PAS - Observational Epidemiology Studies

Question 1

outcome - define?

Answer 1

the disease or occurrence of event being studied - any measurable change in health status

Question 2

exposure factors - define?

Answer 2

any factor that may affect the risk of developing a disease or health outcome

e.g. environmental, lifestyle, biological agents, specific behaviours

Question 3

risk factor - define?

Answer 3

characteristic or exposure that increases the likelihood of developing a disease or health outcome

e.g. smoking, diet, age, genetic pre-disposition = modifiable & non-modifiable

Question 4

types of observational studies?

Answer 4

cohort study
case-control study
cross-sectional study

Question 5

cohort study - study design?

Answer 5

participants who do not have the outcome at baseline are followed over time to see ho develops the outcome of interest

helps estimate the incidence of the outcome

Question 6

prospective cohort study?

Answer 6

follows a cohort of individuals over time, starting before they develop the outcome of interest

risk factor of interest is defined beforehand, and participants are classified as ‘exposed’ or
‘unexposed’ regarding the chosen risk factor

other risk factors/ confounding variables collected at base-line and during follow-up

at the end of the follow-up period, interest assess the manifestation of the outcome of interest in regards to their exposure status

Question 7

retrospective cohort study?

Answer 7

existing data from historical records used to follow a cohort from the past to the present

categorises participants by past exposure to a pre-defined risk factor and assessing current outcomes/ outcomes that have already occurred

Question 8

confounder - define? effect of a confounder on study data?

Answer 8

a variable which is associated with the exposure of interest, and independently associated with the risk of developing the outcome

can lead to an under/ over-estimation of a real association between exposure & outcome

Question 9

how can we deal with confounders in cohort studies?

Answer 9

prospective cohort studies allow pre-specified confounders of interest to be studied

difficult with historical/ retrospective cohort studies as record may lack the information on the confounder of interest

adjust statistical models, or analyse results stratified by the confounder

Question 10

how can risk ration be used as measures of association from cohort studies?

Answer 10

risk ratio = risk of disease occurring in an exposed group / unexposed group

RR = 1 - no difference between groups
RR > 1 - increased risk in exposed group

confidence intervals can be applied - i.e. a range within which the true population relative risk is expected to lie with a 95% degree of certainty with a 95% CL

Question 11

difference between relative and absolute risk?

Answer 11

relative risk = useful for understanding the strength of the relationship between exposure and outcome

absolute risk = clearer picture of the actual chance of a risk happening and its reduction, more meaningful for healthcare decisions

i.e.
absolute risk = AR(control) - AR(treatment) = X%
relative risk = AR(treatment)/ AR(control) = X (ratio below 1 means lower risk in treatment group)

Question 12

advantages of cohort studies?

Answer 12

exposure measured before disease - reduces potential bias

multiple outcomes/ diseases can be studied for any one exposure

incidence of disease can be measured in exposed vs unexposed groups

Question 13

disadvantages of cohort studies?

Answer 13

slow, expensive, complex

need large numbers - harder with rarer diseases/ rarer outcomes

collection of data may alter behaviour

losses to follow-up = can introduce bias

Question 14

case-control study - define?

Answer 14

compares individuals with the outcome of interest (cases) to those without the outcome (controls)

retrospectively assesses and compares their exposures to potential risk factors - starts with the outcome and looks back in time at exposures

Question 15

how to deal with confounders in case-control studies?

Answer 15

collect information on potential confounders and adjust for them at the analysis stage

take cofounder into account at the design stage by matching (during case-control matches)
- but over-matching can introduce bias, with introducing more ocnfoudners and trying to match each one

Question 16

what type of ratio is measured from case-control studies?

Answer 16

odds ratio

Question 17

what is the odds ratio?

Answer 17

first - estimating the odds of being exposed in the diseased group and the odds of being exposed in the non-diseased group

comparing the ratio of these odds with the odds ratio = odds of being exposed in diseased group/ non-diseased group odds

= number of cases exposed/ number of cases not exposed DIVIDED BY number of controls exposed/ number of controls not exposed

interpretation:
OR = 1 - no difference in odds of exposure between diseased & non-diseased groups
OR > 1 = increased exposure odds in diseased group

Question 18

interpreting 95% CL for odds ratios

Answer 18

for odds and risk ratios - a 95% CL containing 1 indicates a non-significant result

e.g. OR (insufficient v sufficient) = 1.17 (95% CI: 0.68 to 2.01)
- 95% certain the true odds ratio in the population from which the sample of taken is between 0.68 to 2.01
- however, as the interval contains 1 = insufficient evidence to reject Ho for an effect/ difference in exposure odds between cases and controls/ diseased and non-diseased groups

e.g. OR (deficient v sufficient) = 1.65 (95% CI: 1.10 to 2.50)
- 95% certain that the true odds ratio in the population from which this sample was taken is between 1.10 and 2.50 = no 1 within the interval means there’s sufficient evidence of the outcome being present more in cases vs controls

Question 19

potential types of bias in case-control studies?

Answer 19

selection bias - if the control group isn’t comparable, over-matching

information bias - differences in collecting data from cases & controls

recall bias - recalling past information may differ between cases and controls

interviewer bias - may influence information with questions asked

Question 20

historical case-control studies?

Answer 20

exposure information obtained from historical records

• reduces recall bias, information recorded prior to disease onset reduces chance of reverse causation
• record may be incomplete with missing cases/controls, extracting data may introduce bias unless blindness is introduced

Question 21

nested case-control study?

Answer 21

both the cases and controls are selected from an established cohort study

• reduces problem of reverse causation and recall bias
• reduces sample size, limitations present with lost follow-ups, needing large sample sizes

Question 22

advantages of case-control studies?

Answer 22

good for rare outcomes
quick and easy to carry out
relatively low cost
can investigate multiple exposures/
risk factors

Question 23

disadvantages of case-control studies?

Answer 23

prone to bias/ recall bias
hard to select controls with matching
can only investigate one disease
can’t measure incidence (with absolute or relative risk) - can only use odds ratios