# Critical appraisal / QIP Flashcards

1
Q

What is absolute risk reduction?

A

Measure of effect
Incidence of outcome in experimental group minus the incidence of outcome in the control group

= CER - EER

2
Q

Of 100 patients in the treatment group, 20 died. 30 of 100 died in the placebo group. What is the absolute risk reduction?

A

30/100 (0.3) - 20/100 (0.2) = 10/100 or 0.1 or 10%

3
Q

What is bias?

A

Systematic difference in the way a study is conducted that leads to less valid results

4
Q

What is bias ascertainment?

A

Bias when measurements are performed differently

5
Q

What is bias expectation?

A

Bias when those collecting data are influenced by their own knowledge

6
Q

A

When loss to follow up differs between two groups

7
Q

What is bias inclusion?

A

When subjects included in groups are not at random

8
Q

What is a confidence interval?

A

Confidence an estimated effect is between two values - 95% normally chosen confidence interval

9
Q

What value makes a confidence interval statistically significant?

A

<1

10
Q

What is a confounding variable?

A

Variable associated with both the exposure and outcome of interest which can lead to inappropriate conclusions re: association

11
Q

What is effectiveness?

A

How well an intervention works in the real world

12
Q

What is efficacy

A

How well an intervention works under ideal conditions

13
Q

What is heterogenecity?

A

The amount of variation of results of trials included in systematic reviews. Less heterogenecity increases likelihood of results being true

14
Q

Describe intention to treat analysis

A

Patients analysed by which group they end up in, not where they were allocated.

I.e if patients stopped take the treatment for whatever reason, they are still analysed in the intervention arm. This increases chance of type 2 error (i.e. less likely to show a difference) but takes into account what happens in real life.

15
Q

A

Rank of methodological quality (0-5)

16
Q

What is kappa?

A

Concordance between observers.
1 = perfect agreement
0 = absolute disagreement

17
Q

What is a negative likelihood ratio?

A

how likely is a negative test to be found in a person without the disease than in a person with the disease?

How good a test is at excluding a disease

18
Q

How is a negative likelihood ratio calculated?

A

(1 - sensitivity) / specificity

i.e. false negative / true negative

19
Q

What values of a negative likelihood ratio show a useful test?

A

Number will be <1.0
<0.2 useful
<0.1 very useful

20
Q

What is a positive likelihood ratio?

A

How much more likely is a positive test to be found in a person with the disease than in someone without the disease

ie. How good is a test at ruling in a disease

21
Q

How is positive likelihood ratio calculated?

A

(sensitivity / (1- specificity)

i.e. true positive/ false positive

22
Q

What positive likelihood ratios values show usefullness?

A

Value will be >1.0
5-10 useful
> 10 very useful

23
Q

What is sensitivity?

A

How many true positives does a test pick up as a proportion of all positives

24
Q

How is sensitivity calculated?

A

A/ A + C

25
Q

What is specificity?

A

How many true negatives dose a test pick up, as a proportion of all negatives?

26
Q

How is specificity calculated?

A

D / B + D

27
Q

What is negative predictive value?

A

If a patient has a negative test, how likely are they not to have the disease?

28
Q

How is a negative predictive value calculated?

A

D/ C + D

29
Q

How is positive predictive value calculated?

A

A / A + B

30
Q

What is positive predictive value?

A

If a patient has a positive test what is the probability of them having the disease?

31
Q

What is number needed to treat?

A

Inverse of absolute risk reduction
(1/ARR)`

32
Q

How is NNT calculated?

A

1/absolute risk reduction
or 100/ ARI as a %

33
Q

How is number needed to harm calculated?

A

1/ absolute risk increase
or 100/ ARR as a %

34
Q

What is an odds ratio?

A

Measure of effect from case control studies

Odds in experimental group / odds in control group

35
Q

What does an odds ratio of 1 mean?

A

No effect

36
Q

How is an odds ratio calculated?

A

Odd of risk in exposed group / odds of risk in non-exposed group

= (a/b) / (c/d)

37
Q

What is a type 1 error

A

Finding a truth, when in fact there is none

38
Q

What is a type 2 error

A

Failing to find a difference, when in fact there is one

ie. accepting the null hypothesis, when it is correct

39
Q

What is regression?

A

Statistical technique where all known confounders are held constant so that an association between variables can be examined

40
Q

What is a relative risk and how is it calculated?

A

It is a measure of effect

Incidence of outcome in experimental group / incidence of outcome in control

EER / CER

41
Q

What is a case control study?

A

Type of observational study. Researchers look at a set of patients with a certain disease and then looks to what risk factors/exposures might have led to the disease.

42
Q

What is a cohort study?

A

Type of observational study. Patients exposed to treatment/risk factor are identified and followed up to see what happens.

NB: the researchers do not affect who gets exposure/medicine, which is what makes it observational

43
Q

What is an observational study?

A

Study where researchers do not ‘do’ anything to the patients or objects. Merely observe and record what happened, has happened

44
Q

What is a ROC curve?

A

Graphical plot of sensitivity against (1-specificity) for the range of possible cut off values for the test.

45
Q

What does an increase in area under the curve represent in a ROC curve?

A

More sensitive test

46
Q

What is the optimum value for area under curve?

What is the AOC value for a useless test?

A
1. 1.0
2. 0.5
47
Q

What is a forest plot

A

Graphical representation of meta-analysis showing results of RCT

Demonstrates degree of data overlap and therefore heterogenecity

48
Q

When is a case control study useful?

A

If the disease is rare

49
Q

When is a cohort study useful?

A

When the you want to known about of lot of different outcomes about a particular exposure/medicine.

50
Q

What is a cross sectional study?

A

Researchers look at either exposure or disease, or both at a single point in time. Most commonly used to demonstrate prevalence. ‘Snapshot’ of population

51
Q

What is a confounder?

A

An extraneous factor that can affect the outcome and lead to invalid inferences

e.g. age, gender, smoker etc.

52
Q

How do we calculate risk?`

A

Risk = number of times an event occurs / total number of events

53
Q

How do we calculate relative risk?

A

Risk in experimental group / risk in control group

EER / CER

54
Q

How do calculate relative risk reduction?

A

Difference in risk between tx and control group / risk in control group

= (CER-EER) / CER
= ARR / CER

55
Q

How do we calculate odds in treatment group?

A

Positive outcome treatment group/ Negative outcome treatment group

= a/b

56
Q

How do we calculate risk in control group?

A

Number with outcome in control group / total number in control group
CER = c/(c + d)

57
Q

How do we calculate risk in experimental group?

A

Number with outcome in experimental group / total number in experimental group
EER = a/(a+b)

58
Q

How do we calculate odds in control group?

A

Positive outcome in control group / negative outcome in control group

= c/d

59
Q

What is on the x and y axis of a ROC curve?

A

x = sensitivity (true positive)
y = (1 - specificity) or false positive

60
Q

What part of a ROC curve shows statistically most sensitive value?

A

Area closest to top left corner

61
Q

What is Lean process?

A

Lean uses process mapping with associated stakeholders to identify inefficiencies in care, enabling actions for improvement.

62
Q

What is Six Sigma?

A

Six sigma uses DMAIC and control charts are used to study adjusted processes over time.

63
Q

What is DMAIC?

A

Define
Measure
Analyse
Improve
Control

64
Q

What type of QIP design works in partnership with patients and families to improve services from their perspective.

A

Experience Based Co-Design (EBCD)

65
Q

What is are run charts and statistical process charts used for?

A

Plots data over time in order to show the effect of interventions

66
Q

What is the difference between run charts and statistical process charts?

A

Run charts simply data plotted over time, SPC uses statistical processes as well.

67
Q

With regards to a run chart what is a ?
1. Shift
2. Trend
3. Run

A
1. 6 or more points above or below median line, signifies change unlikely to be chance
2. 5 or more points consecutively increasing or decreasing
3. Indicates if sufficient data points exist. Number of times the data crosses the median line. Depending on data set, will have prescribed number of runs needed for data to be sufficient
68
Q

What are the 6 S’s of analysis and options (QI )

A
• Strategy
• Skills
• Style
• Shared Values (indefinable)
• Structure (allocation of staff)
• Systems (budgets, training, audit, communication
69
Q

What is a SWOT analysis?

A

Strengths
Weaknesses
Opportunity
Threats

70
Q

What is a stakeholder analysis used for?

A

Establishing how stakeholders will affect change process, and how they should be ‘managed’.

71
Q

What is a forcefield analysis?

A

Method of establishing the drivers and resistors for change (and the magnitude), to assist with planning of change process

72
Q

What is a Healthcare Failure Mode and Effects Analysis (HFMEA)

A

QI method base on asking staff - what could go wrong?

73
Q

What is a Ishikawa (Fishbone) diagram?

A

Visual representation of causes and sub-causes and what actions could be considered to affect change (or could have led to an error)

74
Q

What is the ‘5 whys’ method of root cause analysis?

A

Ask why 5 times, the answer to the fifth is the root cause

75
Q

How do you calculate the false positive rate?

A

1 - specificity

76
Q

How do you calculate the false negative rate?

A

1 - sensitivity