5 - Study designs (part II): Experimental epidemiological studies

Question 1

Randomised control trial

Answer 1

• prospective studies
• each individual is allocated a treatment at random
• disease status of each individual is later measured

Question 2

Why randomize?

Answer 2

• Protect against unsuspected sources of bias

- No guarantee, but an attempt to equalize groups

Question 3

modes of randomisation

Answer 3

• Completely randomised design (equi-probability)
• Randomized block design (block = strata)
• Split-splot
• Crossover
• Sequential

Question 4

simple randomisation

Answer 4

each individual has equal probability of being selected

Question 5

block randomisation

Answer 5

split into groups, and allocate randomly

Question 6

RCT trial hypothesis

Answer 6

superiority trial/ equivalence trial

s: is it better than nothing/placebo/current?
e: is it as good as?

Question 7

RCT phases

Answer 7

I: safety and feasibility
II: pilot test
III: large scale trials
IV: follow up surveillance

Question 8

internal validity

Answer 8

the scientific integrity of the trial

Question 9

external validity

Answer 9

the extent to which the results generalise to the wider population

Question 10

Blinding

Answer 10

procedure whereby people involved in the study do not know which group each participant is in.

• removes bias

Question 11

single blind

Answer 11

the participants don’t know which arm they are in

Question 12

double blind

Answer 12

neither the scientists nor the participants know which arm they are in

Question 13

triple blind

Answer 13

neither the scientists, the participants, nor the oversight committee know which arm they are in

Question 14

decision making

Answer 14

• Eligibility
• Enough subjects?
• Stratification
• What to measure and how to analyse

Question 15

bias

Answer 15

any systematic error that results in an incorrect estimate of the effect of an exposure on the risk of a disease

Question 16

selection bias

Answer 16

People selected for inclusion in study are different from those not selected

Question 17

measurement/ classification bias

Answer 17

The measurement or classification of the exposure or the outcome

Question 18

non-differential bias

Answer 18

bias occurs equally in both case and control groups

Question 19

differential bias

Answer 19

bias occurs at different levels in case and control groups

Question 20

recall bias

Answer 20

Systematic error due to differences in accuracy or completeness of recall to memory of past events or experiences.

Question 21

observer bias

Answer 21

tendency of observers to see what they expect to see

Question 22

accuracy

Answer 22

The degree to which a measurement or an estimate based on measurements represents the true value of the attribute being measured.
= validity

Question 23

precision

Answer 23

The degree to which results can be repeated to give the same value

= Reliability

= Reproducibility

= Stability of measurement

Question 24

Confounding

Answer 24

occurs when a risk factor appears to be associated with a disease, but this association is actually caused by the presence of one or more other (possibly unobserved) risk factors

Question 25

how to correct confounding

Answer 25

In design: matching

In data analysis: ‘stratification’ or ‘adjustment’

Question 26

how to correct bias

Answer 26

Very difficult to correct for bias in data analysis

Possible sources of both bias and confounding should be considered before a study is carried out