Stats 3

Question 1

List the following in terms of the level of evidence they provide: cross sectional, cohort (longitudinal), case control, descriptive case reports and case series, ecological, intervention or clinical trials.

Answer 1

1) descriptive
2) ecological
3) cross sectional
4) case control
5) cohort
6) intervention or clinical trials

Question 2

What do ecological studies do?

Answer 2

Look at the association between exposure and disease on a population or area level rather than on an individual level

Eg: association between smoking and cardiovascular disease: look at average cigs sold pp in USA state and rate of cardiovascular mortality

Question 3

What is ecological fallacy?

Answer 3

With ecological studies any associations seen are on a population level and may not translate to individual level

Question 4

How do you analyse ecological data?

Answer 4

Usually analysed on a group level as a scatter plot

Question 5

Main disadvantage of ecological studies?

Answer 5

Data often not available on confounders

Question 6

What is a cross sectional study?

Answer 6

Collects observations on individuals at one point in time- providing snapshot of the health of a population

Question 7

Do you know the disease prevalence from a cross sectional study?

Answer 7

No- it is not included

Question 8

Can confounding occur in a cross sectional study?

Answer 8

Yes but as long as data on potential confounders are collected they can be dealt with at the analysis stage (stratification or multiple regression)

Question 9

How do you analyse cross sectional studies?

Answer 9

Associations initially assessed by computing appropriate measure of effect (odds ratio/mean dif etc)
95% CI and statistical significance

Question 10

Are cross sectional studies useful for studying rare exposures or rare outcomes?

Answer 10

No

Question 11

What are cross sectional studies useful for?

Answer 11

How much disease there is in a population and to look at cross sectional associations between exposure and disease

Question 12

In what type of study are the people chosen based on presence or absence of disease?

Answer 12

Case control

Question 13

Are case control studies useful for rare diseas s?

Answer 13

Yes - reduces how many normal people are required and is useful for looking at risk factors for rare diseases
But cannot look at how much disease there is!

Question 14

Prevalent or incident cases can be used for case control studies- explain!

Answer 14

Prevalent = people who had the disease at a particular point or period in time 
Incident= new cases that arise within a fixed period of time!

Question 15

How analyse case control studies?

Answer 15

Cross tabulate the outcome against exposure to find % exposed among cases and among controls
Test statistical significance using chi-squared test as variables are categorical

Question 16

What is the only measure of effect suitable for case control studies?

Answer 16

Odds ratio & 95% CI

Question 17

What do you use to calculate adjusted odds ratios (ratio which has had the effect of confounders removed)

Answer 17

A multivariate model called multiple logistic regression

Question 18

What are matched case control studies?

Answer 18

Controls selected so they are matched to a case based on specific variables believed to be confounding

Question 19

Main advantage of case control studies

Answer 19

No need for follow up
Reduces number of disease free people needed
It is possible to study multiple exposures

Question 20

Disadvantages of case control studies

Answer 20

Only possible to look at one outcome

Bias is a problem- specifically selection bias

Question 21

What does a cohort study measure?

Answer 21

Measures incidence of disease (gold standard for epidemiologists)

Question 22

What would an ideal cohort study look like?

Answer 22

Over time- start with population of that population look at those without disease then over time look at those exposed and not exposed and which develop disease and which don’t.

Question 23

Difference between incidence and prevalence?

Answer 23

Incidence is the rate of new (or newly diagnosed) cases of disease.
Prevalence= the number of cases alive with the disease during a period.

Question 24

What is a longitudinal study?

Answer 24

Series of cross-sectional surveys recruiting a study population to participate over time

Question 25

Name 5 types of cohort study's?

Answer 25

1) Longitudinal 2) retrospective (exposure and outcome have already occurred) 3) perspective studies (exposures may not have occurred yet but the disease outcome has NOT) 4) occupational cohorts (for rare exposures) 5) cohorts with no specific exposure in mind- eg black womens health study

Question 26

Benefits of cohort studies?

Answer 26

Selection bias NOT a problem as selection of people occurs before disease

Question 27

Problems with cohort studies

Answer 27

Loss of follow up Expensive Can be inefficient for rare diseases

Question 28

How do you calculate incidence risk?

Answer 28

Number of new cases of disease during a time period/ total population at risk (need cohort of people followed over time)

Question 29

What is an open cohort study?

Answer 29

People recruited into cohorts at different times

Question 30

How do you calculate incidence rate? What type of study is it appropriate for?

Answer 30

Number of new cases of disease/ total persons time at risk

Question 31

What is rate ratio?

Answer 31

Used for cohort studies which look at incidence rate Rate ratio= rate of disease amongst exposed/ rate of disease amongst unexposed Tells you how much more likely the exposed group are to develop the disease than the exposed group

Question 32

What does a rate ratio of 0.13 for veg eaters vs non veg eaters for cancer mean?

Answer 32

High veg eaters 87% less likely to develop colorectal cancer

Question 33

What are randomised controlled trials/ clinical trials/ intervention studies?

Answer 33

Human experiments where the investigator implements the intervention of interest and determines which are exposed and unexposed

Question 34

How do you reduce chance, confounding and bias in a RCT?

Answer 34

Chance minimised by estimating study power Confounding removed my randomisation Bias removed by blinding

Question 35

What is the difference between explanatory trials and pragmatic trials?

Answer 35

Explanatory- measure the efficiency of an intervention under optimal conditions. Pragmatic trials- measure the effectiveness of an interventions implemented in practice

Question 36

What is a type I error?

Answer 36

When you reject the null hypothesis when it is true

Question 37

What is allocation concealment?

Answer 37

Treatment allocated after the subject has been recruited and after baseline measurements taken

Question 38

What are the two approaches for calculating sample size?

Answer 38

1) compute how many subjects are needed to estimate the effect of interest with a specified degree of precision 2) compute the number of subjects needed to test the hypothesis of interest with certain power.

Question 39

What information do you need to obtain a prevalence estimate and 95% CI?

Answer 39

1) rough guess of prevalence (or SD if continuous) | 2) idea of how precise or narrow you want the CI

Question 40

What do you need if you want a narrower CI?

Answer 40

More people

Question 41

If you plan to carry out a statistical test to compare two groups (power approach) what info do u need?

Answer 41

1) rough guess of prevalence- % without come in baseline group. Or SD and mean is continuous outcome 2) minimum difference you want to be able to detect 3) set the probability of type 1 error (same level as p-value, 0.5 or 5%) 4) set the prob of type II error

Question 42

What relation does the prob of type II error have on the power of the study?

Answer 42

1- prob of type 2 =power of study Power of study is usually 80-90%

Question 43

What is the power of a study?

Answer 43

Probability of detecting an effect as significant if it really exists (usually 80-90%)

Question 44

Name two computer packages used to calculate sample size

Answer 44

SamplePower | NQuery

Question 45

What is a cause?

Answer 45

An exposure that produces the diesease

Question 46

What are risk factors?

Answer 46

Exposure that increases risk of disease

Question 47

Risk factors are not necessarily cause because of what?

Answer 47

1) Chance 2) Indirect causation (risk factors associated with disease but not cause) 3) reverse causation (disease maybe causing Rf) 4) confounding

Question 48

How do you calculate is there is a strong association in a study?

Answer 48

Odds ratio or risk ratios of >2 *not true for every case*

Question 49

List Bradford-Hill 9 criteria to assess studies

Answer 49

1) strength of association 2) consistency and replication 3) is the effect specific 4) temporality (cause must precede onset of disease!) 5) biological gradient 6) plausibility 7) coherence (does association fit with what's already known) 8) experimental evidence (does intervention in exposure influence disease outcome?) 9) Analogy- is there a similar cause- effect relationship we already know of?