Multi criteria decision analysis for healthcare

Multi-Criteria Decision Analysis for Healthcare

2014.03.12

Kwon, Sunhong

Value-Based Price From January 2014,

the Pharmaceutical Price Regulation Scheme (PPRS) will be replaced by ‘value-based pricing’ (VBP) for branded medicines sold to the NHS.

VBP is described as ‘‘a mechanism for ensuring patients can get access to the medicines they need by linking the prices the NHS pays drug providers to the value of the treatment.’’

The essence of VBP is that the maximum price the NHS will pay for a medicine is set where the incremental value of us-ing it relative to the comparator treatment (or palliation) just balances the incremental cost.

(Sussex, Towse et al. 2013)

VBP requires disparate types of ‘value’ to be aggregated.

The principal options are:– the net-benefit(NB) approach.

– multi-criteria decision analysis (MCDA) approach

– Weighted QALYs approach

(Sussex, Towse et al. 2013)

National HTA agencies have considered the possibil-ity to use MCDA to support HTA decision-making.– National Institute for Health and Clinical Excellence (NICE – UK),

the German Institute for Quality and Efficiency in Health Care (IQWiG – Germany) the Canadian Agency for Drugs and Technologies in Health

The need to introduce MCDA methods to researchers and decision makers in the healthcare field has be-come apparent.

For explicit decision making

Consistent

Transparent

Predictable

Simple

MCDA

Total Value

MCDA : Multi-Criteria Decision Analysis

the process of making a decision regarding a problem that requires the balancing of multiple factors that may conflict to a substantial extent

용어 Alternative

– a course of action assessed through a decision-making process, ‘option’

Criterion– a construct allowing for the assessment of the performance of an alternative,

‘attribute’, ‘objective’

Weight– a number expressing the relative importance of criteria/levels against which ‘alterna-

tives’ are compared.

Impact matrix– a table displaying the performance of each ‘alternative’ according to the ‘criteria’ and

measured on appropriate scales

Scale– an instrument on which the performance of an alternative is measured. Two types of

data can be measured on these scales, namely qualitative and quantitative.

MCDA 의 단계

Define Prob-lem Identify Criteria

Measure Crite-ria

Weight Criteria

Analysis(sensitivity analysis)

9

MCDA 의 예시

Total Value Impact matrix

Alternatives

Criteria Weight(preference)

A B CPrice 0.3 5 4 3

Design 0.2 2 3 5Color 0.4 1 5 4

Quality of material 0.1 6 5 3

Total Value 2.9 4.3 3.8

4*0.3+3*0.2+5*0.4+5*0.1= 4.3

Total Value Impact matrix

(Diaby and Goeree 2013)

보건 분야에서의 MCDA 현황

Publication pattern of MCDA applications in health over the years 1960–2011.

(Diaby, Campbell et al. 2013)

Number of publications by research topic on MCDA in health over the years 1960–2011.


Top ten countries of residence of corresponding au-thors of MCDA studies in health over the years 1960–2011.



Measure Crite-ria

Weight Criteria


Define Problem What Decisions Were MCDAs Informing?– The published MCDAs informed four decision types

• Investment—prioritization of interventions for coverage• Prescription—selection of intervention;• Authorization—assessment for licensing • Research funding—allocation of research funds.

(Marsh, Lanitis et al. 2014)


Measure Crite-ria

Weight Criteria


Identify Criteria What Criteria Were Included?– Greatest range of criteria was adopted in investment-fo-

cused studies.


Criteria included in multi-criteria decision analyses designed to support investment decisions

Identify Criteria– Criteria proposed for MCDA of orphan drugs

(Hughes-Wilson, Palma et al. 2012)

Identify Criteria– EVIDEM 에서의 Criteria

　 기본 판단기준 가능한 하위 기준 정의

질병영향

D1 질환 중증도 삶의 질 , 장애 , 기대수명에 대한 영향

중재로 인해 치료되거나 예방할 수 있는 건강상태의 중증도 예 ) 사망률 , 장애도 , 삶의 질 , 질병 진행 정도 .

D2 인구수 하위 그룹의 인구수 유병률 , 발생율

특정 인구 , 특정 시점에 중재로 인해 치료 , 예방되는 질병에 영향 받을 인구 예 ) 연간 발생률 , 시점 유병률

중재의 전후맥락

C1 임상 가이드라인 　제시된 중재 혹은 비슷한 대안에 대한 전문가 집단의 합의

C2 비교중재의 한계

효과에서의 불만족 안전성에서의 불만족 환자 보고 결과의 불만족

대상 질환의 치료 , 예방 , 개선 능력에 있어서 비교 대안의 단점 / 한계점

EVIDEM : 투명하고 효과적인 공공 보건분야의 의사결정을 위해 개발된 체계로서 기술 / 경제성 평가에 주로 활용되는 방법으로 MCDA value metrics 를 기반으로 함



중재의 결과물

I1 효과 / 효능의 개선

획득한 건강의 정도 예상되는 건강회복 인구 건강 회복 기간

대안 대비 이상적 , 더 유익한 증후 , 증상 , 질병 진행에서의 이상적인 변화 가능하면 효과 / 효능 개선

I2안 전 성 , 내 성 의

개선

단기안전성 , 장기안전성 중 증 부 작 용 의 위 험 ,

내약성

유해하거나 원치 않는 건강 위험 감소

I3 환자보고결과의 개선 삶의 질의 개선 , QALYs*

DALYs*, 편의성 , 자율성

환자 보고 결과의 유익한 변화예 ) 삶의 질 환자의 편의성 개선

유익의 종류

T1 공공건강의 관심인구 집단에서 위험 감소예 ) 질병전염 위험 감소 , 예방 , 발병 위험요 인의 위험감소

T2 의학 서비스의 종류 　환자 입장에서 임상적 유익성의 성질예 ) 증상완화 , 생명 연장 , 치료



경제성

E1 건강관리 예산 영향 중재 비용 , 획득비용 시행 , 유지 비용

중 재 가 목 표 한 건 강 을 위 해 예 산 에 미치는 궁극적 영향 , 기대되는 예산 절감

E2 비용 / 효용분석 QALY 당 비용 , LYG* 당

비용비용효용 비

E3 그 외의 재정 영향

1 차 치료 , 병원치료 , 장기치료 , 생산성 손실 , 환자 및 부양인의 재정영향

중재로 인한 기타 재정 영향 .예 ) 병 원 비 용 , 특 진 비 , 부 작 용 치료비용 . 장기치료 비용 , 장애비용 , 생산성손실 등


Measure Crite-ria

Weight Criteria


Measure Criteria How was intervention performance measured/scored?– Expert opinion, research evidence or both were used to

measure intervention performance


Several potential sources of bias in the evidence iden-tified by the studies. 1) Lack of resources required

2) Lack of comparability

Most studies used a scale to score intervention per-formance on criteria on equivalent scales.

Most studies used predefined scale . Some used the AHP or the natural units.


Measure Criteria

Measure Criteria EVIDEM 에서의 ScoringOverview중재군 : Tramadol 설정 : 캐나다 적응증 : 중증 , 중등도의 만성적 통증용법 : 1 일 1 회 1 정 ; 최대 8 정중재기간 : 　수일 ; 임상시험에서 4-12 주

질환 : Chronic non-cancer pain의약품 분류 : 비 마약성 진통제비 교 대 안 : 위 약 , NSAIDs, COX-2 억 제 제 , 마약성진통제

질환의 영향 중재군의 점수

D1 질환의 중증도 중증도 낮음

만성 비암성 통증 ; 37% 신경통으로 인한 요통 일상생활의 불편 : 28% 의 환자 수입의 손실 : 49% 의 환자 우울 , 불안 : 40% 의 환자

0 1 2 3중증도 높음

D2 인구수 매우 희귀한 질환

발 생 / 유 병 률 : 캐 나다 통 증 연 구 2004 ( 일 반 인 구 N=1055)

: 만성통증 25% (88% 중등도 , 중증 ) ; 평균기간 9.8 년0 1 2 3

매우 흔한 질환


Measure Crite-ria

Weight Criteria


– Numerous weighting techniques were em-ployed.

– The large majority of prescription-focused studies adopted AHP.

Weight Criteria


How Were Criteria Weights Generated?

Weight Criteria Criteria on defined scales was preferred technique for

pharmaceuticals (35 % of studies) and screening in-terventions

Public health studies used DCE (37.5 %), and all sur-gical intervention studies used AHP

Weights were elicited from many stakeholders



Measure Crite-ria

Weight Criteria


(Sussex, Rollet et al. 2013)

Sensitivity analysis How was Uncertainty Dealt With?– Almost half the studies did not report an assessment of un-

certainty

– Quarter of these incorporated uncertainty as a criterion.


How Useful is MCDA?

Positive aspect Providing a systematic approach to decision making Facilitating knowledge transfer Improving decision makers’ understanding of Identifying data gaps Forcing decision makers to think through all relevant

factors improving the transparency of decisions Communicating their rationale


How Useful is MCDA?

Concerns Time and resources required to implement the MCDA Challenge of absorbing a lot of information Need to participate in the process several times before

participants were familiar with the process Challenges interpreting the MCDA outputs Generalizability of the results beyond the time and

place of the specific decision being informed


THANK YOU

Back up

Where Were MCDAs Undertaken? Various location– The 40 studies selected in review paper were under-taken

in 18 locations

– The most prevalent being: The Netherlands (7), USA (7), and the UK (4).

What Impact Do MCDA Methods have?

Two studies assessed the impact of MCDA methods.

1. Compared supporting patients’ decisions with MCDA rather than with a simple educational inter-vention, -MCDA improved patients’ decisions about colorectal cancer screening.

2. Assessed whether different MCDA methods pro-duced different results. Simple additive weighting (SAW) and technique for order preference by simi-larity to an ideal solution (TOPSIS) were found to produce the same conclusions.

WEIGTH ELICITATION METHODS Method 1: Five point weighting scale (nonhierar-

chical structure)

EVIDEM Collaboration

WEIGTH ELICITATION METHODS Method 2: Five point weighting scale

(hierarchical structure)


WEIGTH ELICITATION METHODS Method 3: Point allocation (hierarchical structure)


WEIGTH ELICITATION METHODS Method 4: Ranking (hierarchical structure)


WEIGTH ELICITATION METHODS Method 5: Pairwise comparison

(hierarchical structure)


WEIGTH ELICITATION METHODS Method 6: Best-worst scaling


WEIGTH ELICITATION METHODS Method 7: DCE ( 이산선택법 )