Interpreting Evidence Flashcards

1
Q

What are variations between people?

A

Effectiveness of treatment

Side effect or no side effect

Differential response to environment

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What are variations within people?

A

Measures of blood pressure over a day

Strength of left and right hands

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Give examples of continuous numerical data

A

Height

BP

Urinary lead concentration

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Give examples of discrete numerical data

A

Number of children in a family

Number of hand nodes

Number of adults consulting for shoulder pain

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Give examples of ordinal categorical data

A

Social class (I, II, IIIM etc)

Pain severity (mild, moderate, severe)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

Give examples of nominal catergorical data

A

Marital status (married, single, divorced)

Blood group (A, O, B, AB)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What is the interquantile range?

A

The middle 50% of data

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Which data summary can be influenced by outliers?

A

Mean

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Why is the median less informative than the mean?

A

Doesn’t use all the data

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What does positively skewed mean?

A

Data falls to the left with more lower scores and few high scores

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What does negatively skewed mean?

A

Data falls to the right with more higher scores and few high score

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What is a normal distribution?

A

Mean, median and mode are all centre of symmetrical graph

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Using a histogram, how many observations will be within 1SD and 2SD of mean?

A

68%

95%

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Risk =

A

Number with disease/ Total number at risk

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

ARR =

A

Risk in group 1 - Risk in group 2

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

Relative risk/ risk ratio =

A

Risk in group 1 / Risk in group 2

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

How can you interpret the RR?

A

< 1 - less risk in group 1
= 1 - same risk
> 1 more risk in group 1

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

NNT =

A

1/ ARR

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

Why can relative risk be misleading?

A

Independent of the original prevalence

Always need to state baseline (absolute) risk as well as relative risks

20
Q

When are odds-ratios used?

A

Case-control studies

Observational studies

i.e. retrospective

21
Q

How do you calculate odds?

A

Number exposed/ number unexposed

22
Q

How do you calculate odds ratio?

A

Odds (cases)/ odds (control)

23
Q

How do you interpret odds ratio?

A

OR = 1 –> equal odds

OR > 1 –> odds higher in cases

OR < 1 –> odds higher in control

24
Q

When are OR and RR roughly the same?

A

In very rare events

RR differs as it takes into account the baseline risk, however if the baseline risk is minute then the RR will be similar to the OR which is independent of baseline risk

25
Q

How does sample size correlate to confidence?

A

Larger sample, more confidence

26
Q

What is the standard error of mean?

A

How effective the sample mean is as an estimate of the population mean

27
Q

Why is SE always smaller than SD?

A

Theres less varability between sample means than there is individual values

28
Q

What does a small SE mean?

A

Not much variability between sample means so the sample is good representation of population mean

29
Q

What is the confidence intervlal?

A

Range of values estimated from sample data that is likely to include the true effect

30
Q

What is the purpose of confidence intervals?

A

Helps determine whether differences are due to chance or likely to be true effect

31
Q

What does a confidence interval = 1 mean?

A

Intervention same as control

32
Q

What does a confidence interval > 1 mean?

A

Intervention worse than control

33
Q

What is the T-test?

A

Test used to compare means between two groups

34
Q

What is the chi-squared test?

A

Test used to statistically compare frequencies

35
Q

What information is given in a T test?

A

Degrees of freedom = sample size - 1

T value

p value

36
Q

What must the p value be for the results to be statistically significant?

A

Less than 0.05

37
Q

What are the 3 variations of the T test?

A

Single value

Independent samples

Paired data

38
Q

What are the degrees of freedom in chi-squared?

A

(no of rows - 1) x (no of columns - 1)

39
Q

What is the chi-squared test for independence?

A

Used to find association between two categorical variables e.g. is cholesterol status associated to gender?

40
Q

What is the chi-squared test for goodness of fit?

A

Tests the difference between frequencies of a single categorical variable and some hypothesised frequency

e.g. is the frequency of depression sufferers in our sample the same as the proportion quoted in the literature

41
Q

What is correlation?

A

Measure of strength of relationship between two numerical variables

e.g. relationship between maternal age and daughter age at menarche?

42
Q

What does it mean if the correlation coefficient (r) = +1?

A

Perfect positive correlation i.e. as one variable increases so does the other

43
Q

What does it mean if r = -1?

A

Perfect negative correlation i.e. as one variable increases the other decreases

44
Q

What is linear regression?

A

Prediction of relationship between independent variable and an outcome

45
Q

What is the regression coefficient?

A

Slope of best fitting straight line i.e. the gradient

46
Q

What does a p value of less than 0.05 mean with regards to regression coefficient?

A

Regression coefficient is not 0 i.e. alternative hypothesis true, relationship exists