2. Clinicians have considerably more
expertise and knowledge than “the
commissioner” in their area of
expertise
but a completely different world view
3. Nobody knows the whole
commissioning and provider archipelago
weak levers
mixed and perverse incentives
focus on service users not populations
4. Starting questions
What is the “framework”
• What does NICE say
• RCP guidance and recommendations
• National or local audits
• Are there gaps.
• If you want to fill gaps, there is no money so give
something up.
• All of this needs to be turned into “the system”
and then into contracts, specifications clinical
pathways and clinical behaviour.
6. I would like to impress on you the IMPRESS
framework– relative value of COPD interventions
• LSE / BTS collaboration.
• Ranked set of interventions to
improve population outcomes in
COPD
• By number exposed, benefit per
person and cost – thus value.
• In effect “the commissioning plan”
http://www.impressresp.com/index.php?option=com_content&view=article&id=131&Itemid=17
http://www.bmj.com/content/345/bmj.e6192
7. Value in COPD – (vs where we
spend the £££)
Thorax Nov 2014
8. However
• How much cold hard cash is ACTUALLY in triple tx (quite
a lot)
• GMS contract and the path of least resistance for the GP
– triple therapy. Smoking cessation
• Number of people, value per person etc - SCALE is
important
• Block contracts with big providers – sweating those.
• Are important interventions under provided? – Pulmonary
Rehab, and is there a specification / pathway
• Are we talking about changing NOW stuff or changing into
the future.
9. Stroke – a slightly different spin
Burden of disease versus
“improving pathways”
10. Stroke is an important problem
• Common
• Deadly and disabling – death and misery
• c100,000 incident events in England
– 20% AF, 25% -30% Diabetic
• £12k and £6k – 1 and subsequent yr cost.
12. Oh and we know the risk factors - we consistently
and continually fail to prevent – 1/3 of DM pop with
BP out of ideal.
13. AF Anticoag - Dear NHS….. Must do
better
Marked under use of a cheap and
effective intervention that cuts stroke risk
by c60%
This is not news.
“overuse” of anti platelet medicine
14. STAR in 3 slides
Describe and line up all steps in a
pathway of care
4 steps
1. No of cases treated
2. Cost per case – thus total population cost
3. Average quality of life gain in each case
treated.
4. Order the pathway
15.
16. Total Value
= 7.36
QALY gain
/ Case = 0.46
Numbers treated = 16
Total
Costs = £2m Costs / case = £120k
Numbers treated = 16
Visualmodel:
intensive residential care
17. Efficiency frontier over care
pathway
QALYs
Costs
Worst VfM
Scale matters
But relatively
small £
So worth the
bother?
18. VfM triangles pathway for treatment of
eating disorders
Residential intensive care
80% costs & 13% benefits
Value
Costs
?
Worst VfM
Lower budget
More QALYs
19. Tested a range of scenarios to increase
value
Prof Bevan LSE
21. Pathways v populations. We all
want good stroke care for our gran.
• Morris Stroke Reconfiguration Study –
centralisation and HASU (n=2 study)
• centralising care to eight (rather than 30) sites
will save lives. £23m in London
• NNT = 100
• ? Evidence around impact of alteplase. FDA
and EMA? Actual impact in real world - % that
get in window?
• At expense of population perspective –
Burden of disease - Salt / exercise / weight /
alcohol – BP prevention
https://gregfellpublichealth.wordpress.com/2016/07/09/stroke-how-many-lives-will-
be-saved-by-the-centralisation-of-hyper-acute-stroke-units-how-much-money-will-
this-save/
22. Lung Cancer
• Burden of disease vs diagnosis treating
• Is your “outcome” incident rate, survival, Lung
cancer (or all cause) mortality, optimisation of
resection
– Smoking
– Screening???
– Early diagnosis???
– Surgery
– RT
– Chemo
– Palliative and end of life care
• What is the balance of your interventions
23. Prof Wight / STAR team. Sheffield
Compare where we
spend most money,
to what is most
‘valuable’
What consumes
most £ for about 5%
of the gain
Does it look like
diminishing marginal
returns curve to you?
24. So where do we focus our energy
and effort?
• It depends on what your objective is
• Reduced burden of disease in patients
WITH condition of interest
• Reduced burden of disease in population
at risk and patient with condition
• Different stakeholders will have different
interests. This RADICALLY affects your
INTERVENTION.
25. “But its not fair to measure whole
pop outcomes – only in the treated
pop”
• It should be only impact in those treated
with that thing
• Yes sure
• But it's the pop more broadly that bear the
opportunity cost
• is that fair?
• Which is least unfair?
26. If policy time, resources, cash, brain
power is limited….
• Marginal improvements ± great cost to stroke
care, stroke thrombolysis
• Primary prevention – salt, AF Anticoagulation,
pop HTN management
• Stroke rehab
• LSE analysis tells us pretty clearly that
primary prevention is best strategy.
Yet to find a context in which upstream primary
preventive strategy doesn’t trump downstream
(by orders of magnitude)
28. Exercise Diabetes
• Common
• Fast growing
• Expensive
• Plot out 5 options for improving value in
this area of spend
• What do you focus on
• What information do you need?
• Where will you look for information.
30. Some thoughts on scoping
• Dm pathway vs Burden of disease
• Screening
• Prevention
• Pre diabetes – in our out of scope. What do you do?
• Glycaemia management / totality of DM care
• Medicines – Glycaemia, BP, Lipids
• Insuling pump – in or out of scope?
• Complications – in or out of scope CV / Opth / Renal /
vascular
• Time horizon
31. Prevention & the future - The epi
isn't pretty. The evidence is
amassed
32. Will intensive lifestyle intervention
REALLY work in practice?
PLOS ONE 8(7): e68605.
doi:10.1371/journal.pone.006860
33. Individual level v population level
See Yudkin and Tomlinson rapid
responses
http://www.bmj.com/content/351/bmj.h4717
35. Early diagnosis will save money
NO IT WONT!
Lancet 2011; 378: 156–67
https://gregfellpublichealth.wordpress.com/2016/02/28/why-
i-argued-against-diabetes-screening/
37. Oh and don’t forget…….
• Screening will find lots of elderly people
with diabetes
• We KNOW we over treat the elderly in our
efforts to achieve tight glycaemic targets
• We KNOW this is limited to no benefit –
individual level and pop.
• And there are risks (hypo)
• And costs (fastest growing area of
prescribing bill)
38. • Although the harms of intensive treatment likely exceed the benefits for older patients with
complex/intermediate or very complex/poor health status, most of these adults reached tight glycemic
targets between 2001 and 2010. Most of them were treated with insulin or sulfonylureas, which may lead
to severe hypoglycemia. Our findings suggest that a substantial proportion of older adults with diabetes
were potentially overtreated.
40. Tight glycaemic control will prevent
complications and save money
5yr NNT – tight A1C vs relaxed.
272 blindness,
120 CVD,
627 renal fail
41. Population cost of preventing complications with
intensive (compared to less intensive) glycaemic
control
can be estimated by combining drug costs & NNT
So assuming an NNT of 272 for blindness
You need to treat 272 pt to tight target (compared to less tight control) for 5yrs to
prevent 1 person going blind
And 5yr cost of £1000 on insulin (£2k for glitazone)
It costs £272k to prevent that CV event, compared to £544k with glitazone.
Costs can be altered in
this model
42. No difference between different types
of 3rd line agent – cost minimisation.
Use the cheapest.
Meta analysis. 19 Articles, c 5k
patients
Cross et al
Ann Int Med 2011
44. "How many people in this room would take a
treatment for years if they didn't know
whether it would do them more good than
harm?"
45. Bad medicine the way we manage
diabetes
BMJ 2013; 346 doi:
http://dx.doi.org/10.1136/bmj.f269
5
46. https://www.youtube.com/watch?v=jOxxHbdyXcg
start at about 23 minutes - lack of evidence that medicines lower the risks (overestimated as
ever) of the disease.
Intensive A1c in elderly – QOF forces us
down this path. We harm people for very
marginal gain.
48. Prev study – out to 5.6 yrs. No diff in CV events
between intensive / standard
N=1719
Long term follow up. 92% of the original cohort.
Primary outcome is CV event
Median A1C at end of trial - 6.9% vs. 8.4%, at 3
yrs after trial difference had declined to 0.2 to
0.3 % points
At median follow-up of 9.8 years HR for CV Event
was 0.83 (95% CI 0.70 to 0.99; P=0.04),
Absolute risk absolute reduction in risk of 8.6
events per 1000 person-years – or .86% (NNT =
115 or so)
No sig diff in mortality. (hazard ratio, 0.88; 95% CI,
0.64 to 1.20; P=0.42).
49. But we consistently and continually fail to
prevent – 1/3 of DM pop with BP out of ideal.
51. To improve value in your system
you need to change…..???
• Change
• no of incident cases
• Outcome – clinical outcome in cases
• Cost per case?
• volume in pop thus pop cost
• You need to change either of these
parameters to improve value.
52. Some questions to ask to get
better value
• variation between geographies
• Do you have clear understanding of the dynamics of
different services in a clinical area.
• “you spend twice as much on teeth as you do eyes,
eyes are more important than teeth – we want some of
their resources”.
• Multi year analysis - what is growing and what is
stable – and thus help identify targets for detailed
consideration.
• What areas of spend within this system or programme
are growing rapidly
• What are the top five areas of spend, or 10.
• Where are we spending more or less than
comparable populations. Where are outcomes better
or worse than comparable populations.
• What is the extent that we can link sets of information
together and focus on whole systems rather than focus
just on admissions or drugs.
53. Key points - systems
• Burden of disease or population
• Under implemented high value
• Over implemented low value
• Know your system, where you spend the
money, who spends it, how
• Population management of risk
• Prevention – where does it fit. How
aggressively do you implement?
54. 3 Whole systems
the things we spend out money on in
the name of "integration" and "reducing
unnecessary admissions“ and “saving
cash”
55. But….back in the real world….
• Programme budgets don’t look like way in which
resources are allocated
• Do commissioners actually allocate resources
• Margins or the totality.
• Whose perspective – that of service provider or
commissioner. Service users v population
• Levers the commissioner has are not strong.
• Vast majority of resource is committed by rote of
what happened last year, existing patterns of
service delivery and behaviours of clinicians
• who may / may not have a stake in financial health
of their institution. If they do it's most likely to be an
institution perspective as opposed to a population
perspective
56. Commissioning is a 20 year failed
experiment in efficiency or at best a
missed opportunity
• Discuss…….
• Marginal investments, often of limited
value
• High transaction costs on both sides –
payer and provider
• Not achieved the integration that was
intended (yet)
• Save between 10-25% on transaction
costs
57. Some contentious statements
• We don't know where we spend the money
• We don't know whether we are spending the
money on the right things
• We don't know whether we get the best
outcomes
• We don’t know whether we get good value
• QIPP plans are highly uncertain, some would
say monopoly money. They are focused on cash
not value.
58. “integration”
• Great deal of policy time & attention given over
to “integrating things”.
• Better Care Fund £ assumptions are stupid
• limited to no evidence to support financial
improvement. NB - quality and user experience
• we often focus on “high risk” patients but
forget the lessons of epidemiology (Roland)
• We focus our energy on things that
research tells us DONT work
• under implement things of highly proven
value and close to cost saving in the short term
Refs in notes page
59. Overestimate
importance of freq
flyers
impactable?
churn?
If you want to make
difference at scale –
focus on big numbers
We ignore regression to
mean
Supply induced
demand
Variation due to
chance
unevaluated
interventions
interventions we know
don't work
60. Prof Roland’s advice
1. Don’t assume that reductions in admissions in a
high risk group are due to your intervention
2. Don’t assume there is a correct level of
admission or referral
3. Don’t assume that fewer admissions or referrals
are necessarily better
4. Be cautious about using data for short time
periods or referrals to single specialties
5. Choose interventions that are evidence based
61. Procedures of “limited clinical value”
• Don’t get me started
https://gregfellpublichealth.wordpress.com/2018/01/02/procedures-of-limited-
clinical-value/
https://gregfellpublichealth.wordpress.com/2016/02/20/how-to-implement-
procedures-of-limited-clinical-value/
62. We aren't short on advice where to
look to increase technical efficiency
– eliminating low value stuff
http://archinte.jamanetwork.com/arti
cle.aspx?articleid=2469079
63. Of course we should be choosing
wisely……
https://www.sciencebasedmedicine.org/choosing-wisely-changing-
medical-practice-is-hard/
64. But - we ignore culture and getting
clinical / £ incentives aligned at our
peril
• Wilson
– Patient
– Clinician
– Organisational cultural issues
– Mixed incentives – incentives to improve
quality, outcome and achieve population level
financial savings are mixed and messy
http://qualitysafety.bmj.com/content/early/2015/07/07/bmjqs-2015-004518.short
65. Referral management
• Don’t get me started
British Journal of General Practice, Volume 63, Number 611, June 2013 , pp. e386-e392(7)
66. Self care will save stacks of cash
• Really?
• I invite you to test,
• ….really test, the ROI 0f £3 to £1 for
Expert Patient Programme
http://www.population-
health.manchester.ac.uk/primarycare/npcrdc-
archive/Publications/EPP%20FINAL%20REPORT%202007.p
df
http://jech.bmj.com/content/62/4/361.full
69. Hospital at home – rehab – step down
Sort of like VW?
Reduce bed days?
Depends on broader system?
We don’t know.
Overall no of non electives will go up
– pop growth
Rate v number
70. H at H
Admission avoidance –
step up
No different to IP care
Same quality / cheaper
setting?
71. Turn it on its head
Same quality of care but
cheaper setting
But DON’T forget Roemers
law!
Also “step up” vs “step
down” – money, tariff,
income.
72. Personal care planning – Cochrane
Review
http://onlinelibrary.wiley.com/doi/10.
1002/14651858.CD010523.pub2/abs
tract
Evid Based Med October 2015 |
volume 20 | number 5 |
73. Case management for frail elders.
Sandberg et al. Health Economics
Review (2015) 5:12
DOI 10.1186/s13561-015-0051-9
79. CLINICAL impact of integration
interventions to reduce non elective
admit
intervention impact on em admission group affected
case management reduce heart failure and frail old
care coordination (HSCI) reduce frail old
specialist clinics reduce heart failure
education and self care reduce adult asthma and copd
telemedicine reduce heart dis, dm, htn, elderly
telecare no impact copd, dm, heart failure
virtual wards no impact high risk
vaccine programmes no impact asthma, copd, elderly
medication reviews no impact elderly, heart failure, asthma
falls prevention no impact [manchester] frail old
hospital at home increases comorbid old
Taken from NIHR 2015
(wilson)
NIHR 2014 (Imoson)
Bristol 2013 (Purdy)
Andy Snell
80. Models of care for high cost
patients
http://www.commonwealthfund.org/publications/issue-
briefs/2015/oct/care-high-need-high-cost-patients
“Overall, the evidence of impact
is modest and few of these
models have been widely adopted
in practice”
81. 5 things to say on integration
• Most integration programme are about reducing non el admit
+ bed days
• Most often by "improving community services"
1) unclear PICO
2) often intervention is very vaguely defined
3) evidence of many interventions being effective is very
poor. ditto cost eff
4) the notion that transferring care to community is cheaper
or more efficient is questionable
5) arguably focus on high risk and non el gives incentive
for provider to transfer poorly covered financial risk to
another sector - likely to lead to less efficient system
overall
• And also arguably crowds out higher value preventive
investments
82. Im not saying some of the
above are NOT good ideas
But from a value for money point
of view – we JUST DON’T KNOW!
84. Why it matters – back to The Don –
triple aim
• don't do harm - primary individual level responsibility
• population health - don't get sick in the first place (much sickness
is not to do with what happens in the HC system)
• nutrition, poverty, violence, environment. mostly out of control
• stewardship of collective limited resource - when we are sick do
the most good (population level) with the resources that are
available
– health care is not ENTITLED to the share of income it gets - there is a
collective responsibility to do best with it.
– constant confiscation of social resources by health care
– if stewards dont take responsibility for common pooled resource - others
loose - social level. Crowds out more valuable social goods
– per capita cost is thus the third aim. just as important as the no harm
– thus per capita cost is the metric
http://www.kingsfund.org.uk/audio-
video/don-berwick-implementing-new-
models-care
85. But………
the whole problem in evidence based policy
1. People believe what suits them
2. we forget the implementation cost and
incremental cost and benefit at our peril
3. vested interests of powerful stakeholders
86. so
• Be savvy with your economics
• Learn it
• Be prepared to streetfight with it
• Try to do the right thing
• Sometimes this is hard