PHEBP - estimates of treatment effect

Question 1

Sampling bias

Answer 1

• We use samples to provide estimates of population parameters
• eg. healthy worker bias if you get men BP and get working men
  => avoid it with random sampling

Question 2

What is 95% reference range and how do you calculate it?

Answer 2

mean = 140mmHg
SD= 20mmHg
N=1500

140 + 1SD(20) = 160
140+2SD(40)= 180 (upper range)
140- 40 = 100(lower range)

-so 95% of men will have BP values between 100 to 180mmHg

Question 3

Can reference range for estimates or group averages?

Answer 3

• ref range is used for continuous numerical data(height, weight)
• no mean or standard deviation so cant calculate ref range

Question 4

Define precision

Answer 4

• inherent error associated with any estimate,
  about being accurate
  -small sample = less precise

Question 5

Define Accuracy

Answer 5

getting the right answer

avoiding bias

Question 6

What are confidence intervals?

Answer 6

• uncertainty in everything you measure
• measure of statistical uncertainty or ‘precision’
• influenced by sample size

Question 7

how would sample estimates and data be distributed?

Answer 7

• spread of Data is standard Deviation

- spread of Estimates is standard Error

Question 8

difference between ref range and confidence error

Answer 8

2 stand dev = ref rate

2 stand error = confidence interval

Question 9

how to work out standard error?

Answer 9

SE= SD/ sample size

Question 10

how to calculate confidence interval?

Answer 10

CI= mean+/- (2x SE)

Question 11

What is risk ratio(RR)?

Answer 11

• unitless ratio
• RR = risk of outcome in drug/risk of outcome in placebo
• ratio is 1 if drug had no effect
  -accept null
  RR - 1=