2 GATE: Graphic Appraisal Tool for Epidemiology - PowerPoint Presentation

Slide1

Slide2

2

Slide3

GATE:

Graphic Appraisal Tool for Epidemiology

1991 - 2015

3

1 picture, 2 formulas & 3 acronyms

Slide4

GATE:

Graphic Appraisal Tool for

Epidemiology

Graphic

Architectural Tool

for

Epidemiology

Graphic

Approach To Epidemiology

making epidemiology accessible

4

Slide5

4

th

year medical students 1991

Slide6

Jerry Morris

n

umerator denominator

epidemiology =

In: Uses of Epidemiology 1977

6

Slide7

Slide8

p

resentation outline

GATE is a framework for:

study design

study analysis

study error

practicing EBM

8

1 picture, 2 formulas & 3 acronyms

Slide9

1

Slide10

GATE: a framework for study design

1 picture

10

every epidemiological

study can be hung on the GATE frame

1

picture

, 2 formulas & 3 acronyms

Slide11

cohort

of British doctors

non-smokers

smokers

lung cancer events counted

yes

no

followed

for 10 years

smoking status allocated by measurement

(

observation)

cohort

/

longitudinal

/

follow-up study

11

1

picture: GATE frame

1

picture

, 2 formulas & 3 acronyms

Slide12

British doctors

placebo

aspirin

MI

yes

no

5 years

r

andomly allocated to aspirin or placebo

12

1

st

acronym: PECOT

Participants

Comparison

Exposure

Outcomes

Time

P

E

C

O

T

r

andomised controlled trial

1 picture, 2 formulas & 3

acronyms

Slide13

m

iddle-aged Americans

‘normal’ weight

overweight

d

iabetes status measured in all participants

yes

no

b

ody mass index measured

c

ross-sectional (prevalence) study

13

P

E

C

O

T

Slide14

m

iddle-aged American women

b

reast cancer

m

ammogram negative

yes

no

receive mammogram screening test

d

iagnostic

test (prediction) study

m

ammogram positive

14

P

E

C

O

T

Slide15

m

iddle-aged American women

m

ammogram

test

no breast cancer

positive

negative

Gold Standard

d

iagnostic (test accuracy) study

b

reast cancer

15

P

E

C

O

T

Slide16

non-smokers

smokers

l

ung cancer

yes

no

smoking status measured

c

ase-control study

16

P

E

C

O

T

cases

controls

(all nested in virtual cohort studies)

Slide17

£100

Slide18

2

Slide19

GATE: a framework for study analysis:

1

st

formula: occurrence = outcomes ÷ population

19

the numbers in epidemiological studies can be hung on the GATE frame

1 picture, 2

formulas

& 3 acronyms

Slide20

British doctors

non-smokers

smokers

Lung cancer

yes

no

10 years

smoking status measured

20

1

st

formula

: occurrence of outcomes =

n

umber of outcomes ÷ number in population/group

Participant Population

Comparison Group

Exposure Group

Outcomes

Time

P

EG

CG

O

T

a

b

Slide21

British doctors

non-smokers

smokers

Lung cancer

yes

no

10 years

smoking status measured

21

Population

Comparison Group

Exposure Group

Outcomes

Time

P

EG

CG

O

T

Exposure Group Occurrence (EGO) = a

÷

EG

= number of outcomes (a) ÷ number in exposed population (EG)

a

b

Slide22

British doctors

placebo

aspirin

MI

yes

no

5

years

r

andomly allocated

22

Population

Comparison Group

Exposure Group

Outcomes

Time

P

EG

CG

O

T

a

b

Comparison Group Occurrence (CGO) = b

÷

CG

= number of outcomes (

b

) ÷ number in comparison population (CG)

Slide23

yes

no

23

E

pidemiology =

N

umerator ÷

D

enominator

Participant Population

Comparison Group

Exposure Group

Outcomes

Time

P

EG

D

O

T

a

N

m

iddle-aged American women

b

reast cancer

m

ammogram negative

receive mammogram screening test

m

ammogram positive

Slide24

British doctors

non-smokers

smokers

Lung cancer

yes

no

10 years

smoking status measured

24

t

he goal of all epidemiological studies is to calculate

EGO and CGO

P

EG

CG

O

T

a

b

EGO:

Occurrence (risk) of cancer in smokers

C

GO:

Occurrence of cancer in non-smokers

Slide25

Middle-aged Americans

Low BMI

High BMI

EGO:

Average blood glucose in EG

high

low

Body Mass Index (BMI) measured

25

P

EG

CG

O

C

GO:

Average blood glucose in CG

Slide26

Middle-aged Americans

Low BMI

High BMI

b

lood glucose

high

low

Body Mass Index (BMI) measured

c

ross-sectional study with numerical measures

26

P

E

C

O

T

Slide27

Middle-aged American women

mammogram

no Breast cancer

positive

negative

Gold Standard

Breast cancer

27

P

E

C

O

T

EGO:

likelihood of a positive mammogram if breast cancer

C

GO:

likelihood of a positive mammogram if no breast cancer

Slide28

EGO ÷ CGO

= Relative

Risk (RR

)

EGO – CGO

= Risk Difference

(RD)

i

ts all about EGO and CGO

measures of occurrence

: risk; rate; likelihood; probability; average; incidence; prevalence

28

1

st

formula:

occurrence

= outcomes ÷ population

Slide29

3

Slide30

GATE: framework

for

nonrandom error

2

nd

acronym: RAMBOMAN

30

R

ecruitment

A

llocation

M

aintenance

B

lind

O

bjective

M

easurements

AN

alyses

1 picture, 2 formulas & 3

acronyms

Slide31

R

AMBOMAN

Recruitment of participants

‘who

are the findings

applicable to?’

P

P

Study setting

Eligible population

31

r

ecruitment process

Slide32

EG CG

O

T

RCT: a

llocated by

randomisation

(

e.g

to drugs

)

EG CG

O

T

Cohort: a

llocated

by measurement

(e.g. smoking)

R

A

MBOMAN:

‘

were participants well

Allocated

to exposure & comparison groups?’

EG & CG similar at baseline?

w

as

Allocation

to EG & CG successful?

32

E & C measures accurate?

Slide33

RA

M

BOMAN

EG CG

O

T

‘were Participants well

Maintained

in the groups they were allocated to?’

P

completeness

of follow-up

c

ompliance

c

ontamination

c

o

-

interventions

33

Slide34

RAM

B

O

M

AN

EG

CG

O

T

‘were

outcomes

well

Measured

?’

w

ere they measured

Blind

to whether participant was in EG or CG

?

P

34

Slide35

RAMB

OM

AN

EG

CG

O

T

‘were

outcomes

well

Measured

?’

w

ere they measured

Objectively

?

P

35

Slide36

RAMBOM

AN

EG

C

CG

C

O

T

P

36

EG

A

CG

A

a

b

‘were the

ANalyses

done well

?’

If RCT were

I

ntention

T

o

T

reat (ITT)

analyses done?

Slide37

RAMBOM

AN

EG CG

O

T

‘were the

ANalyses

done well

?’

P

a

djustment for baseline differences / confounding?

37

Slide38

GATE

:

random error:

2

nd

formula:

random error = 95% confidence interval

38

There is about a 95% chance that the true

value in

the underlying

population

lies

within

the 95% CI (assuming no non-random error)

EGO ± 95% CI

C

GO ± 95% CI

1 picture, 2

formulas

& 3 acronyms

s

ample from a population

Slide39

GATE

:

a framework for error in systematic reviews & meta-analyses:

3

rd

acronym: FAITH

1 picture, 2 formulas & 3

acronyms

39

Slide40

 

study sources

s

tudies appraised & allocated:

included

excluded

s

tudies summarised

& pooled if homogeneous

systematic review: a study

of studies

studies

screened

40

Slide41

 

study sources

s

tudies appraised & allocated:

included

excluded

s

tudies summarised

& pooled if homogeneous

studies

screened

critical appraisal of SR:

FAITH

Find

Appraise

Include

Total

H

eterogeneity?

1 picture, 2

formulas

& 3 acronyms

41

Slide42

4

Slide43

GATE: framework for the 4 steps of EBP

43

Slide44

t

he steps of EBP:

A

sk

Acquire

3. Appraise

4. Apply & Act

44

Slide45

yes

no

45

1. Participants

3. Comparison

2. Exposure

4. Outcomes

5. Time

P

E

C

O

T

EBP Step 1:

ASK

- turn your question into

a focused 5-part PECOT question

Slide46

EBP Step

2:

ACQUIRE

the evidence – use

PE

COT

to help choose search terms

46

yes

no

Participants

Comparison

Exposure

Outcomes

Time

P

E

C

O

T

Slide47

P

E

C

O

T

P

E

C

O

T

R

ecruitment

A

llocation

M

aintenance

b

lind

o

bjective

M

easurements

AN

alyses

47

EBP Step 3:

APPRAISE

the evidence – with the picture, acronyms & formulas

Occurrence = outcomes ÷ population

Random error = 95% Confidence Interval

Slide48

APPLY

the evidence by AMALGAMATING the

relevant information &

making

an

evidence-based decision:’

the X-factor

©

48

Slide49

49

Slide50

e

pidemiological evidence

p

atient’s clinical circumstances

s

ystem features

v

alues & preferences

X-factor: making evidence-based decisions

Practitioner

e

X

pertise

: ‘putting it all together’

- the

art of practice

e

conomic

legal

political

person

family

community

practitioner

social

p

hysical health

psychological

Clinical expertise in the era of evidence-based medicine and patient choice. EBM 2002;736-8 (March/April)

50

Slide51

GATE critically appraised topic (CATs) forms

Slide52

GATE CAT –

4-

sheet workbook (in Excel)

sheet 1: GATE-Ask &

Acquire

52

Slide53

53

GATE CAT – 3-sheet workbook (in Excel)

sheet 2: GATE-Appraise

(with calculator)

Slide54

54

GATE CAT – 3-sheet workbook (in Excel)

sheet 3: GATE-Apply

Slide55

Slide56

Slide57

57