival
(increased survival is being attributed to medicine being tested) |
|
Which type of bias proves why screening is important?
|
Lead-time bias
|
|
Whats recall bias?
|
subjects cant remember accurately
|
|
How do we minimize recall bias?
|
confirm with other sources
|
|
Whats late-look bias?
|
Late detection confused with decreased survival
(decreased survival being attributed to medicine being tested; ie medicine is not effective) |
|
How do we minimize late-look bias?
|
from the moment you include subjects into study, categorize them by severity
|
|
Whats confounding bias?
|
Hidden UNNATURAL factors that do BOTH of the following
1) influence your exposure to the factor in question (smoking makes you drink more) 2) smoking also causes you to have sq cell cancer of the esophagus (like alcohol, which is the factor in question) |
|
How do we minimize confounding bias?
|
-multiple studies
-good research design |
|
Whats latent period?
|
disease being studied is chronic (pts health status fluctuates in chronicity)
|
|
Whats effect modification?
|
NATURAL factors that
1) influences your predisposition to getting the disease that is being attributed to the studied factor (ie in a study to see if OCP's increased your risk for breast cancer, an effect modification would be having women in the study with family history of breast cancer) |
|
Does primary prevention increase/decrease incidence or prevelance?
How? |
DECREASES both incidence and prevalence
its an effort to decrease dz incidence, thus, prevalence |
|
Does secondary prevention decrease/increase the incidence or prevalence?
How? |
decreases prevalence only
secondary prevention= cure |
|
Does tertiary prevention decrease/increase prevalence/incidence?
How? |
INCREASES prevalance
slows progression of disease (ie chemo) |
|
List the 4 types of scales
|
1) Nominal
2) Ordinal 3) Interval 4) Ratio |
|
Nominal scale
-compares: |
compare groups (this or that)
|
|
Ordinal scale
-compares: |
compares: rank, quality order
|
|
Interval scale
-compares: |
compares: NUMBERS (quantity, means, standard deviation)
|
|
Ratio scale
compares: |
compares: ?
|
|
name the 7 types of statistical tests
|
1) Meta-analysis
2) Correlation analysis 3) t-tests 4) Matched pairs t-test 5) Analysis of Variance (ANOVA) 6) ANOVA Repeated measurements 7) Chi-square |
|
How do we measure correlation?
|
+1--> direct correlation
-1--> indirect correlation (this is the one we see graphed as the slope of a line) |
|
In correlation analysis, whats considered a strong/weak correlation/ no correlation?
|
strong: +1 OR -1
weak: closer to 0 none: 0 |
|
Chai-square measures
|
nominal data ONLY (any number of groups: this or that or that or that)
|
|
Which statistical tests are made to compare means?
|
t-test
Matched-paired t-test One-way ANOVA Repeated measure ANOVA |
|
If you want to compare means of 2 sample groups, use which statistical test?
|
t-test
|
|
If you want to compare means of 2 sample groups that are linked (before and after), use which statistical test?
|
Matched pairs t-test
|
|
If you want to compare means of >2 sample groups, use which statistical test?
|
One-way ANOVA
|
|
If you want to compare means of >2 sample groups that are linked (before and after), use which statistical test?
|
Repeated measures ANOVA
|
|
Define null hypothesis
|
opposite of thesis
(thesis= what you're trying to prove) *** "statement of no relationship" |
|
In order for our study to be stastically significant, we must...
Our ability to do this is determined by... |
-reject the null
-determined by how the p value of our study compares to the SET P VALUE |
|
What number is the SET P VALUE?
|
It can be anything, but we want it to be
0.05 (5%) |
|
What if the SET P VALUE does not equal 0.05?
|
Its ability to reject or accept the null is flawed.
|
|
If the SET P VALUE is < 0.05, this is known as...
How does this effect our ability to reject/accept null? |
Type I (alpha) error
-we're more likely to reject the null, erroneously (to prove our theory, goal is to reject the null) |
|
If the SET P VALUE is > 0.05, this is known as...
How does this effect our ability to reject/accept null? |
Type II (beta) error
-we're more likely to accept the null, erroneously (to prove our theory, goal is to reject the null) |
|
What is power?
-verbalized -formula |
power= sample size
power= 1- beta |
|
increasing sample size results in decreased....
|
type II (beta) error
|
|
In a Positive skew, tail is on L/R?
|
R
|
|
In a Negative skew, tail is on L/R?
|
L
|
|
In ANY skewed chart, which (mean, mode, median) is
closer to peak in between closer to tail |
In BOTH Pos. & Neg. skew
mode- closer to peak median- in middle mean- closer to tail |
|
define mode
|
most frequently occurring
"es la moda; its popular" |
|
define median
|
central point
|
|
define mean
|
average
|
|
If you add or remove one data which will change, mean/mode/median?
|
mean: changes
median: changes mode: NO CHANGE |
|
draw distribution curve
|
-1 to +1: uses up 68.2% of area under curve
-2 to +2: uses up 95.5% of area under curve |
|
1) if RR > 1 that means
2) if RR < 1 that means |
1) exposure to risk factor INCREASES your chances of getting dz
2) exposure to risk factor DECREASES your chances of getting dz ex: RR of 0.71 means that exposure to RF dec your chance of getting dz by 29% (1- 0.71) |
|
How do we know whether or not we can believe what the number for RR or Odds ratio implies?
|
If the confidence interval (acquired from the same set of data as RR)
-contains "1": RR means nothing -does NOT contain "1": RR holds true |
|
Formula for confidence interval includes x,z, s, n, which stand for
|
x= mean
s= standard dev n= population # z= if they want a a 95% confidence interval--> 2 if they want a 99% confidence interval --> 2.5 |
|
Birth rate formula
|
(live births / population) x 1,000
|
|
Fertility rate formula
|
(live births/ women of childbearing age) x 1,000
|
|
Maternal mortality rate formula
|
(Maternal death/ LIVE births) x 100,000
|
|
Infant mortality rate formula
|
(Infant death/ live births) x 1,000
*Infant= 0 to 365 days |
|
Neonatal mortality rate formula
|
(Neonatal death/ live births) x 1,000
*neonate= 0 to 28 days |
|
"lost to follow up" gives you what kind of bias?
|
selection bias
|
|
If they give you RR, and they want AR, what formula do you use?
|
ARP= (RR-1)/RR
|
|
Another word for relative risk is
|
Hazard ratio
|
|
a confounder has to meet 2 qualities
|
1) increase exposure to RF in question (smoking makes you drink more)
2) increase incidence of result being evaluated (smoking causes oral cancer too- like alcohol, the RF in question) |
|
What is another word for valid?
|
accurate
|
|
what are the 2 types of validity?
|
internal and external
|
|
what is internal validity?
|
applicability of the obtained results WITHIN the studied population (middle-aged women)
|
|
what is external validity?
|
applicability of the obtained results BEYOND the studied population (middle-aged women)
|
|
What is succeptibility bias?
|
where a treatment regimen in selected for a patient based on the severity of their condition, without taking into account possible confounding variables.
|
|
Name 3 ways to control confounding
|
1) matching
2) randomization 3) stratified analysis |
|
What is matching?
|
Matching: select controls for each subject that match at a potential confounder (age, race, socioeconomic status)
|
|
Of the three ways to control for confounders, which allows you to control for known and unknown confounders?
|
randomization
|
|
What is the relationship of mean, median and mode in a normal distribution curve?
|
they're all equal!
|
|
what is the median of this set?
18, 20, 21, 22, 22 |
21
|
|
What is the median of this set?
18, 20, 21, 22 |
20.5
(20+21)/2 |
|
In a dz with low prevalence, odds ration= ____ ?
|
RR
|