stats

Question 1

What is absolute risk

Answer 1

The probability or chance of an event. Divide number of events that occurred in group by the number of people in the group.
Can be written as fraction or decimal or percentage.

Question 2

What is absolute risk reduction

Answer 2

Also known as risk difference
Risk of bad outcome of a disease in the control group minus the risk of bad outcomes of a disease in the intervention group.
AR% in control group - AR% in intervention group.

Question 3

What is relative risk

Answer 3

AR in intervention group / AR in control group.

If RR is more than 1, the intervention is doing harm

If RR is less than 1, the intervention is likely to be beneficial (the risk of a bad outcome is lowered by the intervention).

Question 4

What is relative risk reduction.

Answer 4

This is absolute risk reduction/AR in the control group

How much risk is reduced in the intervention group compared to the control group.

Question 5

What is Cate plot

Answer 5

Smiley faces - green red and yellow.
Green is good outcome regardless of which group they are in, red is bad outcome regardless of which group they are in. Yellow is the people whos outcome is better with the intervention (so the better the intervention the more yellow faces there will be).

Question 6

What is number needed to treat

Answer 6

Number of patients who need to be treated with the intervention for 1 additional person to benefit.

1/ARR as decimal (or 100/ARR%). Lower NNT means more effective treatment.

Convention with number needed to treat is always to round up (even if NNT is 50.1 would round up to 51).

Question 7

What is NNH

Answer 7

How many patients need to be treated with the intervention for one additional person to be harmed.

Question 8

What is the hazard ratio

Answer 8

hazard (risk) of event in one group/hazard (risk) of event in the other group.

Hazard can be at any point in a study period.
A HR of 1 means the hazard in the first group is the same as in the second.
HR of 2 means twice the hazard over that time period. HR of 0.5 means half the hazard over that time period.

Question 9

Selective reporting bias

Answer 9

When results from research are not fully or accurately reported in order to suppress negative or undesirable findings.

Question 10

What does double blind mean

Answer 10

Neither patients nor researchers and clinicians do not know.

(in single blind researchers know but patients do not)

Main reason is to eliminate all bias and interference with results.

Question 11

Why randomisation

Answer 11

Another way to reduce bias

Question 12

What is primary and secondary outcome measures

Answer 12

Usually 1 or 2 primary outcomes. Most interested in.

Secondary outcomes - useful but not quite as important.

Question 13

What is intention to treat approach

Answer 13

There is always risk that patient will stop taking treatment, but this should happen equally across groups if placebo.
Patients are included for analysis sake in the group that they were alllocated to, regardless of how long they carried on taking the medicine for.
Intention to treat approach is the closest in decision making to real life.

Question 14

What is standard deviation

Answer 14

The variability around the mean.

Question 15

Confidence intervals

Answer 15

Likely range of values that for the whole population for effect. Wider if more variability in results, and wider if small sample size. If the confidence interval crosses 0, not statistically significant.

Question 16

What is P value

Answer 16

P value level of 0.05 is statistical significance.
Confidence interval will always reflect the P value.

Question 17

What is sample size calculation

Answer 17

Calculation before the study starts to work out how many patients they need to recruit to be able to pick up a statistically significant difference if there is one.

Question 18

What is confirmation bias

Answer 18

Confirmation bias relates to tendency to interpret information that confirms someones preconceptions.

Question 19

What is a cohort study?

Answer 19

a cohort study follows a group of people over time looking at a specific outcome and whether exposure to a given risk contributes to that outcome.

Question 20

positive predictive value

Answer 20

The proportion of those who test positive who actually have the disease

Question 21

What graphic is often used to display results of meta analysis

Answer 21

A forest plot. Results of individual studies are shown as squares centered on teh point estimate of the result of each study. The size of the square relates to the size and weighting of the study. The horizontal line through the square shows its confidence interval.

Question 22

Particular type of bias with case control studies

Answer 22

Recall bias.
As cases with the disease are more motivated to recall apparently trivial events in the past compared to controls (disease free)

Question 23

What is lead time bias

Answer 23

This is where it looks like people have a increase in survival, but actually, the condition has just been diagnosed and potentially treated early so there is a longer time living with the disease but no actual affect on survival.

(this is the case with some studies of cancer screening).

Question 24

What is selection bias

Answer 24

This is phenomenon of selecting individuals groups or data for analysis in a way that true randomisation is not achieved so sample is not representative of the whole population.

Question 25

What is a consort diagram

Answer 25

Consolidated standard of reporting trials group - was developed to stop inadequate reporting of RCTs. The consort statement is minimum set of recommendations for reporting RCTs. 25 item checklist and flow diagram showing the progress of all participants through the trial.

Question 26

Difference between calculating odds ratio and calculating risk ratios

Answer 26

In odds ratios, its the odds of having it over the odds of not having it. (so denominator is not having it). In risk ratio its those with the disease /everyone in the group (so denominator is total number in that group)

Question 27

Type 1 error vs type 2 error

Answer 27

Type 1 error is false positives - rejecting the null hypothesis when there is no true difference. This could occur by chance or poor scientific practice. Type 2 error is false negatives - accepting the null hypothesis when there is actually a difference. Problem here is underpowering.

Question 28

How to calculate power of a study

Answer 28

1 - (type 2 error). Therefore high type 2 error means low power /underpowered.