Measuring decision quality: imperatives and tensions

1

• Measuring decision quality imperatives and
  tensions
• June 2009

Richard Thomson Professor of Epidemiology and
Public Health Risk Communication and Decision
Making Research Programme Institute of Health and
Society Newcastle upon Tyne Medical School
2
Thesis

• A good decision should be measurable at
• Individual level
• Provider level (clinician, specialty, health care
  organisation)
• Population level
• Measures should be valid and reliable

3
What is a good decision?
4
Good decision?

• Based on relevant knowledge, consistent with the
  decision-makers values and behaviourally
  implemented (Marteau et al 2001)
• The extent to which a decision reflects the
  considered preference of a well-informed patient,
  and is implemented (Sepucha et al 2007)

5
What are the elements of a good decision?
6
Donabedian triad
Structure
Process
Outcome
7
Structure
8
Process
9
Health outcome and decision quality
10
Outcome
11
Wider issues Differing decision needs and
contexts
12
Differing needs and contexts

• Desire for information and support varies between
  patients and decisions
• Amount
• Format written, internet
• Interaction - face to face, remote
• Timing before, during or after consultation
• Hyper-acute stroke and thrombolysis
• Trust and confidence important
• Face to face verbal
• Did not want any written material
• Clinicians wanted support for eligibility
  assessment and personalised risk communication
• Decision guide/coach

13
Amount of information

• Miller Behavioural Style Scale (Miller/Mangan
  1983)
• Monitors and blunters
• Randomised women (monitors and blunters) about to
  undergo colposcopy to voluminous information
  about procedure and effects or minimal
  information
• Level of arousal lowest when given information
  matching their preference i.e. blunters less
  aroused with low information and monitors less
  aroused with more information.
• Implication high quality SDM requires
  appropriate information and engagement

14
Evaluation context a cautious tale

• Participants attributions about their identity
  fell on a continuum. At one end they identified
  themselves as experienced medical volunteers,
  and at the other as real patients.
• Most people fell closer to an identity of
  volunteer than patient
• Bob When I went I didn't think there was any
  possibility, I didn't know why I was going, just
  to help with some sort of research, I didn't know
  there was any possibility that they might advise
  us to come off treatment A. Em, I didnt think
  I was going to er, I thought I was going, just an
  exercise in finding out what's going on with
  people but I didn't realise there was any
  possibility they'd advise us to come off
  treatment A, and they haven't actually.
• Belle Yes. Well er, actually er, that's what
  I've been wondering about because you hear people
  saying different things, you know, and I know
  I've been on that a long time Yeah and er,
  and I was just wondering IF em, if I did come off
  treatment A and tried treatment B whether it
  would be better or or not or not Right.

Heaven, B., Murtagh, M., Rapley, R., May, C.,
Graham, R., Kaner, E., Thomson, R. Patients or
research subjects? A qualitative study of
participation in a randomised controlled trial of
a complex intervention. Patient Education
Counseling. 62 (2006) 260270
15
Population/individual tensions
16
AF and stroke prevention PDA

• RCT interactive computerised PDA in SDM
  consultation cf. with paternalistically applied
  evidence-based guidelines (control)
• Significantly fewer in the PDA group chose
  warfarin
• Of those not already on warfarin, 25 chose
  warfarin in PDA group cf 94 in control.
• Patients made an informed choice that might be
  seen as counter to evidence-based guidance
• This begs questions
• How would high quality shared decisions impact
  upon population health outcomes and
  government/policy targets and imperatives?
• What population health outcomes?

17
Why might different decisions be made?

• MRC trial of mild hypertension - treat 823
  patients for one year (all exposed to drug
  treatment and consequences) to prevent one stroke
  
• Goals of EBGs/health service health improvement
  policy and patient/public engagement may differ
• Differences in thresholds cf patient preference
  to EBG
• Of 38 patients whose decision analysis indicated
  preference for not taking warfarin, 87 and 58
  would have been advised to take warfarin
  respectively by age related and absolute risk
  guidelines (Protheroe et al BMJ 20003201380-4)
• Differences in treatment thresholds cf
  professionals to patients
• Canadian study patients significantly less likely
  to want treatment than clinicians 49 v 64, 68
  v 92, and 86 v 100 at 2, 5 and 10 five year
  cardiovascular risk (McAlister et al CMAJ
  2000163403-8)
• UK study threshold NNT for willingness to accept
  treatment for saving one life over five years was
  100 (Physicians), 50 (GPs), 33 (practice nurses
  and public) (Steel BMJ 20003201446-7)

18
Population impact of SDM policy tensions -
hypertension

• If only 50 (or 25) of all newly identified
  hypertensive patients over 65 years in a
  population of 100,000 agreed to treatment as a
  result of shared decision making, then 122 (or
  183) strokes would occur that would otherwise
  have been prevented.

19
Implications of hypertension decisions

• Problems
• fewer patients taking effective therapy
• increased incidence of stroke
• failure of health services to achieve targets for
  blood pressure control and disease reduction
• But patients would have experienced
• greater involvement in decision making
• less decisional conflict
• better match with values
• greater satisfaction with their decision
• in those who decide to take treatment,
  potentially better levels of concordance and
  blood pressure control
• What measures of health improvement/outcome
  should we use?

20
WHO definition
Health is a state of complete physical, mental
and social well-being and not merely the absence
of disease or infirmity.
21
Evaluation at system, provider, organisation
level
22
Were you involved as much as you wanted to be in
decisions about your care and treatment?NHS
Inpatient Survey 2005-8
23
Would have liked more involvement in treatment
decisions NHS patient surveys, England 2004/8
Thanks to Angela Coulter
24
UK NHS Survey and feedback
25
Dartmouth Hitchcock Medical Center Breast Cancer
service
26
Implications for policy and research
27
Policy

• Embedding measurement and monitoring of decision
  quality in practice
• Comparative quality measures and feedback
• Shifting choice agenda
• Good decisions require good decision support
• Recognition of tensions between evidence based
  guidance and SDM
• Agenda setting. What is important to policy
  makers?

28
Research

• Development of quality measures, including
  appropriate health outcomes
• Validity and reliability of quality measures
• Correlation of measures
• Matching patient preferences to appropriate
  support and impact of doing (or not doing) so
• Process or outcome
• Timing of assessment
• How best to support good decisions context
  specific and organisation wide. Implementation in
  practice
• Cost effectiveness
• Population impact of SDM

29
Summary

• Good shared decisions have common components
  informed (knowledge) values elicited consistent
  with values
• Many potential measures some primary, others
  secondary but important
• Outcome measures are important
• A good quality decision requires good quality and
  personalised/flexible decision support
• Research needed but enough already to drive
  policy
• Embed research in practice.

30
Thank yourichard.thomson_at_newcastle.ac.uk