Epi

Front 1

descriptive epidemiology is useful for:

Back 1

1) identifying new health problems
2) monitoring how disease patterns are shifting geographically, demographically, and over time. This is important for allocation of resources and planning educational or preventive programs.
3) disseminating information about new or unusual health problems
4) formulating hypotheses about the determinants of disease

**Best use is to generate hypotheses**

Front 2

Case Report

Back 2

a descriptive study.

A case report is a detailed description of disease occurrence in a single person. Unusual features of the case may suggest a new hypothesis about the causes or mechanisms of disease

Front 3

Case Series

Back 3

A case series is a report on the characteristics of a group of subjects who all have a particular disease or condition. Common features among the group may suggest hypotheses about disease causation. Note that the “series” may be small (as in the example below) or it may be large (hundreds or thousands of “cases”).

Front 4

Important limitation of Case Report and Case Series

Back 4

no comparison group and often a small population size.

Front 5

Cross Sectional Survey:

limitations?

Back 5

Cross sectional surveys assess the presence of disease and risk factors at a given point in time and provide a “snapshot” of diseases and risk factors simultaneously in a defined population.

Limitation is that the temporal relationship between the risk factor and disease is frequently unclear. Under these circumstances, they can generate hypotheses, but these associations need to be tested by appropriate analytical studies.

**Note it could be clear for example if the risk factor is gender, we know that that came first!**

Front 6

Correlational Studies

Back 6

These are distinguished by the fact that the unit of observation is not a person; rather it is an entire population or group. In essence, these studies attempt to correlate the average exposure in various populations with the overall frequency of disease within the populations.

Front 7

Advantages of Correlational Studies

Back 7

- data is frequently available so it is quick and inexpensive

- correlation coefficient (r value) can provide a measure of association.

r = 1 (perfectly positive)
r = -1 (perfectly negative)
r = 0 (no association)

varying degrees of strength of association!

Front 8

Limitations of Correlational Studies

Back 8

1. You can’t really link the risk factor to the disease, i.e., it is not clear that the women who ate the most meat were the ones who got colon cancer.

2. Since the exposure levels represent average exposure in a large number of people, correlational studies can mask more complicated relationships. (like J-shaped relationship)

Front 9

cohort studies

Back 9

identify a group of subjects who are free of disease of interest. group them according to their exposure to the risk factor, and then compare the two exposure groups with respect to incidence of disease.

KEY
- start with subjects free of disease
- group subjects based on exposure to risk factor
- measure and compare frequency of disease among the exposure groups

Front 10

case control study

Back 10

instead of starting with non-diseased subjects and comparing incidence of disease, this study identifies diseased people and compares their past exposures to those of non-diseased people.

KEY
- find diseased subjects and non-diseased comparison group
- ascertain exposures prior to disease
- compare odds of having had specific exposures

Front 11

benefit of cohort study

problem of cohort study

Back 11

PRO
- useful for unusual exposures (not outcome)
- easy to know temporal relationship between exposure and outcome
- can calculate incidence, RR...
- also good for evaluating multiple outcomes on a single exposure

CON
- sometimes exposure status is unclear (retrospective)
- need to follow large numbers for a long time
- expensive and time consuming
- not good for rare diseases
- not good for diseases with long latency
- possibility for loss to follow up bias

Front 12

RCT

Back 12

like a prospective cohort but the investigators assign the exposure!

Front 13

fixed vs dynamic population

Back 13

fixed: membership is (relatively) permanent and defined by an event

dynamic: membership is transient

Front 14

ratio vs proportion vs rate

Back 14

ratio: just dividing one number by another. does not necessarily imply a relationship b/w the 2 numbers

proportion: ratio that relates a part to a whole (usually a precentage)

rate: a ratio where the denominator also includes time

Front 15

prevalence

Back 15

the proportion of the population that has the disease at a particular time!

- it is a way of assessing the overall burden of the disease on the population

- can be a point in time or a stated time period! [Point vs Period prevalence]

**Note that this is NOTE a rate**

Front 16

incidence

Back 16

a measure of the occurance of new cases of a disease during a span of time

**Note this IS a rate**

think of this as the probability of developing the disease over a period of time

Front 17

cumulative incidence

Back 17

evaluate the presence of disease at the beginning and the end of a block of time and then divide this number of new cases by the number of people "at risk".

This really is a proportion!!

- # New cases during the period of time / total population that was at risk during that time

- ALWAYS EXPRESS THE TIME PERIOD

Front 18

incidence rate

Back 18

- a true rate measuring incidence with time involved.

- the numerator is the same for both CI and IR, but the denominators differ:
CI: total # at risk
IR: total amount of time of being at risk

- usually expressed as "person-time"

- requires repeated follow-up observations and can be expensive and time-consuming

Front 19

calculation for prevalence

Back 19

P = incidence rate * avg duration of disease

P = IR * D

Front 20

how to calculate average duration of disease

Back 20

D = P / IR

- if average duration of disease is constant, than reducing incidence would reduce prevalence

- if incidence remains constant, reducing duration of disease would reduce prevalence

Front 21

morbidity rate

Back 21

incidence of non-fatal cases of a disease in a specific period of time

- really a PROPORTION!

Front 22

mortality rate

Back 22

incidence of fatal cases of a disease in a specific time period

- really a PROPORTION

Front 23

case-fatality rate

Back 23

number of deaths from a specific disease divided by the total number of cases of that disease

- gives a measure of the severity fo the disease

Front 24

attack rate

Back 24

CI for a disease during a specific period (i.e. an epidemic)

Front 25

live birth rate

Back 25

frequency of live births / 1000 females of childbearing age

Front 26

infant mortality rate

Back 26

frequency of deaths in children under 1 yr of age / 1000 live births

Front 27

how to measure strength of association

Back 27

Relative risk (Risk ratio)

Front 28

relative risk

Back 28

measures the strength of association between a risk factor and the outcome.

indicates the multiple of times of risk a certain exposure has to the disease.

if RR = 1, then there is no difference in risk.

RR = Ie / Io = CI[e] / CI[o]

Front 29

risk difference (AKA attributable risk)

Back 29

RD = Ie - Io

focuses on the excess risk of disease in those who have the factor compared with those who dont!

provides a measure of the public health impact of the risk factor, and focuses on the number of cases that could be potentially prevented by eliminating the risk factor

- means "x amount of excess cases / person-time due to the risk factor"

Front 30

attributable risk%

Back 30

AR% = AR/Ie * 100

calculates the proportion of disease in the exposed group that can be attributed to the exposure!

Front 31

odds ratio

Back 31

- if 2x2 is: a b c d. then OR is ad/bc

- as long as number of outcomes is very low relative to the number of controls, the OR is a good estimate of RR.

- so as the outcome of interest becomes more common, the OR tends to OVERESTIMATE the magnitude of association

Front 32

Confidence Interval

Back 32

- the interval in which the true value lies.

- if the interval includes 1, than the results are insignificant for that power

Front 33

what are you testing in cohort? case- control?

Back 33

cohort: testing the frequnecy of a specific outcome

CC: testing the frequnecy of having the exposure of interest

Front 34

how to state the null hypothesis

Back 34

- incidence is the same in both groups
- RR = 1
- RD = 0

Front 35

p- value

Back 35

The probability of seeing a difference this big or bigger, if the groups were not different, i.e. just by chance.

if p <0.05, then we are 95% certain that it was not due to chance

Front 36

what statistical tests to use?

Back 36

- t-test for measurement data
- chi-square for categorical data
**Note: chi-square can exagerate results in small sample size

Front 37

equipoise

Back 37

a balance in which one has sufficient doubt that the tx is beneficial balanced against sufficient belief that it is beneficial

Front 38

goals of randomization

Back 38

1. unbiased assignment to tx groups
2. achieving baseline comparability of the groups with respect to all other factors that might influence outcome (control of confounding)

Front 39

effects of non-compliance in RCTs

Back 39

1. minimizes any differences between groups
2. statistical power to detect true difference is reduced and the true effect will be biased towards the null

Front 40

how to maintain high compliance

Back 40

1. simple protocol
2. enroll motivated and knowledgeable subjects
3. weed out non-compliant
4. maintain contact frequently

Front 41

how to measure compliance

Back 41

- ask patients
- collect pill packs
- measure biological markers

Front 42

intent-to-treat analysis

Back 42

- all subjects should be included for the primary analysis even if they did not complete or receive the appropriate tx.

why?
- preserves baseline comparability and control of confounding
- prevents bias [since there may be a difference between the two groups in compliance]
- reflects efficacy

Front 43

how to select an exposed group in cohort

how to select unexposed group in cohort

Back 43

exposed:
-can use the general population or a subset (i.e. doctors).
-or can use a special exposure cohort for an unusual risk factor

unexposed:
-needs to be as similar as possible to the exposed group to prevent confounding
-information collection needs to be as accurate and comparable as possible.

Front 44

internal control group

Back 44

pretty much the controls are identical to the exposure group minus the exposure (i.e. the nurses health study)

Front 45

why is the general population a bad control for cohort?

Back 45

- some of the general population may have had the same exposure

- also the general population may not be as healthy as employed workers (healthy worker effect)

Front 46

healthy worker effect

Back 46

employed workers are generally more healthy than the general population so the general population doesnt serve as a good control for cohort studys.

Front 47

missclassification bias

Back 47

occurs when subjects are incorrectly categorized with respect to their risk factor status or outcome

Front 48

random misclassification

Back 48

occurs when errors in classiying risk factors status or the outcome status occur with equal frequency in the groups being compared.

- strength of association can only be UNDERestimated (bias towards the null)

Front 49

nonrandom misclassification

Back 49

occurs if information is more accurate in one of the groups.

- can over or underestimate the association

Front 50

bias from loss to follow up

Back 50

occurs in large cohort studies in which loss to follow up occurs more often in one of the groups

- can over or underestimate the association

Front 51

selection bias

Back 51

results when procedures used to select subjects into the study lead to a result that is different from what you would have gotten if you had enrolled the entire target population

- in Case-control: need the "would" criterion (if the control has the disease, would he have been enrolled as a case?)

- in retrospective cohort: bias can occur if the selection of exposed/unexposed subjects was somehow related to the outcome

- in prospective cohort there is possibility for loss to follow up bias

Front 52

how to avoid selection bias

Back 52

- use reliable motivated group
- use subjects that are easy to track
- get contact info from family and friends
- maintain participant interest
- would criterion

Front 53

confound by indication

Back 53

- situation where patients with more severe diseases are more likely to be treated with certain meds.

it may appear that the drug actually induces the poor outcome.

- also used when a given disease leads to other health complications for which certain drugs may be prescribed. this can create a false association!

Front 54

3 types of observation bias

Back 54

- recall
- interviewer
- misclassification

Front 55

recall bias

Back 55

when there are systematic differences in the way that subjects remember or report exposures/outcomes

Front 56

how to avoid recall bias

Back 56

- use controls who are also sick
- use standardized questionnaires
- examine pre-existing data

Front 57

interviewer bias

Back 57

when there are systematic differences in soliciting, recording, or interpreting info on exposure or outcome

Front 58

how to avoid interviewer bias

Back 58

blind them!
use standard questionnaires

Front 59

misclassification bias

Back 59

occurs when subjects are incorrectly categorized with respect to either exposure or outcome

random or non-random