Review of the General Medical Services
global sum formula
9 February 2007
Contents
Foreword
1
Executive summary
Summary of recommendations
3
7
Chapter 1
Introduction – the global sum and the Carr-Hill formula
9
SECTION A: THE PROCESS OF THE REVIEW
Chapter 2
Chapter 3
Chapter 4
Scope and timeline of the review
Structure of the review
The formula review and resource allocation
10
11
12
SECTION B: THE FINDINGS OF THE REVIEW
Chapter 5
Chapter 6
Chapter 7
Chapter 8
Chapter 9
Review of workload factors
Review of cost factors – variations in labour costs
Review of cost factors – isolation and rurality costs
Combining the formula adjustments
The recommended formula
13
22
27
34
35
SECTION C: TAKING THE RECOMMENDATIONS
OF THE REVIEW FORWARD
Chapter 10
Chapter 11
Chapter 12
Implementation issues
Data recommendations
Impact of the formula review on the devolved administrations
41
44
45
GLOSSARY
47
APPENDICES
Carr-Hill resource allocation formula
Formula Review Group membership
Components of QRESEARCH models
Guide to normalisation in the global sum formula
Calculation of consultation length and home visit adjustment weights
Projected distributional impact of the recommended formula without
the rurality index compared to the current global sum formula
Appendix G Projected distributional impact of the recommended formula with the
rurality index compared to the current global sum formula
Appendix H Projected distributional impact of the recommended formula with the
rurality index compared to the recommended formula without the
rurality index
Guide to the projected distributional impact of the recommended
Appendix I
formula
Appendix A
Appendix B
Appendix C
Appendix D
Appendix E
Appendix F
49
55
56
57
60
61
62
63
64
Foreword
To support the new GMS contract introduced in 2003, Professor Roy Carr-Hill and
colleagues were asked to develop an allocation methodology which became known
as the “the Carr-Hill formula” or the Global Sum formula. When the new GMS
contract was introduced the Department of Health, the General Practitioners
Committee and the NHS Confederation gave a commitment to review the formula
and this report and its recommendations are the result of that review.
The review was carried out by a Formula Review Group led by NHS Employers and
comprised senior colleagues from the General Practitioners Committee of the British
Medical Association and the four UK Health Departments. The review was supported
by independent academic research. I would like to thank each of the contributing
organisations for their commitment and involvement in this important review which
may directly affect the income and livelihood of many of the country’s 35,000 general
practitioners. We were fortunate in having a considerable amount of time in which to
carry out this comprehensive review – a luxury not afforded to Professor Carr-Hill and
colleagues who were originally tasked with developing a formula to a much shorter
timescale.
The review allowed us to collate data not previously available and to obtain
considerably more up-to-date data than had previously been to hand – particularly
that becoming available from the evolution of the new GMS contract. I would
particularly like to thank the Department of Health for their investment in the review
and the commissioned research. I know senior Department of Health officials share
the profession’s desire to secure a robust and credible methodology underpinning the
equitable and transparent distribution of some £1.6 billion of public funds and
informing allocations of a further £2 billion.
The Formula Review Group considered the many comments about the current
allocation formula which fell into two broad areas: those that related to the technical
content and operation of the formula; and those that related to the perceived political
or negotiating decisions which were outwith the original remit of Professor Carr-Hill
and colleagues. We have covered the former through this review and our associated
recommendations and we have included and commented upon the latter where we
felt that such ‘non-formula’ issues had a material impact upon the equitable
distribution of GMS resources. It was not the remit of the review to identify the
resource envelope which would be required to appropriately fund global sum budgets
across all of general practice. The purpose of the formula is to forecast the ‘relative’
costs of one general practice compared to another – that does not tell us how much
funding each practice should recieve.
Not surprisingly the issue of the Minimum Practice Income Guarantee was a
recurring component of discussions during the course of the review and we have
commented on this in Chapter 10. However, it was important that work relating to the
technical components of the formula were considered as objectively as possible and
therefore for individual components of the formula we explicitly excluded any
potential skewing of results through the separate or subsequent application of any
Minimum Practice Income Guarantee. The Government’s White Paper Our health,
our care, our say: a new direction for community services gave a commitment for a
separate review of the Minimum Practice Income Guarantee as well as equity across
Primary Medical Services. The findings of that review may further supplement and
overlap with the recommendations within this report. I believe and hope that this will
increase our ability to distribute available resources in as equitable a way as possible
Review of the General Medical Services global sum formula
1
so that all patients receive the same high quality level of care and enjoy the same
quality of access to services regardless of where they may live or their social
background.
This report rightly focuses on the components of the allocation formula and I hope we
have explained the more technical and analytical elements as transparently and
clearly as possible. I believe much of the criticism of the original formula was due to a
lack of information and clear explanation of the decisions taken at the time of its
implementation. I have therefore encouraged the wider publication and sharing of this
review to raise awareness of these issues as well as the reasons for the
recommendations now being made. However, we should remember that it is not
scientifically possible to forecast exact future workload and the associated resources
required.
I am pleased that the Department of Health, the Welsh Assembly, the General
Practitioners Committee and NHS Employers have agreed to the publication of this
report. In my view this is the right way forward and hopefully offers readers the level
of detail and explanation which would have been helpful at the time of the original
formula. I hope that you will take this opportunity to feedback your comments on the
results of the review and particularly provide us with your views on the individual
consultation questions.
Philip Grant
Chair of the Formula Review Group
2
Review of the General Medical Services global sum formula
Executive summary
Introduction
1.
This report presents the results of the review of the GMS global sum formula
which has been undertaken by the Formula Review Group (FRG) established
by NHS Employers and the General Practitioners Committee (GPC) of the
British Medical Association (BMA).
2.
The FRG included representatives of the GPC, NHS Employers and the four
UK Health Departments as well as independent academic support. We were
required to report our findings and recommendations to the GMS Plenary which
is the main development and negotiating forum between NHS Employers and
the GPC for all matters relating to the GMS contract and funding.
3.
The GMS global sum formula distributes global sum payments to practices in
line with the weighted needs of patients to reflect practice workload and the
relative costs of service delivery. A commitment to review the formula was
made by Plenary following the formula’s introduction in 2003 and concerns
being raised at the time by some GPs regarding the fairness, robustness and
reliability of data supporting the allocation of resources.
4.
In undertaking our review of the formula we have based our findings and
recommendations wherever possible on evidence-based research, much of
which we directly commissioned ourselves.
The recommended formula
5.
Following examination of the factors in the current global sum formula and the
investigation of additional factors for possible inclusion in a revised formula, we
recommend that the revised global sum formula should adjust for the following:
•
workload
•
consultation length and home visits
•
staff Market Forces Factor (MFF)
•
Cost of Recruitment and Retention (CORR)
•
Cost of Unavoidable Smallness (CUS)
•
(possibly) rurality.
Workload adjustment
6.
The single workload adjustment reflects the effect of patient, local area and
practice characteristics upon practice workload. This adjustment is based on
analysis by QRESEARCH (see Chapter 5) and would replace the four separate
workload adjustments used in the current global sum formula. The new
adjustment would be based on the following variables:
•
age-sex bandings
•
newly registered or temporary patients (patients which have joined the
practice in the past twelve months)
Review of the General Medical Services global sum formula
3
•
7.
the Index of Multiple Deprivation (IMD) health domain score for the patient’s
electoral ward of residence.
There are a number of benefits to this approach which estimated all workload
factors in single model. This suggests that its outputs are an improvement on
those produced by the original global sum formula research.
Consultation length and home visits adjustment
8.
We developed a consultation length and home visit adjustment to supplement
the workload adjustment. This adjustment is based on consultation length and
home visits data from the General Practice Research Database (GPRD) and
the age-sex band from QRESEARCH.
Staff Market Forces Factor adjustment
9.
The current global sum formula makes adjustment for a staff MFF to reflect the
geographical variation in staff costs that practices will incur. This current
adjustment is based on data periodically updated by the Advisory Committee
for Resource Allocation (ACRA) which reviews the unified PCT resource
allocation formula. ACRA is currently reviewing the MFF element of that
formula, and we recommend that, until ACRA’s recommendations about the
MFF become available, the existing methodology for the adjustment should be
maintained. This adjustment should be periodically adjusted to use the latest
available data on staff MFF weights as they become available.
Cost of Recruitment and Retention adjustment
10. The CORR adjustment allows for the extra costs of recruitment and retention
that may be necessary to attract GPs to practices in relatively deprived areas.
This adjustment is based on research by the Health Economics Research Unit
at the University of Aberdeen which quantified the relationship between
indicators of GP recruitment and retention difficulties and possible explanatory
variables.
11.
The CORR adjustment formula (used to calculate each practice’s CORR
adjustment index) includes the narrow comparator Standardised Spatial Wage
Differential (SSWD) which is a measure of the wage premium earned by private
sector employees in a given geographical area. The CORR adjustment formula
also includes the average Limiting Long Term Illness (LLTI) ratio for the
practice which is widely used to indicate the chronic health needs associated
with deprivation.
Cost of Unavoidable Smallness (CUS) adjustment
12. The CUS adjustment allows the formula to take account of the lost economies
of scale effects for isolated rural practices which unavoidably have a small list
size. This adjustment, which is based on research by Deloitte (see Chapter 7),
consists of an economies of scale adjustment and an isolation criteria.
13.
4
The economies of scale adjustment reflects the relationship between list size
and expenses per patient that exist for practices with small list sizes. As it
would be inappropriate to reward small practices without recognising the cause,
an isolation criteria is then applied that qualifies the extent to which a small
practice should benefit from the economies of scale adjustment based on the
degree to which its smallness is unavoidable.
Review of the General Medical Services global sum formula
Rurality adjustment
14. The current rurality adjustment is intended to reflect the uncontrollable
additional costs associated with the degree to which the area served is rural.
While the new CUS adjustment compensates for the unavoidable costs of
practices that are necessarily small because of their isolated location, it could
be argued that the rurality adjustment should be applicable to practices
irrespective of list size.
15.
We were unable to recommend whether or not a rurality adjustment should be
included in the revised formula due to a lack of evidence and rationale to
support its inclusion. However, we were able to recommend that if a rurality
adjustment was adopted it should be a specifically updated version of the
current rurality adjustment.
16.
It could be justified that a rurality adjustment be included in the revised formula
because the original analysis showed that rurality was associated with
increased expenses per capita after allowing for list size, so it is arguable that a
rurality adjustment should be applicable to practices irrespective of size.
17.
However, there are also reasons why a rurality adjustment should not be
included in the revised formula. These are:
18.
•
there is an issue around the validity of the original analysis, which was
based on data that preceded the introduction of the nGMS contract. It is
possible that the higher expenses of rural practices are a reflection of
previous payment mechanisms
•
we are aware of a perception that the current adjustment is not particularly
well targeted and that it benefits leafy suburbs as well as the most rural
practices because it is a continuous function of density and distance
•
we appreciate that whilst it is objectively valid, it may appear unnecessarily
complicated to include two adjustments that address seemingly similar
issues.
Additionally, the Carr-Hill rurality adjustment includes patients’ average distance
to practice. Our health, our care, our say: a new direction for community
services states that patients should have more choice to register with the
practice most convenient for their particular needs and circumstances. Patients
may choose to register with practices some distance from their home, and
therefore average distance to practice would no longer be a good measure of
rurality.
Factors considered but not recommended for inclusion in the new
formula
19.
We considered a number of factors that we do not recommend for inclusion in
the new formula. These include:
•
QOF prevalence
•
patients living in nursing and residential homes
•
ethnicity
•
patients who speak a different language from their GP
•
GP Market Forces Factor (MFF).
Review of the General Medical Services global sum formula
5
QOF prevalence
20. QRESEARCH suggested that we could include QOF prevalence in the
workload adjustment, and recommended that patient-level QOF data should be
used. However, we found that an adjustment using patient-level QOF data
would be difficult to implement. Instead, we recommend that there should be a
review of the technical changes required to provide patient-level QOF data in
future years, to allow the option of using these data in the future.
Patients living in nursing and residential homes
21. While the current global sum formula includes an adjustment for nursing and
residential homes, QRESEARCH were unable to adequately define which
patients were living in nursing and residential homes using their database. In
addition, work by us showed that there was a negligible effect on the
distribution of weighted patients from the removal of this adjustment.
Ethnicity
22. The QRESEARCH analysis suggested that consultation rates decreased as the
percentage of the white population increased. We interpreted the negative
effect of ethnicity on workload as evidence of unmet need on non-white groups
and we agreed that it would therefore be inappropriate to reduce practice
payments on the basis of ethnicity.
Patients who speak a different language to their primary health professional
23. We noted that direct information on the number of patients speaking a different
language to their primary health professional was not currently recorded on any
database. We agreed that it would therefore be impossible to implement a new
factor, other than at local level.
GP Market Forces Factor (MFF)
24. While we noted some arguments in favour of the inclusion of a GP MFF
adjustment, this adjustment would be inconsistent with the CORR research,
which found that GP recruitment and retention problems bear little relationship
to private sector pay comparisons.
Modelling the recommended formula
25.
We considered the projected distributional impact of the recommended formula
both with and without the rurality index compared to the current global sum
formula.
26.
The modelling showed that adopting the formula without the additional rurality
adjustment would result in:
6
•
a change in weighted patients for GMS practices ranging from -30% to
+65%. Excluding the 1% most extreme practices (0.5% at each extreme),
the range would be -19% to +29%
•
a general increase in the weighted capitation share of urban practices,
practices with high additional needs, practices with high proportions of new
registrations, practices with low proportions of patients in nursing and
residential homes, practices with low proportions of elderly patients, London
practices and practices in spearhead PCTs.
Review of the General Medical Services global sum formula
27.
28.
When using the additional rurality adjustment the modelling showed that:
•
the change in weighted patients for GMS practices would range from -19%
to +83%. Excluding the 1% most extreme practices, the range would be
-11% to +28%
•
there would be a general increase in the weighted capitation share of urban
practices, practices with high proportions of new registrations, practices with
low proportions of nursing and residential home patients, practices with low
proportions of elderly patients and London practices.
Having compared each recommended formula with the current formula, we
considered the projected distributional impact of the recommended formula with
the rurality index compared to the recommended formula without the rurality
index. It showed that:
•
compared to the recommended formula without the rurality index, including
the rurality index would on average tend to increase the weighted capitation
share of rural practices, practices with low additional needs, practices with
higher proportions of elderly patients and practices outside of London.
Implementation issues
The London adjustment
29. On the basis of improvements to the formula, we recommend that, should the
new formula be implemented, the London adjustment should be discontinued.
The Minimum Practice Income Guarantee (MPIG)
30. We agreed that the historic constitution of MPIG and correction factor payments
protected practices from the negative impact of any redistribution of resource
envelope based on the agreed formula. However, the financial stability of
individual practices was recognised as vital.
31.
We have based our conclusions and financial modelling on the principle that a
revised formula would be applied to the same resource envelope.
Recommendations
32.
Our report makes the following recommendations:
We recommend a revised formula that includes the following
components:
•
workload adjustment
•
consultation length and home visit adjustment
•
staff Market Forces Factor (MFF) adjustment
•
Cost of Recruitment and Retention (CORR) adjustment
•
Cost of Unavoidable Smallness (CUS) adjustment
•
(possibly the) rurality adjustment.
Review of the General Medical Services global sum formula
7
The separate adjustments should be multiplied together to generate the
aggregate formula adjustment.
We recommend that Connecting for Health should review the technical
changes required to provide patient-level data, in particular QOF data, in
future years.
We recommend that the London adjustment should be discontinued if the
revised global sum formula is adopted.
We recommend that the collection of the following could be beneficial to
future developments of the formula:
8
•
the ability to link patient-level QOF data to the National Health
Application and Infrastructure System used to apply the global sum
formula (also known as the Exeter system)
•
workload data for patients living in nursing and residential homes
•
direct information on the number of patients speaking a different
language to their primary health care professional
•
actual patient-level socioeconomic data.
Review of the General Medical Services global sum formula
1
Introduction – the global sum and the
Carr-Hill formula
1.1
In February 2003, the NHS Confederation and the General Practitioners
Committee (GPC) of the British Medical Association (BMA) jointly published
Investing in General Practice: The New General Medical Services Contract
which set out the details of the new GMS contract following the outcome of their
negotiations over the previous 16 months.
1.2
The new contract agreement included movement from the old Red Book
remuneration arrangements to a practice-based contract with core investment
via a global sum, distributed in line with the weighted needs of patients to reflect
practice workload and complexity and the relative costs of service delivery.
1.3
The current GMS global sum formula, developed with the support of a number
of academic teams including Professor Roy Carr-Hill of York University,
provided the basis for the distribution of global sum payments by calculating
each practice’s fair share of the total global sum resource. The formula did not
determine the total global sum resources available nationally.
1.4
The current formula takes account of six key determinants of practice workload
and circumstances:
(i)
patient sex and age for frequency and length of surgery and home visit
contacts
(ii) nursing and residential home status
(iii) morbidity and mortality
(iv) newly registered patients
(v) unavoidable costs of rurality
(vi) unavoidable higher costs of living through a MFF applied to the costs
associated with employing practice staff. In particular, this compensates
for those additional costs involved in delivering services in high cost-ofliving areas such as the south east of England.
1.5
Applying the indices together to a practice’s population creates a practice
weighting. This determines a practice’s global sum entitlement.
1.6
Appendix A presents the core findings from the analysis used to derive the
current formula.
1.7
The global sum also includes two off-formula adjustments:
(i)
an adjustment for the treatment of temporary residents and the provision
of immediately necessary and emergency treatment
(ii) an adjustment to recognise the particular circumstances of practices in
London.
1.8
Some GPs raised concerns about the accuracy and robustness of the current
formula after details of the formula were published in 2003. In response to
these concerns the negotiators moved to reassure the profession and the NHS
by promising that the formula would be reviewed in light of the developing
contract and the availability of additional data.
1.9
This report has been prepared as a result of that review and presents the
findings of the Formula Review Group.
Review of the General Medical Services global sum formula
9
SECTION A: THE PROCESS OF THE REVIEW
2
Scope and timeline of the review
2.1
Our review group was established in December 2004 to:
2.2
•
undertake a thorough review of the payments for GMS essential and
additional services
•
examine the current global sum formula, including all factors currently
included, and investigate additional factors for possible inclusion or
exclusion in a revised formula, subject to evidence
•
propose to Plenary any necessary revisions to the current allocation
formula, taking account of evidence and resources.
Our objectives were to:
•
evaluate whether the current formula delivers a fair distribution of resources,
based on those factors currently included in the formula and the introduction
of additional factors where this is supported by evidence, and make
recommendations about whether fair distribution could be achieved
•
consider redistribution of resources to areas of high health inequalities and
of significant primary care workforce shortfalls
•
consider practices’ relative workload and the relative costs of service
delivery
•
distribute effectively the resources available within the global sum.
2.3
Our membership consisted of experts in the relevant technical, policy and
clinical areas and included representatives from the British Medical
Association’s General Practitioners Committee (GPC), NHS Employers and the
four UK Health Departments as well as independent academic support. The
Chairman of the group was appointed by all parties. A list of FRG members is
included at Appendix B.
2.4
We were established by and therefore required to report to GMS Plenary. The
Plenary, which consists of negotiators from the GPC and NHS Employers,
negotiates all matters relating to the GMS contract and funding. As such, while
we would make recommendations, any decisions regarding implementation
would ultimately be the responsibility of Plenary.
2.5
While it was originally intended that we would report to GMS Plenary so that a
new formula could be implemented in April 2006 together with other changes to
the GMS contract, the GPC and NHS Employers subsequently agreed that the
wider review of the contract should take place in two stages – 2006/07 and
2007/08. Following discussion about the implications of this decision the
negotiators agreed that a significant formula review, in time for April 2006
implementation, was unlikely to be achievable or desirable. Instead, it was
agreed that we should produce a report for consultation in 2006.
10
Review of the General Medical Services global sum formula
3
Structure of the review
3.1
The factors in the current formula are divided into two types: workload factors
and cost factors. Similarly, all proposed new factors are categorised into one of
these types.
3.2
Workload factors are those that impact upon the workload of a practice, or the
time required to provide care to patients. These are often related to the types of
patients seen by the GPs in a practice and account for the fact that some types
of patients impose a higher workload. The workload factors we considered in
this review are described in Chapter 5.
3.3
Cost factors are those that impact upon the expenditure needing to be incurred
by a practice to deliver services to its patients. These are related to the costs
incurred by practices in delivering services. The cost factors we considered in
this review are described in Chapters 6 and 7.
3.4
We also considered how the various components of the formula should be
combined. This included estimating the workload aspects of the formula in a
single model, assessing how the separate numerical components of the model
should mathematically be combined and, reviewing the weight given to the
various components of the formula.
3.5
We undertook a number of pieces of work to assess the potential impact of the
revised formula. This included practice-level modelling of the projected
distributional impact on weighted patients of any changes to existing factors
compared with the current formula. We used a further analysis of this modelling
to project the distributional impact of formula changes on particular cohorts of
practices (for example on practices grouped by indicators of rurality or
deprivation). We also prepared case studies to demonstrate how the formula
would affect specific individual practices.
3.6
We also considered a number of issues that would impact upon implementation
of its recommendations including the London adjustment and the MPIG. These
are discussed in Chapter 10.
Review of the General Medical Services global sum formula
11
4
The formula review and resource allocation
4.1
Primary care trusts receive their main funding through the Department of Health
in the form of a Unified Resource Allocation. The formula which underpins the
distribution of the Unified Resource Allocation is developed and maintained by
the Advisory Committee on Resource Allocation (ACRA) which is an
independent multi-professional body.
4.2
The Unified Resource Allocation formula allocates funding to PCTs for:
•
Hospital and community health services (HCHS)
•
prescribing
•
primary medical services
•
HIV/AIDS.
4.3
We obviously needed to take account of ACRA’s work to avoid potential
significant mismatches between allocations and GMS funding as well as to take
advantage of methodological improvements which might be mutually beneficial.
4.4
There were a number of areas where the work of ACRA and the FRG coincide
and where there was scope for comparing methodologies. These were:
4.5
•
relevant populations
•
additional needs
•
Market Forces Factors.
Issues regarding additional needs and Market Forces Factors will be discussed
in Chapter 5 and Chapter 6 of this report.
Relevant populations
4.6
12
The population base for the current resource allocation formula in England is
the registered population. This is determined using GP registrations in the
Attribution Data Set (ADS) constrained to Office for National Statistics (ONS)
population estimates for the year in question. The ONS constrained registered
population is used for the entire allocation, although, under the GMS global sum
formula PCTs pay practices on the basis of their unconstrained registered list.
ACRA is engaged in a major piece of work at present on appropriate population
bases for revenue allocations after 2007/08. This work forms part of ACRA’s
work programme which is due to be completed in the Autumn of 2007. ACRA
will then make recommendations to Ministers on possible changes to the
overall formula.
Review of the General Medical Services global sum formula
SECTION B: THE FINDINGS OF THE REVIEW
5
Review of workload factors
Current formula
5.1
5.2
Factors
The current global sum formula makes adjustments for four practice workload
factors:
•
the age-sex mix of the practice population
•
the nursing and residential home population of the practice
•
the number of new registrations in the practice population
•
the additional needs of the practice population.
An outline of the research underlying these adjustments is available at
Appendix A.
Age-sex adjustment
The age-sex adjustment reflects the effect of patient age and gender on
workload. This adjustment used the General Practice Research Database
(GPRD) which measures instances of a patient’s computer file being accessed.
In this research, the numbers and length of patient file openings were used to
measure workload. This was compared to the age and sex of patients to
develop the age-sex adjustment. The current adjustment consists of seven age
bandings for each sex.
5.3
Nursing and residential homes adjustment
This adjustment reflects the additional workload associated with patients in
nursing and residential homes based on a survey of home managers (used to
measure age and sex specific consultation rates) and GPs (used to estimate
consultation length and travel time).
5.4
Number of new registrations in the practice population
Using a similar methodology to the age-sex adjustment, Professor Roy Carr-Hill
developed an adjustment which reflects the extra practice workload associated
with newly registered patients and list turnover.
5.5
Additional needs adjustment
This adjustment reflects the other patient characteristics which influence
workload. It was developed by Dr Stephen Morris, Professor Matthew Sutton
and Professor Hugh Gravelle. Using Health Survey for England data on GP
consultations from 1998/2000 and other area-level data, they found that, of the
variables tested, Standardised Limited Long-Standing Illness (SLLI) and the
Standardised Mortality Ratio for those aged under 65 (SMR<65) best explained
variations in workload over and above age and sex. The adjustment was
created accordingly.
5.6
Limitations of the original approach
5.7 All datasets have limitations and the analysis using GPRD was constrained by
the way in which consultations were recorded – GPRD measures instances of a
patient’s computer file being accessed. A receptionist checking an appointment
Review of the General Medical Services global sum formula
13
or a GP issuing a repeat prescription would both contribute to measured
workload. Whilst the original research made adjustments to try and account for
such concerns, there remain limitations with the completeness and accuracy of
the workload data.
5.8
5.9
During the review, concerns about the data used to derive the additional needs
adjustment have also been raised:
•
Health Survey for England data on GP visits were based on patients’ self
reports, which may be less accurate than administrative data
•
Health Survey for England respondents were only asked the number of
visits to the surgery in the previous two weeks, which raises the possibility
of seasonal effects influencing the analysis
•
it did not take account of consultations with practice nurses. If the ratio of
practice nurse to GP consultations is lower in certain areas, then the original
research would overestimate the strength of the additional needs
adjustment on the workload of the primary care team and vice versa
•
it used a “boost sample” of data, where certain groups had been
oversampled in specific years, e.g. patients with chronic vascular disease,
minority ethnic groups, older people and the socially excluded
•
it was an electoral ward-level analysis, built up from individual-level data.
This means that the analysis may ultimately be based on a relatively small
number of individual observations that are not representative of their ward.
Wards with only a small number of individual observations were excluded to
help address this, but this raises a number of issues and potentially skews
the sample to urban areas that are likely to have greater additional needs. If
the relationship between additional needs indicators and consultation
number is non-linear, then the resulting additional needs adjustment may be
unreliable.
Most importantly, this sequential approach involved separately estimated
factors of workload which were then combined. This could lead to double
counting in the allocation process. For example, the impact of nursing and
residential homes on workload is included directly in the formula via the nursing
and residential homes adjustment. However, it is also included indirectly in the
age-sex adjustment as this did not control for the effect of patients in nursing
and residential homes. This is a general risk with the separate estimation of
workload adjustments.
Commissioned research
5.10 We based our recommendation for a workload adjustment in the revised global
sum formula on a multivariate regression analysis linking practice workload to
the full range of various patient and practice characteristics.
5.11 We commissioned QRESEARCH, at the University of Nottingham, to develop a
combined model that accounted for workload factors. QRESEARCH were
supported by a technical steering group drawn from FRG membership which
was responsible for the study design, analysis plan and interpretation of the
analysis.
14
Review of the General Medical Services global sum formula
5.12 The overall aim of the QRESEARCH analysis was to explain variations in
workload in terms of patient, local area and practice level covariates.
QRESEARCH measured the total number of GP and nurse consultations for
each patient in their study practices (Consultations with community/district
nurses were excluded as these nurses are funded by the PCTs).
5.13 The report of QRESEARCH’s analysis is available in the supporting technical
document. This should be read in conjunction with Comparisons of key practice
characteristics between general practitioners in England and Wales and
general practices in the QRESEARCH database, which is also available in the
supporting technical document.
5.14 There are a number of key benefits of the QRESEARCH database and the
multivariate approach taken to estimate all workload factors in a single model:
•
QRESEARCH records actual consultations (instead of file openings)
•
QRESEARCH is a large database (it has over three and a half million
patients) with a largely representative geographical spread throughout the
English regions
•
according to Comparison of key practice characteristics between general
practitioners in England and Wales and general practices in the
QRESEARCH database (available in the supporting technical document),
the QRESEARCH sample of practices is comparable with those in other
national databases (e.g. the Attribution Data Set and QMAS data for
practice disease prevalence rates)
•
the multivariate approach is less likely to lead to double counting
•
unlike the previous research undertaken for the additional needs adjustment
in the current formula, the combined QRESEARCH model:
-
uses practice nurse as well as GP consultations
-
uses administrative rather than self-reported consultation data
-
is based on a large sample of data
-
is based on an entire year of data, rather than annualised self-reported
consultations in the previous fortnight. This prevents the influence of
any seasonal effects.
5.15 The strengths of the QRESEARCH approach suggest that its outputs are more
robust than those produced by the original research.
5.16 However, it is also clear that the QRESEARCH approach has some limitations
compared to the original formula research:
•
the data do not include consultation length or intensity, although the original
research used proxies for these in terms of length of file opening
•
it did not give special consideration to the length of home visits (which were
counted as consultations, but were not weighted differently). Home visits
are a relatively small proportion of patient contacts and we address this
issue in the consultation length and home visit adjustment described later in
this chapter
•
the data had fewer additional needs indicators. However, those that were
available were highly correlated with those in the current additional needs
adjustment. For example, the additional needs indicator used in the
Review of the General Medical Services global sum formula
15
recommended QRESEARCH model (the Index of Multiple Deprivation
Health domain), averaged at practice level, has a correlation coefficient of
over 0.75 with both SLLI and SMR<65.
Recommended model
5.17 On the basis of the QRESEARCH analysis, We recommend the adoption of a
combined workload adjustment for the revised global sum formula based on
QRESEARCH model 20 to replace the four separate workload adjustments
currently used.
5.18 This workload adjustment would be based on the following variables:
•
age-sex band (using the current age bandings)
•
newly registered patients (maintaining the current definition – patients joined
in the last twelve months)
•
Index of Multiple Deprivation (IMD) Health domain score for the patient’s
electoral ward of residence. (An equivalent model using the Townsend
score for deprivation was also available but was decided against on the
grounds that IMD Health was more focused on health-related needs. Full
details of this model are given in the QRESEARCH report and further
descriptions of both measures are given in Appendix C.)
5.19 The coefficients for the factors used to calculate patient weightings are given in
Table 1.
5.20 From this, a workload adjustment for each practice can be calculated as
follows:
Step 1: For each patient in the practice calculate an overall workload weight
using the formula below and the component weights in Table 1
below:
Patient overall
workload weight
= Age-Sex
weight
* Registration
status weight
* IMD Health Domain
score weight
Example: A female aged 50 who is newly registered with the practice
and living in an area with an IMD Health Domain score of 0.7 would
have an overall workload weight of:
2.115
*
(Age-Sex weight)
1.689
*
(Registration status weight)
(1.0540.7)
=
3.706
(IMD Health Score Weight)
Step 2: For each practice calculate the pre-normalised workload weighted
patients by summing the overall workload weights of all the patients
registered with the practice.
Step 3: Apply the standard formula methodology for normalisation (explained
in Appendix D) to calculate practice normalised workload weighted
patients.
Step 4: Divide practice normalised workload weighted patients by practice
unweighted list size to give the practice workload adjustment index.
16
Review of the General Medical Services global sum formula
Table 1: Component weights for the calculation of patient level overall
workload weights
Age-Sex weight
Registration status weight
Band
Weight
Band
Weight
IMD Health Domain
score weight
Weight
Male 0-4 years
Male 5-14 years
Male 15-44 years
Male 45-64 years
Male 65-74 years
Male 75-84 years
Male 85+ years
2.354
1.000
0.913
1.373
2.531
3.254
3.193
Registered with
practice for 12
months+
1.000
The weight is
calculated as:
Female 0-4 years
Female 5-14 years
Female 15-44 years
Female 45-64 years
Female 65-74 years
Female 75-84 years
Female 85+ years
2.241
1.030
1.885
2.115
2.820
3.301
3.090
Registered with
practice in last 12
months
1.689
1.054 to the power of
the IMD Health
Domain score
associated with the
patient’s postcode)
5.21 Within the payment system, this adjustment can be calculated at patient level.
The payment software will calculate a single overall workload weight for each
patient, based on the three patient characteristics above, and then sum these
weights at practice level ready for use in the rest of the weighted patient
calculation. This process will not involve the transfer of patient-level data and
patient anonymity will not be compromised during this process.
5.22 The analysis on which this workload adjustment is based also controls for the
following variables but these will not be applied in the final formula:
•
number of GP principals
•
practice list size in thousands
•
region
•
the proportion of white population in the local area.
5.23 Further information about the components of the QRESEARCH models is
available at Appendix C.
We recommend a revised formula that includes a combined workload
adjustment based on QRESEARCH model 20.
Review of the General Medical Services global sum formula
17
ACRA’s work on additional needs
5.24 In England, the additional needs component of the index for primary medical
services utilises the SLLI and the SMR<65. These variables are the same as
those used in the current global sum formula and while there is commonality of
approach they are applied to different populations (see Chapter 4). Adoption of
the QRESEARCH approach recommended above will represented a
divergence from the current ACRA approach. In addition, the primary medical
services index uses slightly different age/sex weightings and does not include a
weighting for patients living in nursing and residential homes.
5.25 In October 2006, ACRA commissioned further work on additional needs for the
overall allocation formula. This work forms part of ACRA’s work programme,
which should be finalised by the autumn of 2007.
The potential for a supplementary adjustment to the formula
Consultation length and home visit adjustment
5.26 One of the limitations of the QRESEARCH model is the lack of consideration of
consultation length and the extra length of time for home visits. (QRESEARCH
measures home visits, but not the length of time taken for these.) This could be
addressed with a separate formula adjustment that involves the use of
alternative data sources.
5.27 A finalised set of weights for this adjustment are not yet available, but Appendix
E sets out how these will be calculated. Essentially, an estimated average
length of consultation including any home visit effect is derived for each of the
age-sex bands used elsewhere in the formula and then converted into a relative
weight for each age-band. These estimates are produced using estimates of
the average length of consultations (excluding home visits) by age-sex band,
data on home visits as a proportion of total consultations by age-sex band and,
estimates of the average length of home visits including travel time.
5.28 Once the relative consultation length and home visit weights for each age-sex
band are finalised, the consultation length and home visit adjustment index can
be calculated for each practice as follows:
Step 1: For each age-sex band, multiply the practice registered population in
that band by the relevant relative consultation length and home visit
weight.
Step 2: Sum these results across all age-sex bands to produce practice
pre-normalised consultation length and home visit adjusted weighted
patients.
Step 3: Apply the standard formula methodology for normalisation to calculate
practice normalised consultation length and home visit adjusted
weighted patients.
Step 4: Divide practice normalised consultation length and home visit adjusted
weighted patients by practice unweighted list size to give practice
consultation length and home visit adjustment index.
18
Review of the General Medical Services global sum formula
5.29 The status of this methodology is currently provisional, whilst data are obtained
and analysis is conducted to finalise the weights as per the methodology
described in Appendix E. Modelling results presented in this report reflect an
interim version of this adjustment based on currently available data.
5.30 An adjustment for consultation length would also act as a proxy for consultation
complexity which is not otherwise reflected in the formula. We could not
develop a specific adjustment for consultation complexity due to data difficulties
particularly around implementation.
We recommend a revised formula that includes a consultation length and
home visit adjustment.
Workload factors not covered by the formula
QOF prevalence
5.31 We considered models that used QOF prevalence to predict workload. This
could be done using either patient-level QOF data or practice-level QOF data.
However, as QRESEARCH suggest, the models using practice-level QOF data
produce counter-intuitive results:
“Our analyses at patient level produce intuitively acceptable results
that we believe fairly reflect the effects of the variables that we have
entered into the models. The use of practice-level data has, by
contrast, produced some results that are perverse and likely to lead
to considerable controversy. For example it appears that
consultation rates fall with increasing prevalence of some chronic
diseases.” (QRESEARCH Final Report, page 56)
5.32 QRESEARCH recommended models using patient-level QOF data. However,
models using patient-level QOF data are difficult to implement as the National
Health Application and Infrastructure Services (NHAIS) system used to apply
the global sum formula (known as the Exeter system) cannot link patient-level
QOF data to all other patient-level characteristics used in the formula. This has
also meant that the data to project the distributional impact of models using
patient-level QOF data on practice payments have not been available.
5.33 The FRG and Plenary agreed to exclude consideration of QOF prevalence from
the models. However, they agreed that Connecting for Health should be asked
to review the technical changes required to provide patient-level QOF data in
future years, to allow the option of using these data in the future.
We recommend that Connecting for Health should review the technical
changes required to provide patient-level QOF data in future years.
Review of the General Medical Services global sum formula
19
Patients living in nursing and residential homes
5.34 While QRESEARCH derived a binary patient-level surrogate variable for
nursing home/residential home status by looking for groups of four or more
elderly patients living in a single address as nursing/residential home status is
not routinely recorded, it was not sufficiently robust for inclusion in analyses.
5.35 We found that there was a negligible effect on the distribution of weighted
patients from the removal of the nursing and residential homes index on GMS
practices. The small effect of removing the nursing and residential homes
adjustment reflects the low proportion of nursing and residential home patients
for most practices.
5.36 There was also a risk of double counting involving the nursing and residential
homes adjustment under the current formula. Both an age-sex adjustment and
a nursing and residential homes adjustment were included, but neither
controlled for the other. As there is a correlation between the two adjustments,
it is likely that there is some degree of double counting in the current formula.
For example, the higher workload in the elderly identified by the age-sex
adjustment will include the higher workload associated with living in a nursing or
residential home, which is also included separately in the nursing and
residential homes adjustment.
5.37 It is also possible that some of the workload associated with patients in nursing
and residential homes would be picked up by the consultation length and home
visit adjustment.
Ethnicity
5.38 The chosen QRESEARCH model also controlled for, but did not directly include
ethnicity (expressed as the proportion of white population in an area as
individual ethnic status is not routinely recorded). The results of the
QRESEARCH analysis suggested that consultation rates decreased as the
percentage of white population decreased. We interpreted the negative effect of
ethnicity on workload as evidence of unmet needs in non-white groups and we
agreed that it would be inappropriate to reduce practice payments on the basis
of ethnicity. Therefore, ethnicity was not included directly in the formula,
although it was controlled for in the underlying model to prevent the negative
effect contaminating the effect of the other variables correlated with ethnicity.
5.39 In their additional needs formula, ACRA similarly included ethnicity in their
regression model but did not use it to create the needs index.
Patients who speak a different language from their health care professional
5.40 We considered the potential for including an adjustment for patients speaking a
different language to their GP or health care professional. The justification of
such an adjustment is to take into account:
•
possible communication difficulties, which can increase consultation length
and workload
•
possible financial costs incurred, such as the hiring of a translator.
5.41 To isolate the effects of communication problems on practice workload and
cost, we would need either:
•
20
direct information on the number of patients speaking a different language
to their primary health professional at practice level
Review of the General Medical Services global sum formula
•
use of a proxy such as ethnicity.
5.42 Direct information on the number of patients speaking a different language to
their primary health professional is not currently recorded on any database.
Even if direct information were secured, we would still have problems in
devising an appropriate weight from the data and applying to the global sum
payments system. The later would require that the NHAIS system collect more
practice data thus increasing the data collection burden on all practices.
5.43 Because of the problems with obtaining direct information, we considered the
use of a proxy, for example ethnicity. Ethnicity could be linked directly to
workload or cost, but it would be more appropriate to try and link ethnicity as a
proxy to communication difficulties (and then communication difficulties to
workload or cost).
5.44 A base methodology for this approach has been used in the PCT allocations
formula which used the “Family and Working Lives Survey” (1996) to link
country of birth to communication difficulties, and established proportions of
people with difficulties for each country of birth.
5.45 However, even if there is potential to attribute area statistics to estimate
numbers with communication difficulties by practice, it would still be necessary
to establish the workload or cost weights for language difficulties for use in the
formula. Additionally, it would be difficult to isolate the specific “different
language to the primary health care professional effect” on practice workload
and/or cost.
5.46 We therefore agreed that it would be impossible to implement a new factor,
other than at local level.
Review of the General Medical Services global sum formula
21
6
Review of cost factors – variations in labour costs
6.1
In addition to considering workload, the global sum formula must also reflect
differences in relative costs of service delivery across practices. The formula
options we recommend do this through the use of two groups of cost
adjustments – labour cost adjustments and isolation/rurality adjustments.
6.2
We considered three adjustments, that address different aspects of labour cost
differences:
6.3
•
staff market forces
•
GP market forces
•
the additional costs of recruitment and retention that may be necessary to
attract GPs to practices in relatively deprived areas.
The staff Market Forces Factor adjustment reflects differences in the cost of
staff employed across practices whilst the other adjustments reflect appropriate
differences in the net income of the GP partners across practices. We
recommend that the revised formula should not include a GP Market Forces
Factor adjustment – this is described in paragraph 6.20 below.
Staff Market Forces Factor
6.4
The current global sum formula makes an adjustment for staff market forces
and an outline of the research underlying this staff MFF adjustment is included
at Appendix A. This is based on the same research as the MFF adjustment
used in the overall resource allocation formula.
6.5
The adjustment is given a weighting of 48%, as this was the average value of
staff costs as a proportion of the global sum equivalent prior to the introduction
of the new GMS contract. This weighing is subject to change based on the
availability of data, such as that from HM Revenue and Customs.
6.6
The current staff MFF adjustment was developed by the Institute for
Employment Research (IER) at the University of Warwick and is based on the
New Earnings Survey Panel Dataset (NESPD) for 2001/03. This is updated
periodically by ACRA which is currently reviewing the MFF, as described below.
6.7
The staff MFF adjustment index for each practice may be calculated as follows:
Step 1: Calculate practice pre-normalised staff MFF weighted patients using
the formula below.
practice prenormalised staff MFF
weighted patients
practice
= list size
Staff MFF for practice
* location from University
of Warwick reseach
Step 2: Apply the standard formula methodology for normalisation to calculate
practice normalised staff MFF weighted patients.
22
Review of the General Medical Services global sum formula
Step 3: Apply a weighting to reflect that the staff MFF adjustment should only
be applied to the proportion of global sum payments deemed to cover
staff costs using the following formula:
48% weighted
practice normalised
staff MFF weighted
patients
=
practice
unweighted list
size
* 0.52
+
practice
normalised staff
MFF weighted
patients
* 0.48
Step 4: Divide 48% weighted practice normalised staff MFF weighted patients
by practice unweighted list size to give the practice staff MFF
adjustment index.
6.8
We decided that no improvements to the current methodology for the staff MFF
were available other that than that being investigated by ACRA. Until ACRA’s
recommendations about the MFF are available, we recommend that the
existing methodology for the staff MFF adjustment is maintained and that the
adjustment should be periodically updated to use the latest available data on
staff MFF weights as they become available.
We recommend a revised formula that includes a staff Market Forces
Factor, calculated using the existing methodology.
ACRA and Market Forces Factors
6.9
The Market Forces Factor (MFF) is a major element in the national resource
allocation formula. It is designed to compensate for unavoidable differences in
the costs of delivering healthcare in different locations due to external market
forces. From April 2006 the MFF has been paid directly to provider trusts.
ACRA has been reviewing this component of the formula which at present is
based on work carried out by the University of Warwick on New Earnings
Survey (now Annual Survey of Hours and Earnings, or ASHE) data. The same
index is used for the HCHS and the practice staff element of the Primary
Medical Services Component.
6.10 The two methods appropriate for a supply cost adjustment of this sort are a
general labour market (GLM) approach or a specific-cost approach (SCA). The
former is the basis for the current (Warwick) adjustment and the current global
sum formula also uses this approach. Research projects on each methodology
were commissioned by ACRA in late 2005. That on the GLM was carried out by
a team led by the Health Economics Research Unit (HERU) at Aberdeen
University and that on the SCA by a consortium led by Crystal Blue Consulting,
and including economists from York and City Universities.
6.11 The review of the MFF forms part of ACRA’s work programme, which should be
completed by the autumn of 2007. ACRA will then make recommendations to
Ministers on possible changes to the weighted capitation formula.
Review of the General Medical Services global sum formula
23
GP recruitment and retention
6.12 One criticism of the current global sum formula is that it potentially does not
fully capture the higher relative costs of service delivery in deprived areas over
and above the rurality and staff MFF adjustments. In particular, it does not allow
for the additional costs of recruitment and retention that may be necessary to
attract GPs to practices in relatively deprived areas. It is possible that, on
average, there is a disamenity to working in deprived areas that may lead to
recruitment and retention difficulties unless compensated through GP earnings
potential.
6.13 This is a distinct issue from the impact that deprivation has on service demand,
which is captured in the review of workload factors. The workload factors reflect
expected hours of work. Disamenities lead to a requirement for higher hourly
wage rates.
6.14 We commissioned a team led by Professor Bob Elliott of the Health Economics
Research Unit at the University of Aberdeen to:
•
establish whether recruitment and retention difficulties increase the relative
costs of service delivery in deprived areas, and therefore whether an
adjustment for this in the formula would be appropriate
•
if appropriate, establish the data requirements for such an adjustment at
practice level and examine whether they could currently be met
•
if appropriate, develop a methodology that would allow the formula to be
potentially adjusted for recruitment and retention difficulties in deprived
areas.
6.15 The report of this research is presented in the supporting technical document.
This concludes that there appears to be a case for a Cost of Recruitment and
Retention (CORR) adjustment.
Recommended adjustment
6.16 The Aberdeen research quantified the relationship between indicators of GP
recruitment and retention difficulties and possible explanatory variables. By
essentially assessing the effect on GP recruitment and retention difficulties of
GP relative earnings compared to other indicators, such as a measure of
deprivation, the change in GP relative earnings necessary to compensate other
factors, such as deprivation, can be estimated.
6.17 We recommend the adoption of a new CORR adjustment based on Aberdeen’s
‘Narrow comparator group, replacement vacancies, model 4’.
6.18 From this, a Cost of Recruitment and Retention adjustment index for each
practice can be calculated as follows:
Step 1: Calculate practice pre-normalised CORR weighted patients using the
formula below.
practice prenormalised
CORR
weighted
patients
24
=
practice
list size
exponential of
*
narrow
comparator
SSWD
LLTI
* 0.21
Review of the General Medical Services global sum formula
+
100
* 0.73
Step 2: Apply the standard formula methodology for normalisation to calculate
practice normalised CORR weighted patients.
Step 3: Apply a weighting to reflect that the CORR adjustment should only be
applied to the proportion of practice payment that makes up GP
income using the following formula:
42% weighted
practice normalised
CORR weighted
patients
=
practice
unweighted list
size
* 0.58
+
practice
normalised
CORR weighted
patients
* 0.42
(42% is the proportion of practice payment that makes up GP income,
as calculated prior to the new GMS contract. This figure could be
updated to reflect the most recent figures from the GP Earnings and
Expenses enquiry 2004/05.)
Step 4: Divide 42% weighted practice normalised CORR weighted patients by
practice unweighted list size to give the practice CORR adjustment
index.
6.19 In the CORR adjustment formula:
•
The narrow comparator SSWD is the narrow comparator Standardised
Spatial Wage Differential (SSWD). The SSWD is an estimate of the average
differences in wage attributable to geographical location after taking account
of age, gender, industry type and occupation. It is a measure of the relative
wage premium earned in a given geographical area, compared to other
areas, by private sector employees.
It is termed ‘narrow’ as it is based on the narrow comparator group of
occupations defined in line with that suggested in the 2002 DDRB report,
paragraph 1.81: DDRB “use[s] solicitors, actuaries, chartered engineers,
accountants, taxation professionals, and architects in both the public and
private sectors, and from the public sector [it] use[s] civil servants, members
of the armed forces and university academics.” The SSWD includes only
private sector employees since this sector has flexible wage setting.
•
The LLTI is the average Limiting Long Term Illness ratio for the practice. It
is utilised in the current additional needs adjustment. It is a standardised
and relative measure of the proportion of the population that report having a
Limiting Long Term Illness. It is widely used to indicate the chronic health
needs associated with deprivation.
We recommend a revised formula that includes a Cost of Recruitment and
Retention adjustment.
Review of the General Medical Services global sum formula
25
GP Market Forces Factor
6.20 Some people raised concerns that the current formula does not include a
Market Forces Factor for GPs.
6.21 We considered the introduction of a GP MFF adjustment alongside the Cost of
Recruitment and Retention (CORR) adjustment. However, the CORR
adjustment considers a comparator Standardised Spatial Wage Differential
(SSWD) as an indicator of geographical differences in relative earnings. This is
too similar to an MFF indicator to consider both, in the Global Sum Formula,
without a risk of double counting. Therefore, if a GP MFF adjustment were to be
included, the CORR adjustment would need to be modified to exclude
consideration of the comparator SSWD.
6.22 We acknowledged that there are some arguments in favour of the inclusion of a
GP MFF adjustment. On average, a GP MFF and the current London
adjustment are worth a similar amount (equivalent to 4%–5% increase in
weighted patients) to London practices. A GP MFF adjustment could therefore
potentially negate the need for a crude off-formula London Adjustment. This
would better reflect differences within London, prevent cliff edge effects at the
London boundary and, reflect similar pressures existing in other Metropolitan
areas. It also would mean that the global sum formula would better reflect
private-sector earning patterns.
6.23 However, there are also arguments against a GP MFF adjustment. It would be
inconsistent with the CORR research, which found that GP recruitment and
retention problems bear little relationship to private sector pay comparisons. It
also goes against the precedent set by the decision not to use a GP MFF in the
existing formula when there is no clear justification to do so.
6.24 We therefore recommend that a GP MFF adjustment should not be included in
the revised global sum formula.
We recommend that the revised formula should not include a GP Market
Forces Factor adjustment.
26
Review of the General Medical Services global sum formula
7
Review of cost factors – isolation and rurality
costs
7.1
In addition to unavoidable labour costs, practices also face differing costs of
service delivery due to rurality. We considered practices’ costs of isolation and
rurality.
The cost of unavoidable smallness
7.2
Economies of scale mean that bigger practices will have lower expenses per
head but the current formula does not take account of the economies of scale
effects for practices which unavoidably have a small list size. This is often, but
not exclusively associated with practices in isolated rural areas. We therefore
sought to include a Cost of Unavoidable Smallness (CUS) adjustment in the
formula. This is a combination of an economies of scale adjustment and an
isolation criteria.
Commissioned research
7.3 We know that smaller practices face higher average costs per patient because
they benefit less from economies of scale – that is, they spread their fixed costs
over a smaller number of patients so that the average cost per patient is higher.
However, it would be inappropriate to reward or incentivise small practices
without recognising the cause. The global sum formula should only adjust for
the economies of scale effects of unavoidable small scale and it should not
compensate practices for being small when the geographical dispersion of the
population and other causes of higher costs do not warrant this. Otherwise this
would clearly create a perverse incentive for practices to keep only a small
registered list – although this effect is offset as the formula is a capitation
formula.
7.4
We commissioned Deloitte to carry out research into a formula adjustment to
compensate for the cost effects of unavoidable small scale. This focused on the
effects of serving geographically dispersed populations.
7.5
The approach of this research was to:
7.6
•
establish the relationship between practice list size and practice costs per
patient
•
obtain information on the geographical distribution of primary care practices
and of populations requiring access to primary medical care services
•
establish a criteria for unavoidably small practices based on the trade-off
that would exist between the opportunity for practices to benefit from
economies of scale and patient travel costs, under a hypothetical merger of
nearby practices
•
determine what indicators are statistically linked to unavoidable smallness
•
use this information to advise on a economies of scale adjustment that is
qualified by a condition that the small size of practices is unavoidable.
Deloitte’s final report is available in the supporting technical document.
Review of the General Medical Services global sum formula
27
Structure of the CUS adjustment
7.7 We recommend that a Cost of Unavoidable Smallness adjustment, based on
the results of the Deloitte research, is based on two main components:
•
an economies of scale adjustment
•
an isolation criteria.
Economies of scale adjustment
7.8 The Deloitte research established the relationship between list size and
expenses per patient that exists for practices with small list sizes. This can be
used to produce an initial economies of scale adjustment for practices using the
following procedure:
Step 1: Calculate an initial economies of scale adjustment for the practice
using the following formula:
initial
economies of
scale
adjustment
=
35.15664
+
1…
list size
*
34573.21
50.65016
Step 2: The relationship between practice expenses per patient and list size
implied by this formula only holds for smaller practices. Eventually, the
economy of scale effect of practice list size is exhausted and practice
expenses per patient stabilise rather than continue to fall as list size
increases. This is reflected by constraining the economies of scale
adjustment to a minimum value of 1. Any practice with an adjustment
below 1 is then credited with a weight of 1.
The inflation of weights below 1 will affect practices with a list size of
2,232 or more. Only practices of 2,231 patients or fewer can benefit
from the cost of unavoidable smallness adjustment. However, at the
margin, the effect of this cut-off is negligible. The difference in
adjustment for a practice just below the cut-off is only very marginally
different to that of a practice just above the cut-off.
Step 3: For the most extremely small practices, this formula will produce very
high weights. We recommend that practices with an initial economies
of scale adjustment of greater than 2.5 are flagged. This provides the
option to confirm the data relating to such practices and, if
appropriate, replace the initial economies of scale adjustment with an
appropriate value.
Weights of 2.5 or greater are associated with practices with 377
patients or fewer.
28
Review of the General Medical Services global sum formula
Isolation criteria
7.9 Simply applying a formula adjustment based on the above economies of sale
adjustment would not achieve a CUS adjustment. It only examines the effect of
smallness on costs and does not address the question of whether or not small
practice sizes are avoidable. To do this an isolation criteria is applied that
qualifies the extent to a small practices can benefit from the economies of scale
adjustment based on the degree to which its smallness is unavoidable.
7.10 The Deloitte research identified that the key indicator of whether a small
practice was unavoidably small was the degree of isolation of the practice. This
was measured by the distance of the practice to the next nearest practice.
7.11 The Deloitte research focused on the road distance to the next nearest practice,
but implementing an adjustment based on road distance, rather than straightline distance would be difficult. Furthermore, in the range of distance to
practices that are material to the isolation criteria, there is a high positive
correlation between road distance and straight-line distance that allows a
reasonably reliable conversion between the two figures.
7.12 The Deloitte research suggested that at a distance to the next nearest practice
of more than 4km, in straight line terms, additional patient travel costs
associated with the absence of the practice will likely offset any economies of
scale benefit foregone by the practice being small. Likewise, it suggested that
at a distance to the next nearest practice of less than 2.5km, in straight line
terms, the “unavoidability” of practice smallness could be questioned. As such,
we agreed that the impact of the practice list size adjustment described above
should be scaled on the basis of distance to nearest practice.
7.13 From the Deloitte research an isolation criteria based on straight-line distance
to nearest practice can be derived. From this the practice isolation adjusted
economies of scale weight can be calculated from the following table of
scenarios. Essentially, if a practice is closer than 2.5km to its nearest practice
then it receives no benefit from the economies of scale adjustment, if a practice
is further than 4km to its nearest practice then it receives full benefit from the
economies of scale adjustment and, for practices between 2.5km and 4km from
their nearest practice, the benefit of the economies of scale is phased in. This is
described in Table 2.
Table 2: Calculation of practice isolation adjusted economies of scale
weight
Scenario
Practice closer than 2.5km to
nearest practice
Practice 4km of more from its
nearest practice
Practice between 2.5km and 4km
from its nearest practice
Practice isolation adjustment economies
of scale weight
Equal to 1
Full economies of scale adjustment
Equal to 1 plus:
Economies of scale adjustment minus 1
multiplied by
(Distance to nearest practice – 2.5km)/(4km –
2.5km)
Review of the General Medical Services global sum formula
29
7.14 In determining whether a practice is isolated, we agreed that branch surgery
locations should be ignored in the distance to closest practice calculation. This
is because:
•
if branch surgeries were considered the data requirements for the
adjustment would be much greater and there are issues around the
reliability of the current branch surgery data
•
consideration of branch surgeries would greatly complicate the methodology
for the measure at it would be necessary to account for what hours each
branch surgery was open and what range of services they provided
•
it seems inappropriate to penalise practices which improve rural access by
opening branch surgeries.
Calculating the final adjustment
7.15 This isolation adjusted economies of scale weight is converted into a CUS
adjustment as follows:
Step 1: Calculate practice pre-normalised cost of unavoidable smallness
weighted patients using the formula below.
practice pre-normalised cost
of unavoidable smallness
weighted patients
practice
= list size
practice isolation
* adjusted economies
of scale weight
Step 2: Apply the standard formula methodology for normalisation to calculate
practice normalised cost of unavoidable smallness weighted patients.
Step 3: Divide the practice normalised cost of unavoidable smallness
weighted patients by practice unweighted list size to give the practice
Cost of Unavoidable Smallness adjustment index.
7.16 While the other cost adjustments in the revised formula are weighted to reflect
the proportion of practice earnings that cover the relevant set of costs, we
recommend that this principle is not applied to this adjustment. This adjustment
is concerned with promoting the viability of unavoidably small practices and it
should be applied to the full practice list size as both practice expenses and
potential net income are relevant.
We recommend a revised formula that includes a Cost of Unavoidable
Smallness adjustment. This adjustment should be applied to the full
practice list size.
Rurality
Current adjustment
7.17 The current rurality adjustment is intended to reflect the uncontrollable
additional costs associated with the degree to which the area served is rural.
The impact of rurality on costs was modelled using HM Revenue and Customs’
30
Review of the General Medical Services global sum formula
information on GP expenses, and because it only applied to the expenses
element of GMS expenditure, the adjustment was given an overall weighting of
58%.
7.18 The variables that determine the current rurality adjustment are
•
the population density of the area the practice draws its patients from
•
the average distance of patients’ homes from the practice.
7.19 An outline of the research underlying this adjustment is available in Appendix A.
7.20 The new CUS adjustment compensates for the unavoidable costs of practices
that are necessarily small because of their isolated location. Some would argue
that a rurality adjustment should be applicable to practices irrespective of size,
similar to the current rurality adjustment.
7.21 We considered the cases for and against the inclusion of a rurality adjustment
in the revised formula. We were unable to make a recommendation about
whether or not a rurality adjustment should be included in the revised formula
and Plenary is required to make a decision about this. Our considerations are
presented below, together with a recommendation about the form that a rurality
adjustment should take if adopted.
7.22 The delivery costs associated with rurality are a distinct issue from the impact
that rurality has on service demand which was considered as part of the review
of workload factors.
Advantages and disadvantages of including a rurality adjustment in the revised
formula
7.23 It can be justified that a rurality adjustment should be included in the revised
formula because:
•
the Carr-Hill analysis showed that rurality was associated with increased
expenses per capita after allowing for list size, so a rurality adjustment could
be applicable to practices irrespective of size.
7.24 The reasons why a rurality adjustment should not be included in the revised
formula include:
•
the validity of the Carr-Hill rurality analysis, which was based on data that
preceded the introduction of the nGMS contract. It is possible that the
higher expenses of rural practices are specific to previous payment
mechanisms
•
a perception that the current adjustment is not particularly well targeted and
that it benefits leafy suburbs as well as the most rural practices because it is
a continuous function of density and distance
•
it may appear unnecessarily complicated to include two adjustments that
address seemingly similar issues.
7.25 Additionally, the rurality adjustment includes patients’ average distance to
practice. Our health, our care, our say: a new direction for community services
states that patients should have more choice to register with the practice most
convenient for their particular needs and circumstances. Patients may choose
Review of the General Medical Services global sum formula
31
to register with practices some distance from their home, and therefore average
distance to practice may no longer be a good measure of rurality.
Recommended form of possible Carr-Hill rurality adjustment
7.26 We considered a number of forms that a Carr-Hill rurality adjustment could
take:
•
the current Carr-Hill adjustment with no changes
•
an updated Carr-Hill adjustment based on more recent (2003/04) HMRC
data
•
an updated Carr-Hill adjustment based on more recent (2003/04) HMRC
data, which also removes average distance to practice as a rurality
indicator.
7.27 We concluded that if Plenary were to include a rurality adjustment, it should be
an updated version that retains the average distance to practice because:
•
we identified no strong argument to suggest that the latest available data
should not be used
•
the immediate impact of increased patient choice of practice on the validity
of average distance to practice is unclear.
7.28 This suggests that if adopted the rurality adjustment will be calculated as
follows:
Step 1: Calculate practice pre-normalised rurality weighted patients using the
formula below.
practice prenormalised rurality
weighted patients
practice
= list size
practice
* average patient
distance to
practice
0.078
practice average
* population
density
-0.02926
Step 2: Apply the standard formula methodology for normalisation to calculate
practice normalised rurality weighted patients.
Step 3: Apply a weighting to reflect that the rurality adjustment should only be
applied to the proportion of practice payment that makes up practice
expenses using the following formula:
58% weighted practice
normalised rurality
weighted patients
=
practice
unweighted
list size
* 0.42
+
practice normalised
rurality weighted
patients
* 0.58
58% is the proportion of practice payment that makes up GP income, as
calculated prior to the new GMS contract. This figure could be updated to
reflect the most recent figures from the GP Earnings and Expenses enquiry
2004/05
Step 4: Divide 58% weighted practice normalised rurality weighted patients
by practice unweighted list size to give the practice rurality index.
32
Review of the General Medical Services global sum formula
7.29 We recommend that if this adjustment were to be adopted, then it should be
reviewed in the short term. This would address the relatively poor research
base for this formula adjustment compared to the others which are new or
revised, take advantage of new data on both post nGMS expenses and rurality
and, would allow the assessment of the impact of increased patient choice on
the validity of average patient distance to practice indicator.
7.30 However it should be reiterated that we were unable to make a clear
recommendation whether or not to include such an adjustment due to the lack
of evidence and rationale to support its inclusion.
We are unable to recommend whether or not a rurality adjustment should
be included in the revised global sum formula. If a rurality adjustment is
adopted, we recommend that it should be an updated version of the
current rurality adjustment and that this adjustment should be reviewed in
the short term.
Review of the General Medical Services global sum formula
33
8
Combining the formula adjustments
8.1
Currently, the separate adjustments that make up the global sum formula are
multiplied together to generate the aggregate formula adjustment. We termed
this the product method.
8.2
We considered other options for combining the separate adjustments. The
majority of the alternatives involve averaging the formula adjustments in some
way, rather than multiplying them together. We concluded that these options did
not reflect the way that differences, across practices, in relative workload and
relative unit costs interact to produce differences in total resource requirements.
Moreover, they do not reflect how the weights have been estimated in the
empirical research.
8.3
As an example, take a simple formula with a single workload adjustment and a
single cost adjustment. Take two practices (A and B) and say practice A’s
workload adjustment index and a cost adjustment index are both equal to 1. If
practice B has 10% greater workload per patient and 10% greater unit costs
then its workload adjustment and cost adjustment indices will both be equal to
1.1. Multiplying the adjustments together suggests that practice B has a total
resource need premium of 21%. This is intuitively consistent with doing 10%
more work at 10% greater unit cost. In contrast, averaging the adjustments
would only give practice B a 10% total resource need premium. This does not
fully reflect the extra workload and costs of practice B.
8.4
Variations on the averaging method that do not suffer from this problem are
possible, but they produce very similar results to the product method and
increase the complexity of the formula.
8.5
We therefore recommend that the product method should be retained.
We recommend a revised formula in which the separate adjustments are
multiplied together to generate the aggregate formula adjustment.
34
Review of the General Medical Services global sum formula
9
The recommended formula
9.1
Taking into account the research undertaken and evidence available, we
recommend that the revised global sum formula should include the following
components:
9.2
•
workload adjustment (Chapter 5)
•
consultation length and home visits adjustment (Chapter 5)
•
staff Market Forces Factor (MFF) adjustment (Chapter 6)
•
Cost of Recruitment and Retention adjustment (CORR) adjustment
(Chapter 6)
•
Cost of Unavoidable Smallness (CUS) adjustment (Chapter 7)
•
(possibly the) rurality adjustment (Chapter 7).
The formula adjustment for each practice will be calculated as follows:
Step 1: Calculate practice pre-normalised overall weighted patients by
multiplying together the above components of the formula together
with the practice list size.
Step 2: Apply the standard formula methodology for normalisation to
calculate practice normalised overall weighted patients.
Review of the General Medical Services global sum formula
35
Comparison of the current formula with the recommended formula
without the rurality index
9.3
The modelling presented in Appendix F shows the projected distributional
impact of the recommended formula without the rurality index compared to the
current global sum formula (A guide to the projected distributional impact tables
can be found at Appendix I.) It suggests that it is anticipated that:
•
the change in weighted patients for GMS practices would range from -30%
to +65%. Excluding the 1% most extreme practices (0.5% at each extreme),
the range would be -19% to +29%
•
there would be significant redistributive effects across practice cohorts, as
summarised in Tables 3 and 4
•
compared to the current global sum formula, this formula would on average
tend to increase the weighted capitation share of urban practices, practices
with high additional needs, practices with high proportions of new
registrations, practices with low proportions of patients in nursing and
residential homes, practices with low proportions of elderly patients, London
practices and practices in spearhead PCTs.
Table 3: Redistribution across cohorts under the recommended formula
without the rurality index compared to the current global sum formula
(quartile based)
Cohort Category
Number of GPs
List size
Population density
Staff MFF
SLLI
SMR<65
New registrations
Nursing and residential home
patients
Proportion of patients aged over 65
Average % change in weighted patients
Lowest quartile of
Highest quartile of
practices
practices
+3%
-1%
+2%
-1%
-7%
+8%
-3%
+6%
-3%
+3%
-3%
+3%
-1%
+7%
+5%
-2%
+8%
-4%
Note: The practice groupings based on number of GPs and list size are not quartile
based. The low grouping for number of GPs covers single-handed practices
and the high grouping covers practices with 6 or more GPs. The low grouping
for list size covers practices with fewer than 2,001 patients and the high
grouping covers practices with 10,001 practice or more.
Table 4: Redistribution across cohorts under the recommended formula
without the rurality index compared to the current global sum formula
(membership based)
Cohort Category
London practices
Practices in Spearhead PCTs
36
Average % change in weighted patients
Members
Non-Members
+8%
-2%
+3%
-1%
Review of the General Medical Services global sum formula
Comparison of the current formula with the recommended formula with
the rurality index
9.4
The modelling presented in Appendix G shows the projected distributional
impact of the recommended formula with the rurality index compared to the
current global sum formula. It suggests that it would be anticipated that:
•
the change in weighted patients for GMS practices would range from -19%
to +83%. Excluding the 1% most extreme practices, the range would be
11% to +28%
•
there would be significant redistributive effects across practice cohorts, as
summarised in Tables 5 and 6
•
compared to the current global sum formula, this formula would on average
tend to increase the weighted capitation share of urban practices, practices
with high proportions of new registrations, practices with low proportions of
nursing and residential home patients, practices with low proportions of
elderly patients and London practices.
Table 5: Redistribution across cohorts under the recommended formula
with the rurality index compared to the current global sum formula
(quartile based)
Cohort Category
Number of GPs
List size
Population density
Staff MFF
SLLI
SMR<65
New registrations
Nursing and residential home
patients
Proportion of patients aged over 65
Average % change in weighted patients
Lowest quartile of
Highest quartile of
practices
practices
+1%
-1%
+1%
0%
-3%
+4%
-1%
+3%
-1%
+1%
0%
+1%
-1%
+4%
+3%
-2%
+5%
-2%
Note: The practice groupings based on number of GPs and list size are not quartile
based. The low grouping for number of GPs covers single-handed practices
and the high grouping covers practices with 6 or more GPs. The low grouping
for list size covers practices with fewer than 2,001 patients and the high
grouping covers practices with 10,001 practice or more.
Table 6: Redistribution across cohorts under the recommended formula
with the rurality index compared to the current global sum formula
(membership based)
Cohort Category
London practices
Practices in Spearhead PCTs
Average % change in weighted patients
Members
Non-Members
+4%
-1%
+1%
-1%
Review of the General Medical Services global sum formula
37
Comparison of the two recommended formulae
9.5
The modelling presented in Appendix H shows the projected distributional
impact of the recommended formula with the rurality index compared to the
recommended formula without the rurality index. It suggests that it would be
anticipated that:
•
there would be small effects across practice cohorts, as summarised in
Tables 7 and 8
•
compared to the recommended formula without the rurality index, including
the rurality index would on average tend to increase the weighted capitation
share of rural practices, practices with low additional needs, practices with
higher proportions of elderly patients and practices outside of London.
Table 7: Redistribution across cohorts under the recommended formula
with the rurality index compared to the recommended formula without
the rurality index (quartile based)
Cohort Category
Number of GPs
List size
Population density
Staff MFF
SLLI
SMR<65
New registrations
Nursing and residential home
patients
Proportion of patients aged over 65
Average % change in weighted patients
Lowest quartile of
Highest quartile of
practices
practices
-1%
+1%
-1%
+1%
+5%
-4%
+2%
-2%
+2%
-2%
+3%
-2%
+1%
-3%
-2%
+1%
-3%
+2%
Note: The practice groupings based on number of GPs and list size are not quartile
based. The low grouping for number of GPs covers single-handed practices
and the high grouping covers practices with 6 or more GPs. The low grouping
for list size covers practices with fewer than 2,001 patients and the high
grouping covers practices with 10,001 patients or more.
Table 8: Redistribution across cohorts under the recommended formula
with the rurality index compared to the recommended formula without
the rurality index (membership based)
Cohort Category
London practices
Practices in spearhead PCTs
38
Average % change in weighted patients
Members
Non-Members
-4%
+1%
-1%
+1%
Review of the General Medical Services global sum formula
The impact of the two recommended formulae – some case studies
9.6
The case studies below demonstrate the impact of each recommended formula
on various practices. As per the other modelling, this is compared to the results
of applying the current global sum formula on the same date and may therefore
not reflect the payments that practices are currently receiving.
The impact of the recommended formulae upon various practices
Practice A
This GMS practice has a list size of 4,038. It is in a rural environment. It has a
relatively old mix of patients but low additional needs/morbidity. Its list turnover is
also comparatively low (1.1% below average). Under the current formula its overall
practice weighting is 1.11 and total number of weighted patients is 4,474. Under the
revised formula, its weighting would fall to 1.09 and its weighted list to 4,401. A
formula excluding the updated Carr-Hill rurality adjustment would decrease this still
further to 0.95 (4,250). The main driver for the modest decrease would appear to be
a lower benefit from rurality.
Practice B
This PMS practice has a list size of 2,987. It is in central London. It has a relatively
young mix of patients coupled with low additional needs/morbidity. Its list turnover is
2.6% higher than average. Under the current formula its overall practice weighting is
0.74 and total weighted patients 2,196. Under the revised formula its weighting would
rise to 0.81 and its weighted list to 2,419. A formula excluding the updated Carr-Hill
rurality adjustment would increase this still further to 0.90 (2,688). The main driver for
the increase would appear to be the QRESEARCH workload adjustment.
Practice C
This inner London GMS practice has a list size of 5,361. It has a relatively young mix
of patients but high additional needs. Its list turnover is above average but not high
for London. Under the current formula its overall practice weighting is 1.14 and its
total number of weighted patients is 6,099. Under the revised formula, its weighting
would rise substantially to 1.20 and its weighted list to 6,433. A formula excluding the
updated Carr-Hill rurality adjustment would increase this still further to 1.28 (6,862).
The main reason for this increase is the Cost of Recruitment and Retention
adjustment.
Practice D
This urban PMS practice is in a blue collar area of moderate means. Its age-sex
index is above average (1.02) and it has high additional needs. It has a practice list of
6,709. Its weighting is 1.21 under the current formula but this would drop dramatically
to 1.06 (1.07 excluding the updated Carr-Hill rurality adjustment). Its number of
weighted patients would fall from 8,033 to 7,112 (7,179). The main cause of this
seems to be the QRESEARCH workload adjustment and its relationship to additional
needs in the current formula.
Practice E
This large market town PMS practice has a list of 16,567, with an above average
age-sex mix but low additional needs reflecting its comparative affluence. Its rurality
index is 1.047. Under the current formula its overall practice weighting is 1.02 and its
total number of weighted patients is 16,898. Under the revised formula its weighting
would fall to 0.96 and its weighted list to 15,904. A formula excluding the updated
Carr-Hill rurality adjustment would decrease this to 0.92 (15,242). The main reason
for this decrease is the negative effect of the Cost of Recruitment and Retention
adjustment.
Review of the General Medical Services global sum formula
39
Practice F
This GMS practice has a relatively young population of 8,595 with high additional
needs. Under the current formula its overall practice weighting is 1.17 and its total
number of weighted patients is 10,056. Under the revised formula, its weighting
would rise to 1.24 and its weighted list to 10,658. A formula excluding the updated
Carr-Hill rurality adjustment would increase this to 1.30 (11,174). The main reason for
this increase is the combined effect of the QRESEARCH workload adjustment
(1.063) and the Cost of Recruitment and Retention adjustment (1.27).
Practice G
This GMS practice of 6,843 patients is in an affluent area with a very young patient
population although with surprisingly high additional needs (an index of 1.095). Its
overall practice weighting under the current formula is 0.77 and its weighted list size
is 5,235. Under the revised formula, its weighting would rise to 1.17 and its weighted
list to 8,006. A formula excluding the updated Carr-Hill rurality adjustment would
increase this to 1.18 (8,075). The reason for this substantial increase is the
QRESEARCH workload adjustment (1.231).
Practice H
This inner city GMS practice has a list of 2,430 patients with low age-sex weighting
and moderate additional needs. Its overall practice weighting is 0.90 and its weighted
list is 2,197. Under the revised formula its weighting would increase to 0.95 and its
weighted list to 2,309. A formula excluding the updated Carr-Hill rurality adjustment
would increase the overall weighting to a positive value (1.04) and the weighted list to
2,527. The main reason for this increase is the Cost of Recruitment and Retention
adjustment.
Practice I
This GMS practice of 1,839 patients is in an area of moderate means with mostly
blue collar occupations on the outskirts of a conurbation. Its age-sex weighting is
below average and it has above average additional needs. Its rurality weighting is
also above average at 1.016. Its overall practice weighting under the current formula
is 0.98 and it would fall to 0.96 under a revised formula, with a weighted list of 1,765.
A formula excluding the updated Carr-Hill rurality adjustment would decrease the
weighting to 0.94 and the weighted list to 1,729. The reason for this small fall is the
QRESEARCH workload adjustment which is less than but close to the equivalent
components of the current formula (age-sex, additional need and new registrations).
Practice J
This GMS practice of around average practice size (5,576 patients) is in a suburban
area of moderate means with older residents and blue collar occupations. Its age-sex
weighting is above average and additional needs comparatively high. Its overall
practice weighting under the current formula is 1.09 and its weighted list size is
6,103. Under the proposed formula its weighting would decrease to 1.07 and its
weighted list to 5,966. A formula excluding the updated Carr-Hill rurality adjustment
would increase the weighting to 1.10 and the weighted list to 6,134. The reason for
the initial decrease is the QRESEARCH workload adjustment, offset by the Cost of
Recruitment and Retention adjustment.
Practice K
This small rural GMS practice of 1,744 patients has a current weighting of 1.08,
which would fall to 1.07 under the revised formula (0.99 excluding the updated CarrHill rurality adjustment). Its number of weighted patients would fall from 1,884 to
1,866 (1,727). As a beneficiary of the Cost of Unavoidable Smallness adjustment
(1.085), it is spared the impact of a fall in the uprated Carr-Hill rurality adjustment
(1.143 to 1.076).
40
Review of the General Medical Services global sum formula
SECTION C: TAKING THE RECOMMENDATIONS
OF THE REVIEW FORWARD
10
Implementation issues
10.1
Our main remit was to examine the existing payments formula. However,
during the course of the review we acknowledged that there were a number of
additional issues which warranted consideration as they had an impact upon
the implementation of our recommendations and that would effect the
equitable distribution of resources and/or the financial stability of practices.
10.2
The main issues which WE considered outwith the technical content of the
formula were:
10.3
•
the London adjustment
•
the Minimum Practice Income Guarantee (MPIG)
•
growing populations.
During the course of its discussions, we recognised that a technical or
analytical approach to these issues was difficult as these elements of the
original GMS contractual agreement were underpinned by political decisions
at that time. However, we acknowledged the reasons for such decisions and
explored their impact upon the various stakeholders.
The London adjustment
10.4
GMS practices in London currently receive an off-formula “London
adjustment” of £2.18 per unweighted patient. This adjustment is equivalent to
a premium of approximately 4% and was set aside to recognise the
potentially destabilising effects of the current formula on practices in London.
10.5
We agreed that it would be preferable if improvements to the global sum
formula meant that the London adjustment was no longer required. This
would make the calculation of global sum payments for London practices
more sensitive to differences within London and would reduce cliff-edge
effects at the London adjustment boundary.
10.6
On the basis of improvements to the formula, we recommend that the London
adjustment should be discontinued. Appendices F and G show the projected
distributional impact of options for the revised global sum formula.
We recommend that the London adjustment should be discontinued if the
revised global sum formula is adopted.
Review of the General Medical Services global sum formula
41
The Minimum Practice Income Guarantee (MPIG)
10.7
The MPIG was established in April 2003, after financial modelling confirmed
that the introduction of the global sum payments would lead to a reduction in
the basic income of a majority of practices. Under the MPIG agreement, any
practice that would lose out as a result of the introduction of the global sum
arrangements received a guarantee that its allocation would reflect its
previous level of income from those aspects covered by the global sum.
Under this arrangement a “correction factor” would be paid to a practice to
make up the difference between its global sum (as determined by the
formula) and its global sum equivalent income.
10.8
Every quarter, practices’ global sums are recalculated, allowing for changes
in list size and patient characteristics, which may increase or decrease
accordingly. On top of this, MPIG practices continue to receive the correction
factor at the level set from the original calculation at the beginning of 2004/05.
This payment is only one component of practice income, which may also
include payments for QOF and enhanced services.
10.9
The April 2003 agreement included the provision that correction factor
payments would be annually increased for inflation by the same level as the
global sum. However, as part of the agreed 2006/07 contract revision, it was
agreed that future uplifts to the global sum should seek to reduce the reliance
upon correction factor payments. In practice, this would be achieved by
applying a differential level of uplift between the MPIG and the global sum –
thus recognising correction factor payments as a transitional support
mechanism.
10.10 We discussed the reasons why the need for the MPIG had arisen and it was
felt that this was due to the total global sum financial envelope being smaller
than the historic funding levels against which comparison was being made.
Intuitively, it would be reasonable to expect 50% of practices to gain and 50%
of practice to lose if a new allocation formula had been applied to exactly the
same funding envelope. This was not the case as significantly more practices
would have lost resources from this element of the new contract without the
MPIG agreement.
10.11 While many had perceived this imbalance to be a failing of the Carr-Hill
formula, we can confirm that the primary reason was due to the level of the
original global sum resource envelope. We agreed though that the historic
constitution of MPIG and correction factor payments prevented the equitable
distribution of the resource envelope between practices based on the agreed
formula. We also recognised that the financial stability of individual practices
was vital.
10.12 Any new agreement implementing recommendations from our report would
again raise the issue of financial stability at practice level. We based our
conclusions and financial modelling on the principle that a revised formula
would be applied to the same resource envelope and therefore broadly result
in a 50/50 split of winning and losing practices. The potential options for
managing such a transition would be:
(i)
42
establish new practice-level MPIGs based on practice income
immediately prior to the implementation of a new formula (e.g. 31 March
2007). Under this approach:
Review of the General Medical Services global sum formula
(ii)
•
no practice would lose any income through the implementation of a
new formula
•
additional national funding would be required to fund the new
guarantee and provide financial support to potentially losing
practices
•
an agreement would need to be reached regarding the nature of any
such absolute or transitional guarantee
retain previous MPIG levels from the beginning of 2004/05. Under this
approach:
•
some practices may see a reduction in current income but they
would still be protected at the level of their current MPIG
•
an agreement would need to be reached regarding the impact upon
correction factor payments where practices historically reliant upon
MPIG would gain from the implementation of a new formula.
Growing populations
10.13 We discussed how the revised formula would ensure patients’ access to
primary medical care services, particularly how practices would be
incentivised to register new patients.
10.14 As part of its multivariate analysis QRESEARCH considered the impact of
new registrations on practice workload. It explored the effects of new patients
up to 24 months after registration and found only a modest effect of new
registrations on workload after the first 12 months. Hence the final workload
adjustment includes a variable that maintains the current definition of newly
registered patients – that is, those patients who joined the practice in the last
twelve months. The new weight credited to newly registered patients is
greater than the previous weight.
10.15 Both recommended formulae would on average tend to benefit practices with
a high proportion of new registrations compared to the current global sum
formula.
10.16 Additionally, as indicated in the White Paper Our health, our care, our say: a
new direction for community services, the Department of Health will “ask NHS
Employers to consider the case for establishing an Expanding Practice
Allowance for practices that have open lists which are growing significantly
and that offer extended opening hours.”
Review of the General Medical Services global sum formula
43
11
Data recommendations
11.1 This chapter collates and summarises all of our recommendations relating to
data.
We recommend that the collection of the following could be beneficial to
future developments of the formula:
• the ability to link patient-level QOF data to the National Health
Application and Infrastructure System used to apply the global sum
formula (also known as the Exeter system)
• workload data for patients living in nursing and residential homes
• direct information on the number of patients speaking a different
language to their primary health care professional
• actual patient-level socioeconomic data.
Workload survey
11.2 We identified that a new workload survey could usefully inform future work and
we have already commissioned a survey through the Health and Social Care
Information Centre. The last large scale workload survey of general practice
was conducted in 1992/93.
11.3 The new survey will collect up to date information on:
•
the distribution of work in general practice to inform the formula review
•
skill mix changes, particularly the contribution to workload made by practice
staff
•
practice workload as well as that of individual GPs and staff.
11.4 The survey will produce results for the UK but not at country level. It will cover
GMS and PMS practices and any PCTMS GPs attached to any of the practices
in the survey will be included.
11.5 The results of the 2006/07 survey will be reported to Plenary in March 2007.
44
Review of the General Medical Services global sum formula
12 Impact of the formula review upon the devolved
administrations
12.1 While we included membership from each of the devolved administrations, the
findings of the review may impact upon each country differently.
Scotland – report from the Scottish Allocation Formula Review Group
12.2 Concurrent with the UK Formula Review, the Scottish Executive set up its own
review of the Scottish GMS Allocation Formula. They developed a new model
which uses patient-level data to estimate workload, measured by the number of
diagnosis codes per patient contact, based on information regarding age, sex
and deprivation measures (derived from the individuals’ home post code). The
model includes an adjustment for new registrations and care home residents
and is comparable in scope to the model that was developed by QRESEARCH
for the UK FRG.
12.3 If adopted, the new Scottish workload model will replace the old workload
model as well as the current adjustment for morbidity and life circumstances.
12.4 The Scottish formula does not include a factor to adjust for possible costs of
retention and recruitment and the review group did not propose to add such a
factor. The group held the view that problems with retention and recruitment
were best dealt with at a Health Board level.
12.5 Options for the rurality and remoteness adjustment are being developed.
Diseconomies of scale appear to increase per-patient expenditure in small rural
practices and this factor is being considered to provide a fair adjustment to
necessarily small practices.
Wales
12.6 The current global sum formula in Wales is the same as in England. Wales also
has a similar percentage of practices in receipt of the correction factor.
However, there is a belief that a higher proportion of practices in Wales, in
areas of high needs and deprivation, are not appropriately funded because the
redistributive effects of the current formula have been stifled. Wales will
therefore want to be sure that our proposals, if adopted, will help address these
particular problems.
Northern Ireland
12.7 During the original development of the global sum formula, Northern Ireland
established a working group to test, under statutory equality obligations, each
element of the formula, and recommend evidence-based refinements where
necessary to avoid or minimise adverse impact across any of the equality
dimensions. This led to a hybrid formula in Northern Ireland as follows:
•
a Northern Ireland-specific age-sex workload curve (based on consultations
from the Continuous Household Survey and a length adjustment based on a
Scottish study)
Review of the General Medical Services global sum formula
45
•
the UK adjustment for patients in nursing and residential homes
•
the UK adjustment for list turnover
•
a Northern Ireland-specific additional needs index (derived from a database
comprising multiple deprivation indicators, health and socio-economic
indicators from the Northern Ireland Health and Social Wellbeing Survey,
census 2001 data and Standard Mortality Ratios)
•
a Northern Ireland-specific rurality index incorporating economies of scale
(developed from modelling of GP payments using census 2001 data,
mortality and health factors and social security indicators)
•
the UK staff Market Forces Factor.
12.8 With regard to the formula review, the intention in Northern Ireland is to wait for
the recommendations of the FRG and then test each element using NI data and
recommend refinements where necessary, either to meet statutory equality
obligations or to better reflect GMS workload in Northern Ireland.
12.9 As Northern Ireland do not have an equivalent data source to QRESEARCH,
they cannot develop revised elements based on the new methodology
proposed by FRG. They would simply test their current adjustments against our
recommendations. Where possible, they would update their current
adjustments with the most recent data available.
46
Review of the General Medical Services global sum formula
Glossary (including commonly used acronyms and
abbreviations)
ACRA
Advisory Committee on Resource Allocation
coefficient
In a formula, a coefficient is a constant value (e.g.
0.75 or 4) by which a variable (e.g. MFF or SLLI)
is multiplied.
correlation coefficient
This is a number between -1 and 1 which
measures the degree to which two variables are
linearly related. If there is a perfect linear
relationship with a positive slope between the two
variables the correlation coefficient is 1. If there is
a perfect linear relationship with a negative slope
between the two variables the correlation
coefficient is -1. A correlation coefficient of 0
means there is no linear relationship between the
variables.
controlling for (but not including)
certain factors in the formula
In multiple regression, an independent (or
explanatory) variable is said to be controlled for
but not included if it is used in the calculations to
determine the relationship between a dependent
and a number of independent variables, but is not
used subsequently when the derived formula is
applied. This occurs when a variable has a
significant explanatory effect (and therefore
should be taken into account in the calculation of
the coefficients of the other explanatory variables)
but it is either counter intuitive to include the
variable when the regression formula is applied,
considered to duplicate an adjustment made
elsewhere, or is otherwise deemed inappropriate.
convert into an index
This reduces the absolute values of a set of
numbers into a series that reflects their relative
differences. For example, the mean of the set
might be given an index value of 1. A number
25% greater than the mean would have the index
value 1.25 and a number 25% less than the mean
would have the index value 0.75.
CORR
Cost of Recruitment and Retention
CUS
Cost of Unavoidable Smallness
FRG
Formula Review Group
GPRD
General Practice Research Database
HCHS
Hospital and Community Health Services
Review of the General Medical Services global sum formula
47
MFF
Market Forces Factor
MPIG
Minimum Practice Income Guarantee
multivariate regression analysis
This is a statistical procedure that determines the
relationship between a variable and two or more
explanatory variables.
normalisation
Normalisation describes a number of mechanisms
necessary to produce appropriate relative weights
across all practices in England whilst constraining
weighted patients to population totals. See also
Appendix D.
SLLI
Standardised Limited Long-Standing Illness
SMR<65
Standardised Mortality Ratio for those aged under
65
TAG
Technical Advisory Group, a subsidiary of ACRA
48
Review of the General Medical Services global sum formula
Appendix A: Carr-Hill resource allocation formula
This document has been taken from Annex D of “Investing in General
Practice: New GMS contract 2003”. Subsequent negotiations (including
the introduction of MPIG) have caused some changes to the
implementation of the formula. For example, the London Adjustment is
now worth approximately £9 million and the population base for the
formula is the registered list.
Introduction
A.1 This appendix presents the core findings from the analysis used to derive the
Carr-Hill resource allocation formula for the new GMS contract. This will be
used to allocate the global sum and related payments on the basis of the
practice population, weighted for factors that influence relative needs and costs.
The proposed formula includes the following components:
•
an adjustment for the age and sex structure of the population, including
patients in nursing and residential homes
•
an adjustment for the additional needs of the population, relating to
morbidity and mortality
•
an adjustment for list turnover
•
adjustments for the unavoidable costs of delivering services to the
population, including a staff Market Forces Factor and rurality.
A.2 The formula differs from those previously developed for resource allocation
purposes in two key respects. First, the majority of the formula is to be applied
to the four countries within the United Kingdom. Secondly, the formula will be
applied to practice populations, rather than Primary Care Organisation
populations.
A.3 The approach to the formula follows that established elsewhere in the field of
resource allocation: namely, expressing relative need in cost terms. This
involves establishing an age-sex cost curve, estimating the additional resource
implications of additional needs, and then adjusting for other factors that affect
the cost of delivering services. Given the difficulties of collecting data in this
area, a large number of different exercises have been carried out. This annex
summarises the approaches and the main results.
Age-Sex Workload Curve
A.4 The basis of any allocation formula for a set of services is the population
served. For General Medical Services in the UK this is defined by those
registered on the lists of each general practitioner. Whilst those lists are welldefined (although there are well-known problems over list inflation – see
paragraph 31 below) there is no routine dataset that provides the basis for
showing the entire workload generated by different age-sex groups on the
practice list.
A.5 Consultations can take place in the surgery, the patient’s own home or in a
nursing or residential care home. There is no single data source adequately
Review of the General Medical Services global sum formula
49
covering general practice consultations in all of these environments. Whilst
there are routine data available on consultations in the surgery, there are only
limited data on home visits and no systematic data on nursing and residential
home consultations. Consequently, they have to be estimated separately, with
separate databases. Consultations in the Surgery: Analysis of General Practice
Research Database (GPRD).
A.6 The analyses of surgery consultations have been based on the General
Practice Research Database (GPRD). It includes data from 240 practices,
including eight from Northern Ireland, ten from Wales and twelve from Scotland.
The GPRD has individual level consultation frequencies and information on
duration has been recorded for all members of the primary care health team in
most of the practices since 1999.
A.7 The whole dataset covers a period from 1996 to August 2002 and contains
details of 99 million consultations. However, prior to 1999 the vast majority of
the computer systems in these 240 practices did not record when patient files
were opened and closed. Of the whole dataset, 69% (68 million) of
consultations were time-stamped.
A.8 It is important to emphasise that the GPRD material refers to "consultations",
but these are simply instances of a patient's computer file being accessed. So a
receptionist checking an appointment or a computer manager doing data
checks will both count as ‘consultations’. They are more properly referred to as
‘file openings’.
A.9 It can be argued that the relative GP workload associated with different patient
groups may be approximated by the times for which the files were open and
that these data may also be used to estimate consultation rates. The obvious
objections are that the opening of a patient's computer file may not denote a
consultation and that the length of time for which the file is opened may not
reflect the workload associated with the event being recorded. An example of
the latter would be the retrospective entering of details of a home visit. The
retrospective entering of home visit details is unlikely to reflect the full workload
of home visits, which are often longer than surgery visits and also have an
associated travel time. For this reason, home visits have been treated
separately. The age-sex workload adjustment is presented in Table 1 below.
Table 1: Mean total time for which a patient file was opened in minutes
per year: all staff weighted by staff input cost
Males
0–4
5–14
15–44
45–64
65–74
75–84
85+
50
Average time
per person
50.38
12.69
14.64
30.65
53.81
58.62
29.16
Females
Ratio to male Average time
5–14
per person
3.97
46.31
1
13.35
1.15
31
2.42
48.28
4.24
62.38
4.62
63.06
2.3
29.14
Ratio to male
5–14
3.65
1.05
2.44
3.8
4.92
4.97
2.3
Review of the General Medical Services global sum formula
A.10 The counter-intuitive reduction in average time (and hence relative workload)
for the most elderly patients could be explained by a higher proportion of very
short file openings for those groups to record other information about the
patient. Furthermore, these may refer to home or care home visits, the details
of which are recorded post hoc and do not reflect the actual workload
generated by home visits.
Home Visits
A.11 As discussed above, the GPRD does not adequately record home visits.
Although the file may be opened in relation to home visits, this is likely to be for
a relatively short period as the information is added after the home visit has
taken place. This will therefore not reflect the full workload impact of the home
visit.
Length of the Home Visit
A.12 On the whole, a home visit tends to generate a higher workload than a surgery
consultation, as the consultation itself is often longer and a home visit also has
an associated travel time. According to the 1992–93 GP workload survey the
average length of a home visit, including travel time, is 25.2 minutes. General
Medical Practitioners' Workload Survey 1992-93. London: Department of
Health, November 1994. Paragraph 6.8 and Table D5.
Variation by Age and Sex
A.13 The most extensive data available on home visits are those from the Morbidity
Statistics for General Practice 4 (1991/92). A clear ‘J’ shaped relationship
between age and home visiting rates is apparent for both males and females.
This is set out in Table 2 below.
Table 2: Home visit rates per 1000 patient years
Male
Female
0–4
498
454
5–15
126
128
16–24
56
150
25–44
63
163
45–64
136
200
65–74
506
608
75–84
1331
1628
85+
2792
3081
A.14 Applying the home visit length to these weights, and combining with the
consultations in surgery produces an age-sex workload index, which is set out
in Table 3. The formula uses different age-sex workload indices in Scotland and
these are shown in Table 4.
Table 3: Age-sex Workload Index (males aged 5–14 = 1): UK less
Scotland
Male
Female
0–4
3.97
3.64
5–14
1
1.04
15–44
1.02
2.19
45–64
2.15
3.36
65–74
4.19
4.9
75–84
5.18
6.56
85+
6.27
6.72
Table 4: Age-sex Workload Index (males aged 5–14 = 1): Scotland
Male
Female
0–4
2.51
2.24
5–14
1
1.102
15–24
1.153
2.371
25–44
1.272
2.524
45–64
1.823
2.714
65–74
2.672
3.067
75–84
3.207
3.342
Review of the General Medical Services global sum formula
85+
3.468
3.340
51
Nursing and Residential Homes
A.15 Two separate surveys were carried out to analyse the relative workload
generated by patients in nursing and residential homes. One was directed to
nursing and residential homes, to generate information on the frequency of
consultations, and the other to GPs, looking at time spent in nursing and
residential home consultations. Overall, patients in nursing and residential
homes generate more workload than patients with otherwise similar
characteristics who are not in nursing and residential homes. This is mainly due
to the fact that all nursing and residential home consultations involve travelling
time. The workload factor applied to patients in nursing and residential homes is
1.43.
List Turnover
A.16 Areas with high list turnover often have higher workload, as the patients in their
first year of registration in a practice tend to have more consultations than other
patients with otherwise similar characteristics.
A.17 The impact of list turnover was analysed using the GPRD, which contains data
on patient registration date. The results of this analysis indicate that the
average time in ‘consultation’ is some 40 to 50 per cent higher for patients in
their first year of registration in the practice compared with other patients. The
rate varies across age and sex bands, with young males having the strongest
additional effect. Rather than create an entirely separate age-sex cost curve for
new registrations, the average uplift of a factor of 1.46 will be applied to all new
registrations.
Additional needs
A.18 As well as the impact on practice workload generated by differing age and sex
groups, the impact of indicators of mortality and morbidity on consultation
frequency has been modelled.
A.19 This has been modelled using the Health Survey for England data between
1998 and 2000. The survey asks participants whether they have had a GP
consultation in the past two weeks, and if yes the number of such consultations.
The survey also includes information on age, sex, geographic location and a
range of socio-economic indicators. These were combined with a range of other
small area level explanatory variables, including census variables, mortality
rates, and supply variables. The analysis was conducted at ward level, and
wards were excluded where there were less than five observations in the ward.
This reduces the sample size to 2,404 wards.
A.20 Of the variables tested, Standardised Limited Long-Standing Illness (SLLI) and
the Standardised Mortality Ratio for those aged under 65 (SMR<65) were found
to be significant and the best at explaining variations in workload over and
above age and sex. They are related to workload by the following formula:
Practice list adjusted for list inflation * ((48.1198 + (0.26115 * SLLI) + (0.23676 *
SIR<65)) scaled back to the UK population (see paragraph 32)
A.21 As mentioned in paragraph 5.15, special provision is being made for Scotland.
The evidence shows that although the Standardised Mortality Ratio helps to
explain variations in additional need in Scotland, other factors perform better in
combination with it than Standardised Limited Long-Standing Illness. Thus in
52
Review of the General Medical Services global sum formula
Scotland, the explanatory variables used in addition to Standardised Mortality
Ratio for those aged under 65 (SMR<65) are:
•
unemployment rate
•
elderly people on income support (aged 65 or more)
•
households with two or more indicators of deprivation.
Unavoidable Costs
A.22 As well as the impact on workload of practice characteristics, it is also
necessary to analyse the impact on costs. Practices are likely to face differing
costs of delivering a service depending on location and structure. Within the
global sum, we believe there to be three main causes of this: market forces,
rurality and practice size.
Staff Market Forces Factor
A.23 The aim of the staff MFF component is to reflect the geographical variation in
staff costs that practices will incur. The Market Forces Factor (MFF) adjustment
will be used to compensate for this. The MFF was developed by the University
of Warwick, using the same methodology as that used for general NHS
allocations.
A.24 The staff MFF is based on the latest three years of the New Earnings Survey
Panel Dataset. The latest three years of data are used and estimates are given
in terms of three year averages. The analysis uses individual earnings of fulltime employees aged 16–70 in the private sector whose pay is not affected by
absence. Regression analysis is then carried out to isolate the impact of
geographical area on costs, controlling for the effect of other factors such as
age, sex, industry and occupation.
A.25 The results have then been smoothed to prevent cliff edge effects between
neighbouring areas among the 173 zones used in the analysis. This reflects the
fact that the labour market pressures in one area are likely to be influenced by
those of its neighbours. This element of the formula has been given a weighting
of 48%, as this is the average proportion of the global sum accounted for by
practice staff expenses. The adjustment does not apply to general practitioners
or non-staff expenses.
A.26 The equivalent earnings dataset for Northern Ireland was not amenable to
similar analysis. The MFF for Northern Ireland outside Belfast has therefore
been taken as the average between Scotland and Wales, outside of Edinburgh
and Cardiff respectively, whilst the MFF for Belfast has been taken to be the
average between Edinburgh and Cardiff.
Rurality
A.27 The rurality of the practice population is also likely to have an influence on the
costs of delivering services. The impact of rurality on costs has been modelled
using Inland Revenue information on GP expenses. This is under the
assumption that rurality acts as an unavoidable cost impacting on the expenses
associated with delivering services. Because it only applies to the expenses
element of GMS expenditure, this adjustment has been given an overall
weighting of 58%.
A.28 The dataset contained information on around 20,000 GPs across England,
Scotland and Wales. This was aggregated to practice level using practice
Review of the General Medical Services global sum formula
53
identifiers. The impact of population density (persons per hectare in the wards
from which a practice draws its patients) and dispersion (average distance of
patients to practice, in 100 metre units) indicators was modelled against GP
expenses. The modelling controlled for other factors such as the age and
deprivation structure of the population, market forces, and other characteristics
such as dispensing status and list size. The nature of the relationship between
rurality and expenses is such that the equation explaining it works best when
the data are transformed into logarithms. The estimated co-efficients on the
logarithms of population density and dispersion reflect the ‘unavoidable cost’ of
the rurality and remoteness of the area. Thus, for given population densities
and dispersions, variations in the unavoidable costs associated with rurality are
explained by the following formula:
Practice list adjusted for list inflation * ((0.05*log average distance) – (0.06 *log
population density)) scaled back to the UK population (see paragraph 32)
A.29 For Scotland, the formula includes an additional component relating to
economies of scale (see paragraph 30 below) for a limited number of practices.
Other issues
Practice Size (economies of scale)
A.30 Small practices can be expected to incur disproportionately high expenses due
to their inability to secure economies of scale. Many costs (particularly those
associated with premises) are not easily disaggregated and must be incurred
irrespective of practice size. Using data from the Inland Revenue on Schedule
D expenses by practice list size, a strong diseconomies effect was detected at
low list sizes. This effect would, if imported wholesale into the formula, have
had a dramatic and potentially undesirable impact on the distribution of the
global sum, notwithstanding that by definition, it would be applied only to
expenses (58% of spend). The case for including the unavoidable costs
associated with diseconomies of scale in the formula was rejected in order to
avoid any perverse incentives for practices to disaggregate or to avoid
amalgamation.
London
A.31 As mentioned in paragraph 5.15, special provision is being made for London. A
sum (£53 million) has been set aside to recognise the potentially destabilising
effects of the implementation of the Carr-Hill formula. This sum will be
distributed amongst practices in London on the basis of practice populations
after adjustment for list inflation, unweighted for age, sex or additional need.
Combining the Adjustments
A.32 Each adjustment will generate a separate practice index, comparing the
practice score on the adjustment to the national average. The indices are then
simultaneously applied to the practice list (adjusted for list inflation – see
para.5.16).
A.33 This will produce the practice weighted population. The application of the
indices to all practices will produce an overall notional population which differs
from the actual UK population as estimated by the Office for National Statistics
(ONS). Weighted populations are adjusted so as to total to that ONS population
– a process known as normalisation.
54
Review of the General Medical Services global sum formula
Appendix B: Formula Review Group membership
Name
Organisation
Job title
Philip Grant
(Chairman)
NHS Employers
Director of Finance Adviser and Negotiator
Richard Armstrong
Department of Health
Head of Primary Medical Care Contracting
Eleanor Babbington
BMA - Health Policy & Economic Research Unit
Research Analyst
Andrew Clapperton
Department of Health
Head of PMCC Finance
Marie-Claire Demblon
NHS Employers
Secretariat
Francis Dickinson
Department of Health
Economic Adviser
Dr Stewart Drage
GPC
Negotiator
Jon Ford
BMA
Head of Health Policy and Economic Research
Taryn Harding
(Project Manager)
NHS Employers
Project Manager, Primary Care Contracting
Dr Daniel Hinze
Scottish Executive Health Department
Economic Adviser
Felicity Kenn
GPC
Secretariat
Dr David Love
SGPC/GPC
Joint Chairman SGPC
John Maingay
GPC
Secretariat
Andrew Marshall
Department of Health
Economic Adviser
Ivan McMaster
DHSS&PS
Assistant Director, Primary Care
David Melbourne
Heart of Birmingham Teaching PCT
Director of Finance
Dr Stephen Morris
Health Economics Research Group, Brunel University
Reader
Penny Murray
DHSS&PS - Information and Analysis Directorate
Deputy Principal Statistician
David Notman
Scottish Executive Health Department
Policy Adviser – GMS Branch
Ian Reeve
Langbaurgh PCT
Director of Finance and Operations
Katharine Robbins
Information Centre for health and social care
Head of Primary Care Statistics
Dave Roberts
Information Centre for health and social care
Unit Manager - Prescribing Support Unit
Dr Eric Rose
GPC
Negotiator
Martyn Shipp
Welsh Assembly Government
Community, Primary Care and Health Services
Policy Directorate
Prof Matt Sutton
Health Economics Research Unit – University of
Aberdeen
Professor of Health Economics
Dr Ian Trimble, OBE
Nottingham City PCT
GP and PEC Chair
Mike Vickerman
Department of Health
Senior Statistical Officer
Review of the General Medical Services global sum formula
55
Appendix C: Components of QRESEARCH models
With patient years denominator
C.1 Model 20 includes a “with patient years” denominator. This means that when
patient consultation rates are calculated, they are annualised so that they
reflect the proportion of the year that patients were registered with practices.
Deprivation indicator
C.2 QRESEARCH models incorporate the use of IMD deprivation or Townsend
measures. We recommended a workload adjustment based on QRESEARCH’s
model 20 which uses an IMD deprivation measure.
Index of Multiple Deprivation (IMD) – Health domain
C.3 The IMD health domain score identifies areas with relatively high weights of
people who die prematurely or whose quality of life is impaired by poor health
or who are disabled. It considers:
•
years of potential life lost
•
comparative illness and disability ratio
•
measures of emergency admissions to hospital
•
adults under 60 suffering from mood or anxiety disorders.
C.4 The larger the score, the more deprived an area is.
Townsend score
C.5 The Townsend score is a measure of multiple deprivation that is calculated for
an area of residence by combining four census variables:
•
percentages of households that are not owner occupied
•
percentage of households with no car
•
percentage of households with more than one person per room
•
percentage of persons that are unemployed.
C.6 The larger the score, the more deprived an area is.
C.7 QRESEARCH used a modified version of the Townsend score created from
2001 census variables in the same manner as 1991 Townsend scores.
56
Review of the General Medical Services global sum formula
Appendix D: Guide to normalisation in the global sum
formula
Normalisation and the global sum formula
D.1
D.2
There are a number of broad steps in the calculation of global sum formula
weighted patients:
•
calculate the PCT normalising indices
•
calculate practice weighted patients for each of the component
adjustments of the formula
•
calculate overall global sum formula weighted patients by combining the
formula components.
The process of normalisation is used, in some way, during each of these
stages.
Normalisation during the calculation of PCT normalising indices
D.3
The global sum formula is designed to reflect relative workload and relative
costs of service delivery across all practices in the country. However,
because of the delay involved in collating data for all practices nationally,
each quarter the global sum formula calculation only assesses each practice
relative to other practices in the same PCT.
D.4
This means a mechanism for indirectly assessing practice workload and costs
relative to practices in different PCTs is required. The PCT normalising
indices fill this need. They become an adjustment to the global sum formula
weighted patients, calculated within PCTs that reflect the relative workload
and costs of serving that PCT’s registered population compared to that of all
other PCTs in the country. This adjustment is one of the processes termed
normalisation.
D.5
For example, say the within PCT application of the global sum formula
suggests that a particular practice should be credited with 2.5% of its PCT’s
total weighted population and that PCT has a total registered population of
500,000 and a PCT normalising index of 1.1. That practice’s weighted list size
would be 13,750, i.e. 2.5% of 500,000 multiplied by 1.1.
D.6
To calculate the PCT normalising indices, all of the patient and practice
characteristics data that informs the global sum formula is compiled nationally
each quarter. The global sum formula is then run at a national level, directly
assessing the relative workload and costs of each practice compared to all
others nationally. Within this calculation, another process termed
normalisation is also employed.
D.7
At each stage of the calculation, practice weighted patients are normalised or
constrained to the national total registered population whilst maintaining the
relative practice shares of weighted patients.
Review of the General Medical Services global sum formula
57
D.8
For example, say the national registered population is 50m and the
application of some formula weighting suggests that half of these patients
should be, in terms of weighted patients and global sum payments, worth
double the ‘value’ of the rest. The initial application of these weightings would
suggest a national weighted population of 75m [(25m multiplied by a factor of
2) plus (25m multiplied by a factor of 1)].
D.9
Normalisation, in this context, constrains the national weighted population
back to the national registered population of 50m whilst maintaining the
relative weighting of patients. The ‘high value’ group of patients account for
two-thirds of the pre-normalised national weighted population i.e. (25m
multiplied by 2) divided by 75m. The ‘high value’ group of patients is then
credited with normalised weighted patients equal to two-thirds of the national
registered population i.e. 33.33m. Likewise the ‘low value’ group of patients is
credited with 16.66m normalised weighted patients.
D.10
This broad process with a more complicated set of relative weightings occurs
at practice level as part of the application of this specific use of normalisation.
D.11
This ensures that the impact of each of the formula adjustments is equal and
that no adjustment dominates the others. It also ensures that the weighting of
practice populations is a cost neutral exercise that redistributes capitation
payments rather than increasing or decreasing them.
D.12
Under this methodology, a practice’s normalised weighted patients for a
formula component is equal to its percentage share of the national total
weighted patients applied to the national registered population.
D.13
For each PCT, the ratio of the sum of its practices’ global sum formula
weighted patients to its total registered population is then its PCT normalising
index. These PCT normalising indices are then used to adjust the following
quarter’s global sum formula weighted patients calculated within PCTs.
Normalisation during the calculation of practice weighted patients for
each of the component adjustments of the formula
D.14
As described above, apart from the calculation of PCT normalising indices,
the global sum formula calculations are performed at PCT level.
D.15
During these PCT-level calculations, normalisation, in the sense of scaling
weighted patients back to a specified level whilst maintaining relative practice
shares, is used repeatedly to ensure that the impact of each of the formula
adjustments is equal, and no adjustment dominates the others, and that the
application of formula weightings is a purely distributive cost neutral exercise.
D.16
When calculating the practice indices for each of the formula components it is
used to constrain the PCT total weighted patients for that component back to
the PCT’s overall global sum formula weighted patients (calculated as its total
registered population multiplied by its PCT normalising index as calculated
using the previous quarter’s data).
D.17
Under this methodology, a practice’s normalised weighted patients for a
formula component is equal to its percentage share of the PCT total weighted
58
Review of the General Medical Services global sum formula
patients for that component applied to the PCT total registered population
multiplied by its PCT normalising index.
D.18
For example, say the application of weightings for a particular adjustment in
the formula suggests a particular practice has a pre-normalised weighted
patient total of 20,000 and that the PCT total pre-normalised weighted patient
total is 1m. If that PCT has a total registered population of 500,000 and a PCT
normalising index of 0.9 then the practice’s normalised weighted patient total
for that adjustment is 9,000 i.e. a 2.0% share (20,000 of its PCT’s 1m prenormalised weighted patients) of its PCT’s registered population of 500,000
multiplied by its PCTs normalising index of 0.9.
Normalisation during the calculation of overall global sum formula
weighted patients by combining the formula components
D.19
Normalisation, in the sense of scaling weighted patients back to a specified
level whilst maintaining relative practice shares, is used again in the
combination of the formula components to calculate overall global sum
formula weighted patients.
D.20
When combining practice list size and the practice indices for each of the
formula components to produce overall weighted patients, it is used to
constrain the PCT total overall global sum formula weighted patients to the
PCT total registered population multiplied by its PCT normalising index.
D.21
Under this methodology, a practice’s normalised overall global sum formula
weighted patients is equal to its percentage share of the sum of prenormalised overall global sum formula weighted patients for the practices in
its PCT applied to the PCT registered population multiplied by its PCT
normalising index.
D.22
For example, say the application of each of the formula’s component
adjustments to unweighted practice list sizes suggests a particular practice
has a pre-normalised weighted patient total of 25,000 and that the PCT total
pre-normalised weighted patient total is 1m. If that PCT has a total registered
population of 500,000 and a PCT normalising index of 1.2 then the practice’s
overall normalised weighted patient total is 15,000 i.e. a 2.5% share (25,000
of its PCT’s 1m pre-normalised weighted patients) of its PCT’s registered
population of 500,000 multiplied by its PCTs normalising index of 1.2.
Review of the General Medical Services global sum formula
59
Appendix E: Calculation of consultation length and
home visit adjustment weights
The general method for calculating the weights for a consultation length and home
visit adjustment is:
Step 1:
Estimate average consultation length by age-sex bands using the
length of file opening from GPRD (updated to use the latest data).
Step 2:
Estimate the proportion of consultations that are home visits by agesex band (which is available from QRESEARCH).
Step 3:
Increase the estimated average consultation lengths to reflect the
additional length of home visits, including travel time, weighted by the
proportion of consultations that are home visits. The preferred data
source for the length of home visits is the Workload Survey currently
underway.
Step 4:
Calculate relative consultation length and home visits weights for each
age-sex band by scaling the adjusted consultation lengths relative to
the value for males aged 5–14.
60
Review of the General Medical Services global sum formula
Appendix F - Projected distributional impact of the recommended formula
without the rurality index compared to the current global sum formula
Option:
Compared with:
Source data:
Revised Global Sum formula with no rurality adjustment
Current Carr Hill formula (baseline)
2006-07 Q2 payments data
Percentile
Overall
Overall
formula Average
formula
weighting: weighting: percent
change
option
baseline
All
1.006
1.011
0.49%
Min 0.5th
5th
-30% -18% -11%
10th
-8%
25th
-4%
50th
0%
75th
4%
90th
10%
95th 99.5th
15% 31%
Max
72%
GMS
PMS
1.007
1.005
1.006
1.017
-0.06%
1.21%
-30% -19% -12%
-28% -18% -10%
-9%
-7%
-4%
-3%
0%
1%
4%
5%
9%
11%
14%
15%
29%
32%
65%
72%
Number of GPs in practice
1
2 or 3
4 or 5
6 or above
1.006
1.010
1.012
0.994
1.030
1.009
0.996
0.977
2.70%
0.24%
-1.29%
-1.49%
-20%
-23%
-30%
-19%
-15%
-7%
-5%
-19% -14% -10%
-22% -13% -11%
-18% -12%
-9%
-2%
-4%
-5%
-5%
2%
0%
-1%
-2%
6%
5%
2%
2%
12%
10%
7%
5%
16%
14%
12%
10%
31%
28%
24%
25%
65%
55%
38%
58%
Practice list size
2,000 and under
2,001 to 5,000
5,001 to 10,000
10,001 and above
1.022
1.009
1.011
0.983
1.038
1.016
1.002
0.973
1.87%
1.02%
-0.57%
-0.71%
-20%
-26%
-30%
-19%
-16%
-7%
-5%
-19% -13%
-9%
-18% -13% -10%
-17% -10%
-8%
-3%
-3%
-5%
-4%
1%
1%
-1%
-1%
5%
5%
3%
2%
11%
10%
8%
5%
15%
14%
14%
12%
31%
27%
24%
30%
65%
41%
55%
58%
Population density
Least dense quartile
Quartile 2
Quartile 3
Most dense quartile
1.028
0.993
1.015
0.989
0.953
0.988
1.029
1.062
-6.99%
-0.35%
1.66%
7.58%
-30% -23% -17% -15% -11%
-13%
-9%
-6%
-5%
-3%
-10%
-9%
-5%
-4%
-2%
-8%
-5%
-2%
0%
3%
-7%
-1%
1%
6%
-3%
2%
4%
12%
0%
4%
7%
17%
2%
6%
10%
21%
12%
15%
36%
31%
65%
21%
55%
58%
Staff MFF
Lowest MFF quartile
Quartile 2
Quartile 3
Highest MFF quartile
1.045
1.032
0.995
0.959
1.014
1.014
0.984
1.013
-2.65%
-1.55%
-0.95%
5.82%
-30%
-26%
-18%
-17%
-22% -15% -12%
-19% -13%
-9%
-16% -12%
-8%
-13%
-7%
-4%
-7%
-5%
-4%
1%
-3%
-1%
-1%
5%
1%
2%
3%
11%
4%
5%
5%
17%
8%
8%
7%
20%
37%
18%
20%
27%
55%
65%
58%
34%
SLLI
Lowest SLLI quartile
Quartile 2
Quartile 3
Highest SLLI quartile
0.959
0.988
1.010
1.079
0.927
0.979
1.024
1.110
-2.79%
-0.53%
1.69%
3.08%
-30% -20% -15% -13%
-23% -19% -14% -10%
-20% -15%
-8%
-6%
-10%
-9%
-5%
-4%
-8%
-4%
-3%
-2%
-3%
-1%
1%
1%
2%
4%
5%
6%
6%
8%
11%
14%
10%
12%
15%
20%
22%
18%
29%
37%
65%
33%
55%
54%
SMR<65
Lowest SMR quartile
Quartile 2
Quartile 3
Highest SMR quartile
0.965
0.982
1.015
1.073
0.936
0.969
1.029
1.105
-2.63%
-0.90%
1.71%
3.26%
-30% -22% -15% -13%
-26% -19% -14% -10%
-20% -16%
-7%
-5%
-12%
-9%
-7%
-5%
-8%
-5%
-2%
-3%
-3%
-1%
1%
1%
2%
3%
5%
7%
7%
7%
10%
15%
9%
10%
15%
20%
21%
20%
26%
38%
65%
58%
33%
55%
New Registrations
Lowest New Reg quartile
Quartile 2
Quartile 3
Highest New Reg quartile
1.026
1.018
0.994
0.984
1.011
0.987
0.986
1.048
-1.35%
-2.84%
-0.57%
6.92%
-26%
-30%
-20%
-19%
-18% -10%
-8%
-21% -15% -12%
-18% -13% -10%
-13%
-5%
-2%
-5%
-7%
-5%
2%
-2%
-2%
0%
6%
1%
1%
3%
12%
5%
4%
7%
18%
7%
6%
10%
21%
24%
29%
22%
32%
65%
41%
54%
58%
Nursing and Residential Home residents
Lowest NRH quartile
0.970
1.012
Quartile 2
0.992
0.997
Quartile 3
1.018
1.002
Highest NRH quartile
1.040
1.014
4.64%
0.72%
-1.46%
-2.34%
-22%
-30%
-26%
-22%
-19%
-17%
-18%
-19%
-10%
-6%
-11%
-7%
-13% -10%
-14% -11%
0%
-3%
-5%
-6%
4%
0%
-1%
-2%
9%
4%
2%
1%
17%
9%
6%
5%
20%
13%
8%
7%
32%
28%
23%
24%
65%
41%
39%
36%
Patients age > 65
Lowest P(pat>65) quartile
Quartile 2
Quartile 3
Highest P(pat>65) quartile
0.937
0.994
1.019
1.064
1.009
0.999
0.998
1.017
7.72%
0.61%
-2.00%
-4.18%
-15%
-8%
-2%
0%
-19% -16%
-8%
-7%
-22% -18% -12% -10%
-30% -22% -16% -13%
2%
-3%
-5%
-8%
6%
0%
-2%
-3%
12%
4%
1%
0%
18%
7%
4%
3%
22%
10%
7%
4%
36%
24%
21%
16%
58%
54%
36%
65%
London
In London
Outside London
0.970
1.015
1.050
0.996
8.36%
-1.57%
-12%
-7%
-2%
0%
-30% -19% -13% -10%
3%
-5%
7%
-1%
13%
2%
18%
5%
21%
8%
30%
29%
34%
65%
Spearhead PCTs
In Spearhead PCT
Not in Spearhead PCT
1.051
0.984
1.077
0.970
2.67%
-1.03%
-20% -17%
-7%
-5%
-30% -19% -13% -10%
-3%
-5%
1%
-1%
6%
3%
15%
7%
20%
10%
36%
22%
55%
65%
GMS only
Review of the General Medical Services global sum formula
61
Appendix G - Projected distributional impact of the recommended formula
with the rurality index compared to the current global sum formula
Option:
Compared with:
Source data:
Revised Global Sum formula with updated rurality adjustment
Current Carr Hill formula (baseline)
2006-07 Q2 payments data
Percentile
Overall
Overall
formula Average
formula
weighting: weighting: percent
change
option
baseline
All
1.006
1.008
0.18%
Min 0.5th
-23% -11%
5th
-7%
10th
-5%
25th
-3%
50th
0%
75th
3%
90th
6%
95th 99.5th
10% 28%
Max
83%
GMS
PMS
1.007
1.005
1.005
1.011
-0.15%
0.62%
-19% -11%
-23% -11%
-7%
-6%
-6%
-5%
-3%
-3%
-1%
0%
2%
3%
6%
7%
10%
11%
28%
28%
83%
69%
Number of GPs in practice
1
2 or 3
4 or 5
6 or above
1.006
1.010
1.012
0.994
1.016
1.008
1.003
0.986
1.20%
0.08%
-0.72%
-0.61%
-13% -10%
-14% -11%
-19% -12%
-11%
-9%
-7%
-8%
-7%
-6%
-5%
-6%
-6%
-5%
-3%
-3%
-4%
-3%
0%
0%
-1%
-1%
4%
3%
1%
1%
8%
6%
4%
4%
12%
9%
9%
6%
33%
25%
24%
23%
83%
53%
42%
53%
Practice list size
2,000 and under
2,001 to 5,000
5,001 to 10,000
10,001 and above
1.022
1.009
1.011
0.983
1.032
1.010
1.005
0.980
1.19%
0.30%
-0.36%
-0.09%
-12%
-9%
-16% -12%
-19% -10%
-11%
-9%
-7%
-8%
-7%
-6%
-5%
-6%
-6%
-5%
-3%
-3%
-3%
-3%
0%
0%
-1%
-1%
4%
3%
2%
2%
9%
6%
5%
4%
13%
10%
9%
8%
33%
27%
21%
28%
83%
39%
53%
53%
Population density
Least dense quartile
Quartile 2
Quartile 3
Most dense quartile
1.028
0.993
1.015
0.989
0.999
0.987
1.015
1.022
-2.66%
-0.45%
0.22%
3.56%
-19% -13%
-12%
-9%
-11%
-9%
-12%
-7%
-9%
-6%
-6%
-5%
-8%
-5%
-5%
-3%
-5%
-3%
-3%
0%
-3%
-1%
-1%
3%
0%
2%
2%
6%
2%
4%
5%
11%
4%
6%
8%
14%
16%
14%
35%
27%
83%
21%
53%
53%
Staff MFF
Lowest MFF quartile
Quartile 2
Quartile 3
Highest MFF quartile
1.045
1.032
0.995
0.959
1.033
1.020
0.984
0.987
-0.94%
-0.98%
-0.97%
3.07%
-19% -13%
-13% -11%
-12% -10%
-12%
-9%
-8%
-8%
-7%
-4%
-7%
-6%
-6%
-3%
-4%
-4%
-3%
-1%
-2%
-1%
-1%
2%
1%
1%
1%
6%
4%
4%
4%
10%
8%
6%
5%
13%
37%
19%
17%
19%
53%
83%
53%
31%
SLLI
Lowest SLLI quartile
Quartile 2
Quartile 3
Highest SLLI quartile
0.959
0.988
1.010
1.079
0.949
0.982
1.012
1.090
-0.72%
-0.42%
0.38%
1.18%
-19% -12%
-16% -11%
-12% -10%
-12%
-9%
-8%
-7%
-7%
-7%
-6%
-6%
-5%
-5%
-4%
-3%
-3%
-3%
-1%
-1%
0%
0%
2%
2%
2%
3%
5%
5%
6%
9%
8%
7%
10%
13%
20%
13%
28%
36%
83%
25%
53%
50%
SMR<65
Lowest SMR quartile
Quartile 2
Quartile 3
Highest SMR quartile
0.965
0.982
1.015
1.073
0.961
0.974
1.017
1.079
-0.19%
-0.61%
0.38%
0.76%
-19%
-14%
-12%
-12%
-12%
-11%
-10%
-10%
-7%
-8%
-6%
-8%
-6%
-6%
-5%
-6%
-3%
-3%
-3%
-3%
-1%
-1%
0%
-1%
2%
2%
2%
3%
5%
5%
7%
9%
7%
7%
11%
13%
20%
14%
24%
37%
83%
53%
29%
53%
New Registrations
Lowest New Reg quartile
Quartile 2
Quartile 3
Highest New Reg quartile
1.026
1.018
0.994
0.984
1.015
1.000
0.988
1.019
-0.97%
-1.61%
-0.40%
3.96%
-16% -11%
-19% -13%
-13% -10%
-10%
-8%
-7%
-8%
-7%
-5%
-6%
-7%
-5%
-3%
-4%
-4%
-3%
0%
-1%
-2%
-1%
3%
1%
0%
2%
7%
4%
3%
5%
12%
6%
4%
6%
14%
27%
28%
17%
28%
83%
39%
50%
53%
Nursing and Residential Home residents
Lowest NRH quartile
0.970
0.993
Quartile 2
0.992
0.993
Quartile 3
1.018
1.010
Highest NRH quartile
1.040
1.023
2.56%
0.26%
-0.68%
-1.52%
-13%
-19%
-16%
-14%
-10%
-10%
-10%
-12%
-6%
-7%
-7%
-8%
-5%
-5%
-6%
-7%
-1%
-3%
-3%
-4%
2%
0%
-1%
-2%
5%
2%
1%
1%
10%
6%
4%
3%
14%
9%
7%
5%
32%
28%
21%
23%
83%
42%
38%
35%
Patients age > 65
Lowest P(pat>65) quartile
Quartile 2
Quartile 3
Highest P(pat>65) quartile
0.937
0.994
1.019
1.064
0.980
0.993
1.006
1.036
4.72%
-0.04%
-1.24%
-2.46%
-10%
-8%
-12%
-9%
-12% -10%
-19% -13%
-3%
-6%
-7%
-9%
-2%
-5%
-6%
-7%
1%
-3%
-4%
-5%
4%
0%
-2%
-3%
7%
2%
1%
0%
12%
4%
3%
2%
15%
6%
5%
3%
35%
26%
21%
16%
53%
50%
35%
83%
London
In London
Outside London
0.970
1.015
1.007
1.005
3.98%
-0.80%
-12%
-9%
-19% -11%
-4%
-7%
-3%
-6%
0%
-4%
4%
-1%
7%
1%
12%
4%
14%
6%
20%
28%
31%
83%
Spearhead PCTs
In Spearhead PCT
Not in Spearhead PCT
1.051
0.984
1.062
0.976
1.22%
-0.53%
-12% -10%
-19% -11%
-7%
-7%
-5%
-6%
-3%
-3%
0%
-1%
4%
2%
10%
5%
14%
7%
35%
18%
53%
83%
GMS only
62
Review of the General Medical Services global sum formula
Appendix H - Projected distributional impact of the recommended formula with the
rurality index compared to the recommended formula without the rurality adjustment
Option:
Compared with:
Source data:
Revised Global Sum formula with updated rurality adjustment
Revised Global Sum formula with no rurality adjustment
2006-07 Q2 payments data
Percentile
Overall
Overall
formula Average
formula
weighting: weighting: percent
with rur. change
w/o rur.
0.5th
-8%
5th
-5%
10th
-4%
25th
-2%
50th
-1%
75th
1%
90th
5%
95th 99.5th
8% 13%
Max
18%
All
1.011
1.008
-0.30%
Min
-24%
GMS
PMS
1.006
1.017
1.005
1.011
-0.09%
-0.58%
-13%
-24%
-8%
-8%
-5%
-6%
-4%
-4%
-2%
-3%
-1%
-1%
2%
1%
6%
4%
8%
7%
13%
13%
18%
18%
Number of GPs in practice
1
2 or 3
4 or 5
6 or above
1.030
1.009
0.996
0.977
1.016
1.008
1.003
0.986
-1.34%
0.11%
0.82%
1.09%
-13%
-9%
-8%
-8%
-8%
-8%
-8%
-8%
-6%
-5%
-5%
-4%
-5%
-4%
-3%
-3%
-3%
-3%
-2%
-1%
-2%
-1%
0%
0%
0%
2%
3%
3%
2%
7%
7%
6%
5%
9%
9%
8%
11%
13%
14%
12%
13%
15%
18%
14%
Practice list size
2,000 and under
2,001 to 5,000
5,001 to 10,000
10,001 and above
1.038
1.016
1.002
0.973
1.032
1.010
1.005
0.980
-0.55%
-0.46%
0.47%
0.78%
-13%
-9%
-8%
-8%
-9%
-8%
-8%
-7%
-5%
-5%
-5%
-4%
-4%
-4%
-4%
-3%
-3%
-3%
-2%
-1%
-1%
-1%
0%
0%
0%
1%
2%
2%
4%
6%
7%
5%
8%
8%
9%
7%
11%
13%
13%
10%
13%
18%
16%
13%
Population density
Least dense quartile
Quartile 2
Quartile 3
Most dense quartile
0.953
0.988
1.029
1.062
0.999
0.987
1.015
1.022
4.82%
-0.08%
-1.40%
-3.67%
-5%
-4%
-13%
-13%
-2%
-4%
-5%
-9%
0%
-3%
-4%
-7%
0%
-2%
-3%
-6%
2%
-1%
-2%
-5%
4%
0%
-1%
-3%
7%
1%
0%
-2%
10%
2%
0%
-1%
11%
2%
1%
-1%
14%
4%
2%
1%
18%
7%
5%
4%
Staff MFF
Lowest MFF quartile
Quartile 2
Quartile 3
Highest MFF quartile
1.014
1.014
0.984
1.013
1.033
1.020
0.984
0.987
1.96%
0.74%
0.12%
-2.40%
-5%
-6%
-6%
-13%
-4%
-5%
-5%
-9%
-2%
-3%
-4%
-7%
-2%
-3%
-3%
-6%
-1%
-2%
-2%
-5%
0%
0%
-1%
-3%
4%
2%
1%
-1%
8%
6%
5%
2%
10%
9%
7%
4%
14%
13%
11%
9%
15%
18%
14%
11%
SLLI
Lowest SLLI quartile
Quartile 2
Quartile 3
Highest SLLI quartile
0.927
0.979
1.024
1.110
0.949
0.982
1.012
1.090
2.43%
0.36%
-1.17%
-1.74%
-13%
-13%
-9%
-9%
-7%
-7%
-8%
-8%
-4%
-5%
-6%
-6%
-3%
-4%
-4%
-5%
-1%
-2%
-3%
-3%
2%
0%
-1%
-1%
6%
2%
0%
0%
9%
6%
2%
1%
11%
9%
3%
2%
14%
14%
9%
4%
18%
16%
12%
7%
SMR<65
Lowest SMR quartile
Quartile 2
Quartile 3
Highest SMR quartile
0.936
0.969
1.029
1.105
0.961
0.974
1.017
1.079
2.80%
0.51%
-1.19%
-2.31%
-13%
-7%
-13%
-9%
-6%
-7%
-8%
-8%
-3%
-4%
-5%
-7%
-2%
-3%
-4%
-6%
-1%
-2%
-3%
-4%
2%
0%
-1%
-2%
6%
2%
0%
-1%
9%
5%
2%
0%
11%
8%
3%
1%
14%
13%
9%
3%
15%
18%
12%
4%
New Registrations
Lowest New Reg quartile
Quartile 2
Quartile 3
Highest New Reg quartile
1.011
0.987
0.986
1.048
1.015
1.000
0.988
1.019
0.50%
1.49%
0.40%
-2.62%
-8%
-9%
-9%
-13%
-5%
-6%
-7%
-9%
-3%
-4%
-4%
-7%
-3%
-3%
-4%
-6%
-1%
-1%
-2%
-5%
0%
0%
-1%
-3%
1%
4%
2%
-1%
4%
8%
7%
1%
7%
10%
9%
2%
13%
14%
12%
9%
18%
15%
16%
13%
Nursing and Residential Home residents
Lowest NRH quartile
1.012
0.993
Quartile 2
0.997
0.993
Quartile 3
1.002
1.010
Highest NRH quartile
1.014
1.023
-1.73%
-0.27%
0.96%
1.05%
-13%
-9%
-8%
-8%
-9%
-8%
-7%
-6%
-7%
-5%
-3%
-4%
-6%
-4%
-3%
-3%
-4%
-2%
-1%
-1%
-2%
-1%
0%
0%
-1%
1%
2%
3%
4%
4%
7%
7%
7%
8%
9%
9%
13%
13%
13%
12%
15%
18%
15%
16%
Patients age > 65
Lowest P(pat>65) quartile
Quartile 2
Quartile 3
Highest P(pat>65) quartile
1.009
0.999
0.998
1.017
0.980
0.993
1.006
1.036
-2.67%
-0.51%
0.94%
2.01%
-13%
-9%
-9%
-6%
-8%
-8%
-5%
-5%
-7%
-5%
-3%
-3%
-6%
-4%
-3%
-2%
-4%
-3%
-2%
-1%
-3%
-1%
0%
1%
-1%
1%
2%
5%
0%
3%
7%
8%
1%
6%
9%
10%
7%
11%
12%
14%
12%
13%
14%
18%
London
In London
Outside London
1.050
0.996
1.007
1.005
-3.95%
0.96%
-13%
-6%
-9%
-5%
-8%
-3%
-7%
-3%
-5%
-1%
-4%
0%
-3%
2%
-2%
7%
-1%
9%
2%
13%
4%
18%
Spearhead PCTs
In Spearhead PCT
Not in Spearhead PCT
1.077
0.970
1.062
0.976
-1.26%
0.75%
-9%
-13%
-8%
-8%
-6%
-5%
-5%
-4%
-3%
-2%
-1%
0%
0%
3%
2%
7%
3%
9%
11%
13%
15%
18%
GMS only
Review of the General Medical Services global sum formula
63
Appendix I: Guide to the projected distributional
impact of the recommended formula
I.1
Throughout the formula review process we considered the impact of its
decisions through detailed modelling on the distribution of weighted patients
across practices. This was to ensure that when decisions regarding
implementation are made, the full impact of any resource redistribution and
practice stability are known.
I.2
This report presents the distributional impact of our’s recommended formulae.
This modelling work is based on 1 July 2006 data for all GMS and PMS
practices in England. For every practice the weighted list sizes that would result
from applying the options for the revised global sum formula on that date are
calculated. This is compared to the results of applying the current global sum
formula on the same date.
I.3
This provides results for close to 8,500 practices, including almost 5,000 GMS
practices. To allow conclusions to be drawn from such a large amount of data,
projected distributional impact tables (as used in Appendix F) were developed.
I.4
The tables provide results for various practice groups or practice cohorts. This
allows the specific consideration of the distributional impact of particular groups
of practices that we considered to be of particular interest.
I.5
Although all practices, regardless of size, are included in the modelling to
produce these results, the tables exclude figures relating to practices with fewer
than 100 patients. This is because such small practices are often atypical and
could appear to skew the distributions.
I.6
We considered the impact of its recommendations upon all practices and also
looked at the impact of particular groupings of practices as determined by:
I.7
64
•
contract type (GMS or PMS)
•
number of GPs
•
list size
•
population density
•
staff Market Forces Factor (MFF)
•
Standardised Limiting Long-Standing Illness (SLLI) scores
•
Standardised Mortality Ratio for those under 65 (SMR<65) scores
•
proportion of new registrations
•
proportion of patients living in nursing and residential homes
•
proportion of patients aged over 65
•
London practices
•
practices in Spearhead PCTs.
The first two columns of numbers provide the average overall formula weights
for each practice cohort, for both the current and the option formula. This is the
Review of the General Medical Services global sum formula
average ratio of weighted list size to unweighted list size and is a measure of
the average strength and direction of the formula on the practice cohort.
Although average weighted list size is equal to average unweighted list size, the
average ratio for all practices need not equal exactly 1. This is because the
average is calculated with equal weight applied to each practice irrespective of
variations in practice size.
I.8
For example, a larger practice with a list size of 8,000 and a ratio of 1.5 would
have 12,000 weighted patients. Two smaller practices, each with a list size of
4,000 and a ratio of 0.5 would each have 2,000 weighted patients. In total,
across the three practices, this gives 16,000 unweighted patients (i.e. 8,000 +
4,000 + 4,000) and 16,000 weighted patients (i.e. 12,000 + 2,000 + 2,000).
However the average ratio is 0.853 (i.e. [1.5 + 0.5 + 0.5]/3).
I.9
The next column, “average % change”, is the average percentage change in
the overall formula weight in moving from the baseline formula to the option
formula. This is equivalent to the average percentage gain in weighted patients
that would be expected for practices in the cohort of interest from making the
formula change.
I.10 The remaining columns show the distribution of percentage changes across
practices. For each practice the percentage change is calculated from moving
from one option (e.g. the current formula) to another (a revised formula). The
changes are then ranked across all process and the largest negative (labelled
“Min”), median (labelled “50%”) and largest positive (labelled “Max”) percentage
changes are reported, along with those at other percentiles in the distribution.
Review of the General Medical Services global sum formula
65
GMS global sum formula
19/1/07
10:13
Page 2
NHS Employers
General Practitioners Committee
www.nhsemployers.org
Email enquiries@nhsemployers.org
NHS Employers
29 Bressenden Place
London SW1E 5DD
www.bma.org.uk
British Medical Association
BMA House
Tavistock Square
London WC1H 9JP
2 Brewery Wharf
Kendell Street
Leeds LS10 1JR
Published January 2007
© NHS Employers and BMA 2007. This document may not be reproduced in whole or in part without permission
The NHS Confederation (Employers) Company Ltd
Registered in England. Company limited by guarantee: number 5252407
Ref. EGUI02401