9 - Observational studies

Question 1

What are observational studies?

Answer 1

Studying a group of individuals
Looking at the association/frequency between exposures and outcomes
Can be prospective or retrospective

Question 2

Define a prospective study:

Answer 2

Identify group of individuals with an exposure, follow them FORWARDS in time to monitor outcome occurrence

Question 3

Define a retrospective study:

Answer 3

Identify group of individuals with an outcome, work BACKWARDS to monitor exposures they experienced

Question 4

What are the three controlled aspects of observational studies?

Answer 4

Size of sample
Nature (prospective/retrospective)
Characteristics of subjects

Question 5

Describe the limitations of observational studies:

Answer 5

• they only demonstrate ASSOCIATIONS between exposure and outcome, and not causality
• confounding factors between exposure and outcome may mask true/create false associations

Question 6

Describe cohort studies and their benefits and limitations:

Answer 6

FORWARDS in time
- 2 groups (one with exposure and one without, well balanced otherwise)
- rates of outcome occurrence in each group are measured and compared
GOOD: follows natural history of disease
BAD:
- time consuming and expensive for rare outcomes
- subjects lost to follow up
- exposures may change over time

Question 7

Describe case control and their benefits and limitations:

Answer 7

BACKWARDS in time
- two groups artificially created (one with outcome and one without)
- analysed on presence of past exposures
GOOD: simple, quick, good for rare outcomes
BAD:
- difficult to get exact matching controls
- recall bias/inaccuracy
- prevalence/incidence cannot be determined as the controls are an artificially selected group

Question 8

What is relative risk and what kind of studies is it used to analyse?

Answer 8

Also called risk ratio
Used for cohort studies
It is the ratio of getting the outcome in the exposed vs unexposed group (probability that an event will occur)
On formula sheet (a/(a+c) / b/(b+d))
- RR = 1 means no difference between groups
- RR = <1 means lower risk of outcome in group 1
- RR = >1 means greater risk of outcome in group 1
Calculate CI using log equations (if CI contains 1 then the result is statistically insignificant)

Question 9

What is odds ratio and what kind of studies is it used to analyse?

Answer 9

It is the odds that an outcome will occur when exposure is present compared to when exposure is absent (ratio of event happening to not happening)
Used for case control studies
On formula sheet (a/c / b/d)
- OR = 1 means no difference in odds of exposure occurring between groups
- OR = <1 means lower odds of outcome in group 1
- OR = >1 means greater odds of outcome in group 1

Question 10

Why can we not use RR for case control studies?

Answer 10

As you are artificially enriching the sample with controls as it is a retrospective study, meaning the prevalence is not true

Question 11

What is adopted analysis?

Answer 11

When there is individual matching of cases and controls
Concordant pairs (both the exposed group of individuals) and discordant pairs (both the unexposed group of individuals) are compared, and OR = b/c