pa presentation cadth april 2016

18
A FRAMEWORK FOR COST- EFFECTIVENESS ANALYSIS OF PERSONALIZED MEDICINE CO- DEPENDENT TECHNOLOGIES CADTH April 12, 2016 Philip Akude – MSc, Reza Mahjoub – PhD, Mike Paulden – MSc, Christopher McCabe – PhD University of Alberta

Upload: cadth-acmts

Post on 12-Apr-2017

73 views

Category:

Healthcare


1 download

TRANSCRIPT

Page 1: Pa presentation cadth april 2016

A FRAMEWORK FOR COST-EFFECTIVENESS ANALYSIS OF PERSONALIZED MEDICINE CO-DEPENDENT TECHNOLOGIES

CADTHApril 12, 2016

Philip Akude – MSc, Reza Mahjoub – PhD, Mike Paulden – MSc, Christopher McCabe – PhD

University of Alberta

Page 2: Pa presentation cadth april 2016

DISCLOSURE This study was conducted under the PACEOMICS project, funded by

Genome Canada, Genome Quebec, Genome Alberta and the Canadian Institutes for Health Research (CIHR).

The following authors are funded by PACEOMICS project: Philip Akude, Reza Mahjoub, and Michael Paulden.

Christopher McCabe is funded by the University of Alberta, Faculty of Medicine and Dentistry.

Page 3: Pa presentation cadth april 2016

OUTLINEBackgroundObjectiveGeneral model descriptionSimplified model with probability of response set exogenouslyOptimal cut-offs for the general model under perfect informationConclusions

Page 4: Pa presentation cadth april 2016

CURRENT HTA PRACTICEHTA for treatment technologies are increasingly required to have randomized controlled trial evidence of efficacy. Test technologies are frequently adopted on the basis of evidence of laboratory validity and clinical test performance. The ascent of personalised medicine, specifically test guided therapies is bringing these two evidentiary traditions together.Stakeholders of personalized medicine product seek a coherent framework to appraise these technologies.

Page 5: Pa presentation cadth april 2016

OBJECTIVE Develop methods for combining evidence on the test(s) and treatment

components of co-dependent technologies, and to identify the cost effective cut offs on the test components for pre-specified values of the willingness to pay for health.

Page 6: Pa presentation cadth april 2016

CO-DEPENDENT TECHNOLOGIES FLOWCHART

Genotypic Test

(Test d)Positive? Yes

No

No Treatme

nt

Therapy Responde

r Test(Test π)

TP? Yes

FP

No Treatme

nt

Test π

Π≥ΠCYes

No

Stand. Care

Phenotypic Test

(Test u)UR≥UC

Yes

No

Stand. Care

New Tx

TN

Page 7: Pa presentation cadth april 2016

MODEL PARAMETERS: Health benefit (phenotype) resulting from the new treatment or standard care, 1,

{ , , Responding, Non-responding, Standard Care}

: Expected health benefit, { , , }: Cost of treatme

i i

i

i

U U

i R N S R N S

U i R N SC

nt/standard care per unit time, { , , }

: Expected cost, { , , }: Cost of test, { , , }

: Error in test measurment; ~ (0, ), { , };ˆ : Observed health benefit(phenotypic expression) f

i

tj

j j j

R

i R N S

C i R N SC j d u

N j u

U

ˆor a responding patient; ˆ ˆ: Estimated probability of response,

: Inverse of CE ratio

R R uU U

g

Page 8: Pa presentation cadth april 2016

INCORPORATING HETEROGENEITY1. The patient population heterogeneous with

respect to the “success rate”, i.e., Π:Fπ(π)=Pr{Π≤π} is CDF of Π and fπ(π) is PDF of Π

2. The patient population heterogeneous with respect to their phenotypic expression for patients who respond to treatment, i.e., UR:

Fu(uR)=Pr{UR ≤ uR} is CDF of UR and fu(uR) is PDF of UR

Page 9: Pa presentation cadth april 2016

DECISION TREE FOR CO-DEPENDENT TECHNOLOGIES

( )R R td t tuU g C C C C

( )S S td t tuU g C C C C

( )R R td t tuU g C C C C

( )S S td t tuU g C C C C

1

ˆC

1 c

C

ˆC

1R R CuU U U

   R C uU U

New Tx

Stand. Care

ˆR CU U

ˆR CU U

1R R CuU U U

 R C uU U

New Tx

Stand. Care

ˆR CU U

ˆR CU U

Stand. Care

p

1 p

f

1 f

q

1 q

( )SN SN tdU g C C

( )HN HN td tU g C C C

( )HN HN tdU g C C

Test d

Sick

Healthy

TP

Test

TPE NB

FN

FP

TN

Test 2

Do Not Treat

Do Not Treat

Responding

Not Responding

R CU U

R CU U

( )S S td tU g C C C

( )R R td t tuU g C C C C

( )S S td t tuU g C C C C

( )R R td t tuU g C C C C

( )S S td t tuU g C C C C

ˆC

1 c

C

ˆC

1R R CuU U U

   R C uU U

New Tx

Stand. Care

ˆR CU U

ˆR CU U

1R R CuU U U

   R C uU U

New Tx

Stand. Care

ˆR CU U

ˆR CU U

Stand. Care

R CU U

R CU U

( )S S td tU g C C C

C

ˆ

ˆR R uU U

ˆR R uU U

ˆR R uU U

ˆR R uU U

Phelps & Mushlin Framework

Same tree as on responding

† FP patients will be correctly diagnosed as TN as a result of second test

Test u

Page 10: Pa presentation cadth april 2016

( )R R td tuU g C C C

( )R R td tuU g C C C

( )S S td tuU g C C C

( )S S td tuU g C C C

1R u R CU U U

1R u R CU U U

 R C uU U

 R C uU U

R CU U

R CU U

ˆR R uU U

ˆR R uU U

ˆR CU U

ˆR CU U

New Tx

Stand. Care

ˆR CU U

ˆR CU U

New Tx

Stand. CareTPE NB

( )N N td tuU g C C C

( )N N td tuU g C C C

(   )S S td tuU g C C C

( )S S td tuU g C C C

1R u R CU U U

1R u R CU U U

   R C uU U

 R C uU U

R CU U

R CU U

ˆR R uU U

ˆR R uU U

ˆR CU U

ˆR CU U

New Tx

Stand. Care

ˆR CU U

ˆR CU U

New Tx

Stand. Care

1

p

1 p

f

1 f

q

1 q

Test 1

Sick

Healthy

TP

FN

FP

TN

Test u

Do Not Treat

Do Not Treat

Phelps & Mushlin Framework

( )SN SN tdU g C C

( )HN HN tdU g C C

• Π exogenous• Imperfect information (error in measurement)• One test for magnitude of response UR

Test u

WHEN THE PROBABILITY OF RESPONSE IS EXOGENOUS

Page 11: Pa presentation cadth april 2016

OPTIMALITY CONDITION FOR UC

1

0FOC: 0,

where .

TP Ru RR

uuR R RCuC

R N N N NS SR

NBE K f u U f u duU

K u U C Cg U U g C C

1

1

1

1

1

)( uR R Cu RuR R

uR R Cu RuR R

uRu RuR R C

u u Uu u uTP R N R N R Ru u

u u Uu u uN N R Ru u

uu u u R RS Su u U

t

E NB u U g C C f d f u du

U gC f d f u du

U gC f d f u du

g C

.tud C

Page 12: Pa presentation cadth april 2016

WORKED EXAMPLE

Us = 0.69UN = 0.35CN = 1,245$ CR = 1,245$ CS = 15,958$ π = 0.6

~ (2.7,0.3) Mean=0.9 and Stand. Deviation=0.15

~ (0,0.1)

R

u

U

N

Example:

*Note that under perfect information, 0.51.CU

* 0.461CU

Page 13: Pa presentation cadth april 2016

( )R R td t tuU g C C C C

( )S S td t tuU g C C C C

C

1

New Tx

Stand. Care

p

1 p

f

1 f

q

1 q

( )SN SN tdU g C C

( )HN HN td tU g C C C

( )HN HN tdU g C C

Test 1

Sick

Healthy

TP

Test

TPENB

FN

FP

TN

Test 2

Do Not Treat

Do Not Treat

Not Responding

R CU U

R CU U

Stand. Care ( )S S d tU g C C C C

Phelps & Mushlin Framework

BOTH Π AND UC ARE ENDOGENOUSLY DETERMINED

† FP patients will be correctly diagnosed as TN as a result of second test

( )N N td t tuU g C C C C

( )S S td t tuU g C C C C

C

New Tx

Stand. Care

R CU U

R CU U

Stand. Care ( )S S td tU g C C C C

Test u

• No error in measurement

Page 14: Pa presentation cadth april 2016

E[NBTP] FOR USE OF TEST

1 1

0

1 1

( ) ( ) ( ) ( ) ( )

( ) ( )

1 ( ) ( ) ( ) ( )

C

R RC C

C

C

R RC C

U

TP R R td t tu u R R S S td t tu u R RU

S S td t

U

N N td t tu u R R S S td t tu u R RU

E NB u g C C C C f u du U g C C C C f u du f d

U g C C C f d

U g C C C C f u du U g C C C C f u du

0

( )

1 ( ) ( )C

S S td t

f d

U g C C C f d

1 1

1

0

( ) 1 ( ) ( ) ( )

( ) ( ) ( ) ( ) ( )

RC C

C C

RC

TP R R td t tu N N td t tu u R RU

U

S S td t S S td t tu u R R

E NB u g C C C C U g C C C C f u f du d

U g C C C f d U g C C C C f u du f d

Page 15: Pa presentation cadth april 2016

OPTIMAL CUT OFF CALCULATIONS - UC

1( ) 1 ( ) ( )

( ) ( ) ( )

RC

C

R

TP R R td t tu N N td t tu u R RU

U

S S td t S S td t tu u R R

E NB u g C C C C U g C C C C f u du

U g C C C U g C C C C f u du

UC* is equal to the clinical expression for a responding patient, at which the payer becomes indifferent between the new treatment and the standard care.

* 1( ) ( )C N R N S N S NU U g C C U U g C C

*FOC to find the optimal phenotypic cut off, :

0

C

TP

C

U

E NB

U

1R R N N S Su gC U gC U gC

Net benefits from new TxNet benefits from Stand. Care

Page 16: Pa presentation cadth april 2016

OPTIMAL CUT OFF CALCULATIONS - ΠC

*

*

To find the optimal cut-off probability of response :

: 0

At :

C

TP

C

C C

E NBFOC

* * * *

1 1* *

( ) ( )1 ) ( ) ( )

R RC C C C

C R R C N N u R R tu S S u R RU Uu gC U gC f u du gC U gC f u du

Page 17: Pa presentation cadth april 2016

CONCLUSIONSThis study develops a formal decision analytic framework for the economic evaluation of personalised medicine co-dependent technologies.The method presented offers decision makers the impact a changing cost effectiveness threshold has on the optimal cut-off value for co-dependent technologies.The optimal test cut-off must be set at the point where the marginal payoff from new treatment is equal to the marginal payoff from standard care.

Page 18: Pa presentation cadth april 2016

THANK YOU

[email protected]