MEAN

:

Average

continuous

data-normally

distributed

MEDIAN

:

Middle

ordinal

data

continuous

data-skewed

*

#him

MODE

:

most

frequent

nominal

data

ora

!

Queried

mirr

Des

difference

between

categories

is

not

equal

EX

:

gender

,

ethnicity

,

Marital

status

,

mortality

EX

:

NYHA

FXNI

Class

1- 11

,

Painscale

0-18

Ex

:

Age

,

height

.

Weight

,

time

,

BP

EX

:

celcius

&

farenheit

temp

scales

NO

SET

ORDER

ORDERED

EQUAL

I

ONE-SAMPLE

PAIRE D

STUDENT

HYPOTHESIS

FOR

SIGNIFICANCE

·

product

is

significantly

better

(superior)

to

current

treatment

or

placebo

·

Trial

needs

to

demonstrate

null

hypothesis

is

not

true

and

should

be

rejected

and

alternative

can

be

accepted

.

NUll

Hypothesis

:

no

statistically

significant

difference

Al

What

reacher

triso

promo reje

ht

difference

↳

What

researcher

tries

to

prove

or

accept

Error

margin

=

Alpha

(c)

=

0

.

85

p-value

is

compared

to

Alpha

P

>

8

.

85

9

-

885

-

NULL

:

REJECTED

$$$

FAILED

to

REJECT

NULL

NUL

acce

T

Alternative

Hypothesis

Accepted

Alternative

Hypothesis

Rejected

STATISTICALLY

SIGNIFICANT

NOT

STATISTICALLY

SIGNIFICANT

MOST

RELIABLE

confidence

Intervais

(CI)

:

same

into

as

p-value

+

precision

of

result

.

Cl

=

1

-

c

EX

:

0

=

0

.

05

-

11

:

95

%

Narrow

=

Precision

Wide

:

less

precise

SYSTEMATIC

comparing

Difference

Data

(means)

:

REVIEWS

-

NOTiNudeUOSSF

:

X

META-ANALYSIS

comparing

Ratio

Data

(RR

,

OR

,

HR)

:

RANDOMIZED-CONTROL

15

%

(1

does

NOT

include

one

=

statistically

significant

:

10

.

12-0

.

99)

/

95

%

(1

includes

one

:

NOT

statistically

significant

:

10

.

61-1

.

29)

X

I

COHORT

TYPE1

ERROR

:

FALSE

POSITIVE

TYPE2ERROR

:

FALSE

NEGATIVE

CASE-CONTROL

Cl

=

1

-

D

(type

<error)

Null

hypothesis

accepted

when

shoulde

been

rejected

CASE

SERIES/

REPORTS

B(neta)

usually

8

.

1

or

0

.

2

meaning

risk

of

Type

11

=

18

%00

20

%

Power

to

Avoid

LEAST

RELIABLE

EXPERTOPINION

STUDY

POWER

:

Type

11

error

power

=

1-B

RISK

ODDSRATIO

HALARDRATIG

RISK

=

#

subjects

in

group

w

unfavorable

event

CASE

CONTROL

STUDIES

-

OR

used

to

estimate

risk

of

unfavorable

events

Total

number

or

subjects

in

group

B

Risk

in

treatment

group

B

-UTCOME

NO

OUTCOME

Risk

in

control

group

Treatment

lowers

Equal

risk

between

Treatment

increases

TREATMENT

EXPOSURE

A

B

A

=

#

that

have

outcome

w

exposure

risk

of

outcome

treatment

-

control

risk

of

outcome

-

B

:

I

that

don't

have

outcome

w

exposure

calculate

risk

of

HF

progression

Risk

metoprolol

=

=3

=

0

.

16

CONTROL

NO

EXPOSURE

C

D

=

#

that

have

outcome

w

no

exposure

in

each

group

then

calculate

RR

.

Risk

Control

:

=

0

.

28

Patients

treated

w

metoprolol

were

57

%

as

likely

to

RR

:

AS

likely

(vs

.

control)

D

:

I

that

don't

have

outcome

who

expose

F

MVOO

CONT

O

RR

:

=

0

.

57

:

37

%

have

progression

of

disease

as

placebo-treated

patients

.

RRR

:

LESS

likely

(vs

.

control)

RR

+

RRR

=

100

%

metoprolol

pts

&3

%

·

RELATIVE

RISK

REDUCTION

HOW

MUCHM

=

I-RRIO

like

A

placebo

treated

pts

.

ARR

=

0

.

28

-

0

.

16

=

8

.

16

or

16

%

·

ABSOLUTE

RISK

REDUCTION

-risin

control-riskin

treatment

m

will

have

-

promission

ODDSRATIO

:

BC

>

Additional

benefit

of

ARR

is

to

use

the

inverse

to

determine

number

needed

to

treat

(NNT)

and

number

needed

to

harm

(NNH)

NUMBER

NEEDED

TREAT

MWhmh3VVpamemprone

,

As

HAZARDRATIG

*

ALWAYS

round

up

for

NNT

1

For

everyo

patients

receive

metoprolol

for

one

year

,

one

additional

case

of

HF

progression

will

occur

.

NUMBER

NEEDED

HARM

METScWhon

T

Ke

down

for

NNI

#YES

or

STUDY

DATA

Range

:

Difference

between

highest

o

lowest

values

STANDARD

DEVIATION

:

Dow

Spread

out

dais

,

And

What degree

Highly

Dispersed

=

large

a

EX

:

L

:

6

.

2

=

3

.

8

(mean

=

so)

-

3

.

2

%.

4

2

.

4

6

.

2

18

13

.

8

17

.

6

68

%

as

.

Dose

TYPES

or

STATISTICALTESTS

Independ

variable

:

changed/manipulated

by

researcher

to

determine

if

it

has

an

effect

on

DV

.

EX

:

drugs

,

doce

,

placebo

,

pts

included

(age

,

gender

,

comorbid

conditions)

-OGROUPS

DISCRETE

CATEGORICALDATA

CONTINUOUSDATA

Dependent

Variable

:

outcome

EX

:

HF

progression

.

HbAlC

,

Bo

.

Cholesterol

,

Mortality

NORMAL

DISTRIBUTION

NOT

NORMAL

DISTRIBUTION

Primary

Endpoint

:

main

result

measured

to

see

if

treatment

had

significant

benefit

Ex

:

Death

from

CV

disease

or

nonfatal

stroke

or

nonfutal

M1

I

=

PARAMETRIC

I

=

NONPARAMETRIC

I

composite

Endpoint

:

combines

multiple

individual

endpoints

into

one

measurement

.

(With

BEFORE

Op

ER

MEASURES

CHISQUARED

TTEST

Sign

test

EX

:

Death

from

CV

disease

AND

nonfatal

stroke

AND

nontutal

M1

Wilcoxon

Signed-Rank

TTEST

Wilcoxon

Signed-Rank

2

CHISQUARED

Or

FISCHERS

EXACT

INPAIRED

TTEST

/TREXTMENT

O

CONTROL

3

Kouskal -Wallis

ANOVA

Mann-Whitne

a

CHISQUARED

:

T-TESTS

:

·

Nominal

or

Ordinal

data

↳

invin

Movingroups

mywhodrum gema

popular

↳

EX

:

comparing

reduction

in

HbA/C

between

metformin

placebo

A: ANALYSIS

O

NaviANCIFTS

or

groups

Intent-to-treat

:

includes

data

from

ALL

TYPES

or

MEDICAL

STUDIES

pts

,

even

if

they

didn't

J

complete

trial

Equivalence

trial

:

see

if

new

treatment

has

same

effect

Non-inferiority

:

No

worre

than

current

Summary

of

clinical

literature

on

specific

topic

or

question

Results

from

multiple

studies

for

greater

statistical

power

Pts

randomized

-

Egoal

chance

of

getting

treatment/sometimes

blinded

:

Double

blind

:

Both

pt/investigator

unaware

·

single

blind

:

It

is

unaware

·

Open

label

:

All

parties

know

treatment

PARALLEL

CROSSOVER

COHORT

=

EXPOSUR

!

PROSPECTIVE

compares

group

of

pos

exposed

o

not

exposed

Limitation

:

other

factors

can

affect

outcome

C ASE

=

DISEASE

!

RETROSPECTNE

compares

pts

w

disease

to

those

without

disease

Few

pts

=

Series

/

Single

Pt

=

case

report

series

more

reliable

than

report

Just

simply

No

101

R

Pe

treatment

is

lower

EX

:

A

B

OR

=

=So

23

%

increase

risk

of

falls

w

fracture

on

Serotenergic

A

Hazard

rate

in

treatment

group

EX

:

Niacin

Placebo

N

:

=

0

.

16

HR

=

I

Hazard

rate

in

control

group

N

=

18

N696p

=

=

0

.

1

No

benefit

when

adding

Niacin

-