Medical Statistics Flashcards
Which of the following matches the following definition?
The ability of a test to correctly identify people with a disease, or true positives. A highly sensitive test has few false negatives, so it misses fewer cases of disease.
1 - specificity
2 - sensitivity
2 - sensitivity
Gives a TRUE positive
Which of the following matches the following definition?
The ability of a test to correctly identify people without a disease, or true negatives. A highly specific test has few false positives.
1 - specificity
2 - sensitivity
1 - specificity
Gives a TRUE negative
A patient has a suspected PE, which of the following tests is the MOST sensitive?
1 - chest X-ray
2 - CTPA
3 - d-dimer
4 - troponin
3 - d-dimer
> 99% sensitive
Essentially means it is very very unlikely that someone with a DVT or PE will not be positive with a d-dimer
A new test to detect PE is designed. The results are in the image below. What is the specificity of this test?
1 - 85
2 - 75
3 - 71
4 - 65
5 - 35
4 - 65
What is the positive predictive value of the new test below for detecting PE?
1 - 85%
2 - 65%
3 - 71%
4 - 81%
3 - 71%
True Positive / (True Positive + False Positive)
Negative Predictive Value = 81%
True Negative / (True Negative + True Negative)
A new drug is developed to reduce the risk of stroke, but some patients developed gout when taking it. The data below is from a cohort study identifying how many people with and without diabetes developed gout when taking the new drug. Using the data below, what is the relative risk of a patient with diabetes developing gout?
1 - 6.7
2 - 3.2
3 - 1.6
4 - 0.6
5 - 0.2
Risk = risk of developing gout on new drug
Relative risk = risk of developing gout in patients WITH T2DM vs WITHOUT T2DM
Risk ratio and relative risk are the same thing
2 - 3.2
Means patients with T2DM are >3 times more likely to develop gout than patients without T2DM
Risk of gout in T2DM = 54 / (54+33) – 0.6
Risk of gout in non-T2DM = 23 / (23+96) – 0.2
Relative risk = (54 / (54+33)) / (23 / (23+96))
Score of 1 means thing being studies has no effect
Score >1 means thing being studies has detrimental effect
Score <1 means thing being studies has beneficial effect
An odds ratio (OR) is a statistical measure that compares the odds of an event occurring in one group to the odds of it occurring in another group. What is the odds ratio of patients with T2DM developing gout?
1 - 6.7
2 - 3.2
3 - 1.6
4 - 0.6
5 - 0.2
1 - 6.7
OR
- With T2DM + gout / With T2DM + NO gout = 1.636
- Without T2DM + gout / Without T2DM + NO gout = 0.242
1.636 / 0.242 = 6.7
This means patients with T2DM are 6.7 times more likely to develop gout if taking this new drug compared to those with T2FM
Score of 1 means thing being studies has no effect
Score >1 means thing being studies has detrimental effect
Score <1 means thing being studies has beneficial effect
Look at the question in the image:
- High blood pressure is the correct answer
Score of 1 means thing being studies has no effect
Score >1 means thing being studies has detrimental effect
Score <1 means thing being studies has beneficial effect
If a CI spans 1 (below and above 1, meaning NO effect), it is unlikely to be significant. In the image we have:
- CI of family history = 0.8-4.5
- CI of high BMI = 0.7-2.11
Here high BP had the higher OD (>1 meaning high BP has a detrimental effect), so therefore is correct
Look at the question in the image:
B. 2
When looking at the boxes, the size relates to the weighting in the meta analysis. So larger boxes mean more patients in that study.
The lines either side of the box relate to CI, where anything spanning 1 for a CI is NOT statistically significant
The overall box:
- middle of diamond relates to overall RR or OR
- width represents the confidence interval
Look at the question in the image:
D. 25%
Absolute risk reduction = risk in controls - risk in treatment
- Control 20% deaths vs Treatment 15% deaths = 5% reduction
Relative risk reduction = absolute risk / risk in control group
- Absolute risk reduction = 5% / 20% = 25%
This is because the drug being studied is relative to the control group.
Look at the question in the image:
D. 20 patients
- Here treating 100 people leads to a 5% (or in this case 5 patients out of 100) reduction in vascular events.
- 100/5 = 20.
- So treating 20 people means 1 person benefits
- Also do 100 / % absolute risk reduction
