Written evidence from Professor Gwyn Bevan[53]
(COM 131)
EXECUTIVE SUMMARY
(i) This note aims to give evidence relevant
to following issues in the terms of reference of the Health Committee's
inquiry into Commissioning:
(a) Clinical involvement, clinical practice variation,
risk management and budgeting;
(b) Integration of primary and secondary health
care and designing a market for new entrants into commissioning;
and
(c) Engaging GPs, patients and the public in
making hard choices in commissioning.
(ii) Clinical involvement, clinical
practice variation, risk management and budgeting. GP consortia
are likely find that GPs have interests in being involved in commissioning
different services and it is difficult to involve GPs in budgeting.
Allocating resources to practices with reference to targets derived
from a formula using estimates of the relative needs of populations
(and unavoidable variations in costs) will show large variations
for some practices between their estimated past expenditure and
their estimated future target allocations. Information on Clinical
Practice Variation (CPV) can help make sense of these differences.
(iii) Integration of primary and secondary
health care and designing a market for new entrants into commissioning.
The division in British medicine between hospital specialists
and General Practitioners (GPs) and between commissioners and
providers makes it difficult to create high-performing Integrated
Insurance and Delivery Systems (IIDSs) as in the US, such as the
Group Health Cooperative of Puget Sound and the Kaiser Permanente
Health Maintenance Organisation. Evidence suggests that these
IIDSs are more cost effective than the NHS. But US experience
shows that it has proved difficult to replicate this model in
the US. It would also be difficult to satisfy the characteristics
that result in high-performing IIDSs in the English NHS. Allowing
GP consortia to form integrated organisations with hospitals has
potential to create much better care than now, but could also
result in some dysfunctional organisations (as happened in the
"backlash" against managed care in the US). Such integration
is hence risky if there is no choice between GP consortia. One
policy option would thus be to design choice and competition between
commissioners and allow new entrants subject to regulation by
Monitor (as the new proposed economic regulator). Over time, commissioners
could be free to negotiate with providers, could develop into
IIDSs, could restrict patient choice of provider on grounds of
quality, and replace rationing by Ministers with individuals choosing
between benefits packages.
(iv) Engaging GPs, patients and the
public in making hard choices in commissioning. GP consortia
will have to make hard choices in commissioning. To do this it
is helpful to use a "socio-technical" approach designed
to generate information on costs and benefits so that this can
be used systematically by patients, carers and staff providing
social and clinical services. The Systems Modelling for Performance
Optimisation and Service Equity (SyMPOSE) research programme at
LSE has developed such a socio-technical approach, in a five-year
research programme funded by the Health Foundation. This research
has developed a visual geometry which makes transparent three
key concepts for stakeholders: population health gain, Value for
Money, and scenarios that order interventions in terms of VfM.
This approach has been developed with PCTs in the Isle of Wight
and Sheffield and used by these PCTs to make strategic changes.
In Sheffield the strategy of eating disorders has potential materially
to reduce costs and produce health gain.
EVIDENCE FROM
GWYN BEVAN
1. This note aims to give evidence relevant
to following issues in the terms of reference of the Health Committee's
inquiry into Commissioning:
(a) Clinical involvement, clinical practice variation,
risk management and budgeting;
(b) Integration of primary and secondary health
care and designing a market for new entrants into commissioning;
and
(c) Engaging GPs, patients and the public in
making hard choices in commissioning.
Clinical involvement, practice variation, risk
management and budgeting
2. GP consortia may be seen as extensions,
in both scale and scope, of GP fundholding in the 1990s. Some
GP fundholders piloted such extensions as Total Purchasing Pilots
(TPPs): in principle GPs of TPPs could opt to commission all of
Hospital and Community Health Services (HCHS); and TPPs varied
from being single fundholding practices to large networks.
3. In practice TPPs opted to commission
selected services only (Mays et al, 2001). GP consortia may need
to work out how to organise commissioning with GPs may also decide
to focus on their involvement in different services.
4. The evaluation of TPPs gives findings
relevant to managing budgets and risks in GP consortia. The study
of TPP found that it proved much easier to involve GPs in the
single practice TPPs than in TPPs that were networks of practices.
The consequences were that the latter found it more difficult
to keep expenditure within budget than the former; and single
practice TPPs were no more likely to experience problems in managing
the risks of rare costly referrals than TPPs of networks (Baxter
et al, 2000). GP consortia are hence likely to face challenges
in involving GPs in budgetary control across their practices.
This may prove to be more difficult than it was for multi-practice
TPPs, as their practices had the experience of managing budgets
as GP fundholders, and they had opted to be part of the same TPP.
5. The funding of each GP consortium will
be determined with reference to its estimated past expenditure
and a target allocation, determined by a formula that takes account
of estimates of the size of its population, its relative need
and variations in unavoidable costs (such as from labour markets).
Its estimated past expenditure and target allocation will be aggregates
of estimates of each of these for its constituent practices. There
will be large differences between these two estimates for some
practices, with some targets implying the need for large reductions
in expenditure. Although this process offers potential to produce
fairer allocations of resources than the haphazard outcomes of
the past, at the level of the single practice, estimates of both
past expenditure and the target allocation will be subject to
large errors. Hence consortia will need a way of making sense
of these differences (Bevan, 1997).
6. Target allocations are based on average
rates of treatment for estimated relative need. So one reason
for differences between estimated past expenditure and a target
allocation will be that rates of hospital admissions are higher
or lower than the estimated average. A good way of assessing whether
these variations do or do not reflect differences in need is to
use information on Clinical Practice Variations (CPV)[54]
from analysis of small area variations in admission rates (as
reported in the US by the Dartmouth Atlas).[55]
If higher than expected volumes of admissions are concentrated
in categories of admission that are known to be high variation
with high levels of inappropriateness (such as tonsillectomy and
disc surgery), the cause is likely to be CPV, which suggests that
volumes ought to be reduced to average levels. If, however, higher
than expected volumes of admissions are concentrated in categories
of admission known to be low variation and with high levels of
appropriateness (such as acute myocardial infarction and hip fractures),
this suggests that the estimates of need are inadequate in capturing
the high risk of this population, and the target is too low (Bevan,
1997). Hence it would be helpful for GP consortia to have information
on the nature of CPV in England and advice on how to use this
information.
Integration of primary and secondary health care
and organising new entrants into commissioning
7. Enthoven (1985) compared the NHS in the
1980s with two successful high-performing Integrated Insurance
and Delivery Systems (IIDSs) in the US: Group Health Cooperative
of Puget Sound and the Kaiser Permanente HMO. In doing so, he
identified as a key weakness of the NHS that "it appears
locked forever into a model of separation between GPs and hospital-based
specialists" who "communicate with each other mostly
by mail" (Enthoven 1985: 46-47). This separation was observed
in studies of the NHS in the 1960s (Stevens, 2003) and still strikes
US visitors to-day. Enthoven argued that the NHS could benefit
from being reorganised into IIDSs "in which primary care
physicians are partners in regular contact with specialists, sharing
the same offices, records and equipment". The advantages
of this included easy, quick and informal consultation in a collegial
and collaborative atmosphere, with formal and informal learning
and quality assurance by peer review, and a shared comprehensive
medical record. Although Enthoven's preferred reform was to transform
the NHS into competing IIDSs, he deemed this to be not politically
feasible. So he advocated instead that each District Health Authority
would become an IIDS with a local monopoly for its defined population
by employing GPs and hospital specialists, and be empowered to
use contestable markets for delivery in an "internal market"
(ie contract out unsatisfactory services). He recognised that,
as these IIDSs would not compete, there would be a lack of economic
incentives for them to improve the delivery of care.
8. We now know that it proved virtually
impossible the US, in the 1990s, to replicate the model of high-performing
IIDSs. One symptom of the failure to do so was the backlash against
"managed care". The lessons from the literature on the
characteristics of high-performing IIDSs in the US identify obstacles
to developing these in England (Bevan and Janus, in press).
(a) High-performing IIDSs are organised as vertically
integrated systems that provide the consumer with health care
services across the whole continuum of care. This runs counter
to the historic division in British medicine (Honigsbaum, 1979)
and the organising NHS principle of the purchaser/provider split.
(b) High-performing IIDSs design systems of reimbursement
so that whatever is financially adverse for one of the components
carries over to the organisation as a whole. This does not apply
to the NHS because of the purchaser/provider split: with GP consortia
funded with reference to a formula (based on the needs of their
population) and paying providers according to the volume and complexity
of care they supply.
(c) High-performing IIDSs' commitment to cost
control and high quality care follows from the need to signal
high quality care to attract staff and enrollees from competitors.
As GP consortia do not compete, they will not face such pressures.
(d) High-performing IIDSs have sophisticated
management and information systems, which the NHS lacks.
(e) High-performing IIDSs have been stable over
decades and are large enough to provide integrated care across
a broad range of services. In contrast the NHS has been subjected
to fundamental reorganisations of commissioning bodies every few
years since 1991 and it is not yet clear if GP consortia would
be large enough to be a high-performing IIDS.
9. Developments in which GP consortia develop
integrated care with hospital specialists are welcome. But the
US experience shows that simply putting them in the same organisation
does not of itself result in high performing IIDSs. Indeed some
may turn out to be dysfunctional local monopolies. Hence, whilst
the path of integration is worthy of exploration, it seems vital
to allow choice of commissioner as part of such developments.
10. Patients are more likely to choose their
GP as a provider of primary care than on the commissioning effectiveness
of the GP's consortium. It would be possible to design a contestable
market for commissioning, in which individuals have an explicit
contract for an insurance package with their commissioner (as
in the Netherlands) and to open up this market to new entrants:
such as private insurers, or NHS Foundation Trusts offering integrated
services (Bevan and van de Ven, 2010). Monitor, as the new proposed
economic regulator, could be made responsible for developing a
contestable market for commissioning, regulating market entry
or changes to insurance coverage based on two requirements: demonstration
of key competences in commissioning; and definition of catchment
areas for which they could guarantee a duty of care (to commission
or deliver all necessary care). Over time, commissioners could
be free to negotiate with providers, could develop into IIDSs,
could restrict patient choice of provider on grounds of quality,
and replace rationing by Ministers with individuals choosing between
different insurance packages offering different benefits.
Engaging GPs, patients and the public in making
hard choices in commissioning
11. GP consortia will have to make hard
choices in commissioning. The Systems Modelling for Performance
Optimisation and Service Equity (SyMPOSE) research programme at
LSE has developed a socio-technical approach that would enable
them to do so, in a five-year research programme funded by the
Health Foundation. This approach develops a requisite model (Phillips,
1984) that can generate information on costs and benefits so that
this can be used systematically by key stakeholders (patients,
carers and staff providing social and clinical services) make
hard choices that take account of impacts on resources, population
health benefit and inequalities in health. We have collaborated
with Primary Care Trusts in the Isle of Wight and Sheffield and
worked with their key stakeholders to examine options and make
strategic decisions. This research has developed a visual geometry
which makes transparent three key concepts for stakeholders (see
Appendix 1):
(f) Rectangles of population health gain (with
numbers who benefit and the degree of benefit for a typical individual).
(g) Value for Money (VfM) triangles (with value,
costs, and VfM), where value includes population health gain and
can incorporate other criteria (such as reducing inequalities
in health and the probability of success of the implementation
of proposed changes).
(h) An efficiency frontier developed by ordering
interventions in terms of VfM.
12. Our approach requires stakeholders to
focus remorselessly on ensuring that they have all the data they
need to make decisions and to appreciate that it is much better
to have approximate estimates than none. Through a series of working
meetings led by an impartial facilitators (Decision Conferencing:
see Phillips, 2007) stakeholders were able to estimate the relative
health gains of a typical individual from a series of interventions
for mental health, cancers, dental health, cerebrovascular disease,
cancers, respiratory disease (and long term conditions), and children.
In this way they were able to generate rectangles of population
health gain and set priorities within and across disease areas
using VfM triangles that took account of: NHS costs and various
assessments of value.
13. This approach was first developed in
collaboration with the Isle of Wight PCT in 2008 to agree a strategy
for spending £1m of growth money (Health Foundation, 2009).
It was further developed with Sheffield PCT in 2009 to examine
scope for reallocating resources along care pathways for the treatment
of eating disorders, the prevention and treatment of three cancers
(breast, colorectal and lung) and dental care. The analysis of
the care pathway for the treatment of eating disorders informed
a new strategy to re-allocate resources to early interventions.
This appears to have reduced the need for intensive care for those
who are seriously ill and is likely to produce more population
health gain at reduced total costs. The analysis of the current
system of fees paid to dental practices for their Units of Dental
Activity showed that this was not designed to create incentives
for them to provider good VfM and we explored how that system
could be redesigned to do so.
14. Hence the SyMPOSE research programme
has developed new ways or organising the collection of data and
the presentation of information designed for deciding on the few
changes that can release substantial efficiency savings and gains
in value. This approach is of wide generalisibility: it can inform
those responsible for national and local policies, including the
design of systems of reimbursement to providers to generate VfM.
Furthermore this approach is designed so that to enable a method
of analysis to which clinicians, patients and the public can contribute
to the analyses and understand the results. This will be increasingly
important for the hard choices that the NHS now faces.
November 2010
REFERENCESBaxter
K et al. (2000) Primary care groups: Trade-offs in managing budgets
and risks. Public Money and Management, 20(1): 53-62.
Bevan G. (1997) Resource Allocation within Health
Authorities: Lessons from Total Purchasing Pilots. London:
King's Fund Publishing.
Bevan G, Janus K. Why hasn't integrated healthcare
developed widely in the US and not at all in England? Journal
of Health Politics, Policy and Law (in press).
Bevan G, van de Ven WPMM. (2010) Choice of providers
and Mutual Healthcare Purchasers: can the English NHS learn from
the Dutch reforms? Health Economics, Policy and Law, 5
(S3): 343-363.
Bevan G et al. (2004) Using information on variation
in rates of supply to question professional discretion in public
services. Financial Accountability and Management, 20
(1): 1-17.
Enthoven A. (1985) Reflections on the management
of the NHS. London: Nuffield Provincial Hospitals Trust.
Glover A J. (1938) The Incidence of Tonsillectomy
in Schoolchildren, Proceedings of the Royal Society of Medicine,
31: 1219-36.
Health Foundation (2009) Improvement in practice:
Commissioning with the community. London: Health Foundation.
Honigsbaum F. (1979) The Division in British Medicine.
London: Kogan Page.
McPherson K et al. (1996) Systematic Variation
in Surgical Procedures and Hospital Admission Rates. London,
School of Hygiene and Tropical Medicine).
Mays N et al. (eds) (2001) The Purchasing of health
care by primary care organisations. An evaluation and guide to
future policy, Buckingham: Open University Press.
Phillips L D. (2007). Decision Conferencing. Chapter
19 in W. Edwards et al. (Eds) Advances in Decision Analysis.
From Foundations to Applications. New York: Cambridge University
Press; p375-99.
Phillips L D. (1984). A theory of requisite decision
models. Acta Psychologica 56: 29-48.
Stevens R. (2003) Medical practice in modern England.
Yale University Press.
Wennberg J E et al. (1984) Will Payment Based on
Diagnosis-Related Groups Control Hospital Costs? New England
Journal of Medicine 311: 295-300.
APPENDIX 1
THE SyMPOSE APPROACH TO VALUE FOR MONEY
1. The SyMPOSE approach generates a series
of four estimates.
(a) Quantitative estimates of the population
health gain for each policy option:
(i) the numbers who are likely to benefit;
(ii) the typical individual who is likely to benefit;
(iii) a standardised health gain score between
100 (for the best) & zero (for no benefit).
(b) The total population health gain for each
policy option as the product of the numbers who are likely to
benefit and its health gain score.
(c) The development of a Value for Money (VfM)
triangle for each policy option, with:
(i) health benefit score as the vertical line
which can combine multiple criteria (population health gain; reduction
in inequality; and the ease or difficulty of successful implementation);
(ii) cost as the horizontal line; and
(iii) VfM as the slope.
(d) The ranking of policy options to develop
an efficiency frontier in terms of VFM (or scale of health benefit).
2. Table 1 illustrates the way estimates
for quantitative estimates of the population health gain for three
policy options in treatment and prevention of cancers were derived
in the Isle of Wight (IoW) collaboration and Figure 1 illustrates
the total population health gain for policy option (the product
of the numbers who are likely to benefit and its health gain score).
Figure 2 illustrates the concept of the VfM triangle.
Table 1
ASSESSING HEALTH BENEFITS FOR POLICY OPTIONS
Initiative
| No Benefit | "average" beneficiary
| Health gain score |
| Early detection & diagnosis | 200
| Femail, mid-60s, "hard to reach"
| 100 |
Palliative & End of life care |
1,500 | Late 70s, life limiting long term health condition
all socio-economic groups
| 75 |
Relocate active treatment | 300
| Mid-60s, female, very sick | 25
|
Figure 1: population health gain for each policy option
Figure 2: The VfM triangle
 |
Risk-adjusted
benefits(compared
to current care)
taking account of
doability |
3. Figure 3 shows the outcomes of applying this approach
for eating disorders in Sheffield. The pattern of care in 2009
was analysed as producing in total 50 units of estimated benefits
at a total cost of £2.2 million; with 90% of the costs spent
on a small number of patients who had become acutely ill and required
costly intensive care that produces only limited benefits for
each patient. Stakeholders agreed to expand services at an early
stage, when interventions are low cost and highly effective, to
reduce the need for costly intensive care later. This strategy
has potential to reduce total costs (to £1.4 million) and
double the estimated benefits (to 100 units).
Figure 3: Analysis of treatment of eating disorders in Sheffield
53
Gwyn Bevan is Professor of Management Science and Director of
the MSc in Public Management and Governance in the Department
of Management at the London School of Economics and Political
Science where he is an associate of two research centres: LSE
Health and the Centre for Analysis of Risk and Regulation. His
current research includes: comparisons of performance of health
care and schools across the UK; developing methods for reducing
NHS expenditure for least harm; methods of equitable funding of
insured populations; and implications of introducing purchaser
competition into the English NHS. He is a member of the Department
of Health's two advisory groups on formulas used to allocate resources
to Primary Care Trusts. His report for the Secretary of State
for Health on the developments of these formulas was published
in 2008. From 2001 to 2003 he was Director of the Office for Information
on Health Care Performance at the Commission for Health Improvement
(CHI) where he had lead responsibility (for the NHS in England)
for: developing "star ratings"; national surveys of
staff and patients; developing national clinical audits; and analyses
for CHI's clinical governance reviews, investigations, and national
studies. He has worked for the National Coal Board, HM Treasury,
and an economic consultancy; and, as an academic, at Warwick Business
School, and the Medical Schools of St Thomas's Hospital and Bristol
University. Back
54
CPV exists when different doctors make different decisions about
the same or similar patients. The classic example is that of adenoidectomy/tonsillectomy
described in the 1930s as "a prophylactic ritual carried
out for no particular reason with no particular result" (Glover,
1938). Since then there has been a vast literature on CPV in GP
referral rates and hospital admission rates. (Although variations
in demands by patients are likely also to contribute to these
variations). Studies in the US (Wennberg et al, 1984) and
in England (McPherson K et al 1996; Bevan et al, 2004)
have found about 90% of hospital admissions were in a high variation
category. Back
55
See: http://www.dartmouthatlas.org/ Back
| 
