13
Social Science & Medicine 58 (2004) 539–551 The pursuit of equity in NHS resource allocation: should morbidity replace utilisation as the basis for setting health care capitations? Sheena Asthana a, *, Alex Gibson b , Graham Moon c , John Dicker d , Philip Brigham e a Department of Social Policy and Social Work, University of Plymouth, Drake Circus, Plymouth PL4 8AA, UK b Department of Geography, University of Exeter, Amory Building, Rennes Drive, Exeter EX4 4PN, UK c Institute for the Geography of Health, University of Portsmouth, Milldam, Burnaby Rd, Portsmouth PO1 3AS, UK d Information Management and Technology, Iechyd Morgannwg, 41 High Street, Swansea SA1 1LT, UK e Department of Social Policy and Social Work, University of Plymouth, Drake Circus, Plymouth PL4 8AA, UK Abstract Although the English NHS has been described as a world leader in pioneering methods of distributing expenditure in relation to population needs, concerns about the legitimacy of using the current utilisation-based model to allocate health service resources are mounting. In this paper, we present a critical review of NHS resource allocation in England and demonstrate the feasibility and impact of using direct health estimates as a basis for setting health care capitations. Comparing target allocations for the inpatient treatment of coronary heart disease in a sample of 34 primary care trusts in contrasting locations in England, we find that a morbidity-based model would result in a significant shift in hospital resources away from deprived areas, towards areas with older demographic profiles and towards rural areas. Discussing the findings in relation to a wider policy context that is generally concerned to direct more health care resources towards the poor, the paper concludes by calling for greater clarity between the goals of health care equity and health equity. Whilst the former demands that the legitimate needs of demographically older populations for more health care resources are acknowledged, the goal of health equity requires real political commitment to resource broader social policy initiatives. r 2003 Elsevier Science Ltd. All rights reserved. Keywords: NHS resource allocation; Health and equity; Urban and rural health needs; Morbidity-based capitation; England Introduction As the first health care system to distribute the majority of its revenue resources between geographical areas in relation to population need (Mays, 1995), the English NHS has received international attention. Indeed, the NHS has been described, albeit by the main architects of the current resource allocation system, as a ‘world leader in pioneering scientific methods of equitable resource allocation’ (Carr-Hill et al., 1997, p. 69). Weighted capitation has become the principal method of allocating health care finance to regions in many countries (Rice & Smith, 1999). As in England, most of these capitation systems are based on empirical models of health care expenditure. A transparent, empirically based approach to match- ing health spending to population need is generally regarded as an improvement upon allocation based on historical patterns of expenditure. However, as we discuss in the first part of this paper, the current system of NHS allocation in England is not without its critics. ARTICLE IN PRESS *Corresponding author. Tel.: +44-1752-233262; fax: +44- 1752-233209. E-mail addresses: [email protected] (S. Asthana), [email protected] (A. Gibson), [email protected] (G. Moon), [email protected] (J. Dicker), [email protected] (P. Brigham). 0277-9536/03/$ - see front matter r 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0277-9536(03)00217-X

The pursuit of equity in NHS resource allocation: should morbidity replace utilisation as the basis for setting health care capitations?

  • Upload
    soton

  • View
    0

  • Download
    0

Embed Size (px)

Citation preview

Social Science & Medicine 58 (2004) 539–551

The pursuit of equity in NHS resource allocation:should morbidity replace utilisation as the basis for setting

health care capitations?

Sheena Asthanaa,*, Alex Gibsonb, Graham Moonc, John Dickerd, Philip Brighame

aDepartment of Social Policy and Social Work, University of Plymouth, Drake Circus, Plymouth PL4 8AA, UKbDepartment of Geography, University of Exeter, Amory Building, Rennes Drive, Exeter EX4 4PN, UK

c Institute for the Geography of Health, University of Portsmouth, Milldam, Burnaby Rd, Portsmouth PO1 3AS, UKd Information Management and Technology, Iechyd Morgannwg, 41 High Street, Swansea SA1 1LT, UK

eDepartment of Social Policy and Social Work, University of Plymouth, Drake Circus, Plymouth PL4 8AA, UK

Abstract

Although the English NHS has been described as a world leader in pioneering methods of distributing expenditure in

relation to population needs, concerns about the legitimacy of using the current utilisation-based model to allocate

health service resources are mounting. In this paper, we present a critical review of NHS resource allocation in England

and demonstrate the feasibility and impact of using direct health estimates as a basis for setting health care capitations.

Comparing target allocations for the inpatient treatment of coronary heart disease in a sample of 34 primary care trusts

in contrasting locations in England, we find that a morbidity-based model would result in a significant shift in hospital

resources away from deprived areas, towards areas with older demographic profiles and towards rural areas. Discussing

the findings in relation to a wider policy context that is generally concerned to direct more health care resources towards

the poor, the paper concludes by calling for greater clarity between the goals of health care equity and health equity.

Whilst the former demands that the legitimate needs of demographically older populations for more health care

resources are acknowledged, the goal of health equity requires real political commitment to resource broader social

policy initiatives.

r 2003 Elsevier Science Ltd. All rights reserved.

Keywords: NHS resource allocation; Health and equity; Urban and rural health needs; Morbidity-based capitation; England

Introduction

As the first health care system to distribute the

majority of its revenue resources between geographical

areas in relation to population need (Mays, 1995), the

English NHS has received international attention.

Indeed, the NHS has been described, albeit by the main

architects of the current resource allocation system, as a

‘world leader in pioneering scientific methods of

equitable resource allocation’ (Carr-Hill et al., 1997, p.

69). Weighted capitation has become the principal

method of allocating health care finance to regions in

many countries (Rice & Smith, 1999). As in England,

most of these capitation systems are based on empirical

models of health care expenditure.

A transparent, empirically based approach to match-

ing health spending to population need is generally

regarded as an improvement upon allocation based on

historical patterns of expenditure. However, as we

discuss in the first part of this paper, the current system

of NHS allocation in England is not without its critics.

ARTICLE IN PRESS

*Corresponding author. Tel.: +44-1752-233262; fax: +44-

1752-233209.

E-mail addresses: [email protected] (S. Asthana),

[email protected] (A. Gibson), [email protected]

(G. Moon), [email protected] (J. Dicker),

[email protected] (P. Brigham).

0277-9536/03/$ - see front matter r 2003 Elsevier Science Ltd. All rights reserved.

doi:10.1016/S0277-9536(03)00217-X

Notwithstanding the statistical sophistication of the

empirical models that underpin the system, the legitimacy

of deriving needs indicators from existing patterns of

health care utilisation is questionable (Sheldon, Smith, &

Bevan, 1993; Mays, 1995). Concerns have been raised

about the impact of applying formulae designed to

allocate resources at Health Authority level to smaller

primary care trusts (PCTs) (Bevan, 1998). The limitations

of the weighted capitation approach have also become

more apparent with the growth of interest in the role of

targeting clinically effective health services in improving

health and reducing health inequalities. The current

formula offers an approximation of general health service

needs. However, it is difficult to apply to more specific

areas of clinical activity such as those targeted in key

National Service Frameworks (DoH, 2000).

Against this background, interest is growing in a

radically new approach to health resource allocation,

one that distributes NHS resources on the basis of direct

measures of morbidity rather than indirect proxies such

as health service utilisation or deprivation. Despite the

perceived legitimacy of the morbidity-based approach,

there has been little practical testing of such a model for

resource allocation. The aim of this paper is to explore

the feasibility and impact of replacing utilisation with

morbidity as a basis for setting health care capitations.

In the second part of this paper, we illustrate a

methodology by which symptom-based capitation esti-

mates for the inpatient treatment of coronary heart

disease (CHD) can be derived. Comparing morbidity-

based allocations with indicative allocations based on

the hospital and community health services (HCHS)

component of the national weighted capitation formula,

we demonstrate the impact of replacing utilisation with

morbidity as the basis for setting health care capitations.

Our findings suggest that, as proposed by critics of the

utilisation-based approach, the HCHS formula does

introduce systematic biases against particular popula-

tion groups in the way in which it allocates resources.

Perhaps against expectations, the evidence we present

suggests that the current formula discriminates in favour

of deprived urban areas.

In the final part of the paper, we discuss the feasibility

and desirability of developing an alternative approach to

NHS resource allocation. We highlight some of the

practical issues that would need to be addressed before a

morbidity-based capitation methodology could be im-

plemented. These include the need to validate epidemio-

logical data and to establish consistent definitions of

how symptom-based morbidity relates to service need.

More fundamentally, by demonstrating the ways in

which utilisation- and morbidity-based capitation meth-

odologies result in very different distributions of

resources, we propose that there needs to be greater

clarity and transparency about the purpose of resource

allocation.

NHS resource allocation in England: a critical review

The principle that health care expenditure should be

geographically distributed in relation to population need

was established in response to the widespread perception

that, when it was created, the NHS inherited gross

inequalities in provision. Several commentators have

proposed that the pre-NHS distribution of health care

resources reflected ‘past philanthropy, municipal pride

and local affluence rather than a planned response to

population needs’ (Beech, Bevan, & Mays, 1990, p. 44).

This fuelled concerns that access to British health care

services was not only unequal but also inequitable, areas

with the greatest needs having the lowest service

provision. Assertions about inequity date back to the

beginnings of the NHS (Powell, 1997, p. 34). However,

despite developments such as the 1962 Hospital Plan,

existing geographical inequalities were largely perpetu-

ated by a system of incremental budgeting until the 1970s.

The most significant break from this system came with

the appointment of the Resource Allocation Working

Party (RAWP). In 1976, RAWP recommended that

revenue resources for HCHS should be distributed on

the basis of population, weighted according to differ-

ences in the age/sex structure, the need for health care

and in the costs of providing services (ACRA, 1999).

The RAWP formula, and a revised version introduced in

1990, continued to be used to allocate HCHS resources

to Regional Health Authorities until 1994. However, its

use of standardized mortality ratios (SMRs) as a proxy

for relative needs was criticized for failing to fully reflect

the demand for health care resources produced by

chronic disease and deprivation.

In 1995 a new weighted capitation formula for HCHS

was introduced. This comprises an age index (based on

estimates of national spend per head in eight age bands)

and an ‘additional needs’ index (over and above

demographics). The latter is derived from an empirical

model (the York model) that identified its needs

indicators as those census-derived health status and

socio-economic variables that, having adjusted for the

independent effects of supply, were most closely

correlated with the national average pattern of hospital

utilisation (Carr-Hill et al., 1994).

Whilst the 1995 formula was applied at Health

Authority level, the election of the Labour Government

in 1997 maintained the move initiated by the conserva-

tives towards a more devolved management structure

and has since resulted in the establishment of PCTs. This

is now the spatial level at which the national resource

allocation formulae are applied.

Critique of the utilisation-based approach

The technical analyses that have informed the current

system of NHS resource allocation in England

ARTICLE IN PRESSS. Asthana et al. / Social Science & Medicine 58 (2004) 539–551540

(Carr-Hill et al., 1994; Rice, Dixon, Lloyd, & Roberts,

2000) are undoubtedly sophisticated. For all its merits,

however, the English formula and the empirically based

models on which it is based have been criticised. The

most fundamental criticism of these relates to the use of

utilisation-based models to assess need for health care.

This implies that historical patterns of service uptake

between different care groups (as revealed by utilisation)

are appropriate (Mays, 1995), a problematic assumption

given the concerns that are regularly expressed about

ageism, sexism, racism and socio-economic bias in

access to health care.

For example, a growing body of evidence suggests

that the use of potentially life saving and life enhancing

investigations and interventions decline as patients get

older (Bowling, 1999). If, on the basis of clinical

evidence, it is accepted that older patients can and

should gain from more intensive treatment, then use-

based per capita allocations for older age bands could be

regarded as conservative.

There is also strong evidence to suggest that

geographical access to services has a profound effect

on health care utilisation (Rice & Smith, 2001). This is

demonstrated by the distance decay effect, where use of

health care drops as distance to a service increases.

Rural areas have been found to exhibit lower levels of

health service use than their urban counterparts (Haynes

& Bentham, 1982; Bentham & Haynes, 1986; Jones,

Bentham, & Harrison, 1998; Gibson, Asthana, Brigham,

Moon, & Dicker, 2002). Concerns have been expressed

particularly about uptake of mental health services in

rural areas where, due to isolation, stigma and low

service expectations, residents are less likely to seek help

for mental health problems than urban dwellers (Fearn,

1987; Gift & Zastowny, 1990). This is not acknowledged

in the resource allocation formula for psychiatric

services which responds to the relatively high rates of

utilisation in inner cities.

Interpreting the appropriateness of weightings given

to socio-economic status (a major component of the

additional needs index within the Hospital and Com-

munity Services Formula) is more complex and contest-

able. Since Tudor-Hart first proposed that the

availability of good primary care varies inversely with

the need of the population served (Tudor-Hart, 1971),

claims that the accessibility and use of NHS services

(primary or secondary) are subject to the ‘inverse care

law’ have become received wisdom. This would suggest

that need, as revealed by use will be under-estimated in

deprived areas.

In fact, evidence of inverse care is equivocal. Whilst

some studies suggest that deprived populations have

significantly lower rates of health service use according

to need (Payne & Saul, 1997; Hippisley-Cox & Pringle,

2000), others find little difference between deprived and

more affluent populations (Manson-Siddle & Robinson,

1998). Indeed, some research shows that the residents of

more deprived areas experience higher rates of use

relative to need (O’Donnell & Propper, 1991; Black,

Langham, & Petticrew, 1995; Gibson et al., 2002). Ten

years on from when Julian Le Grand suggested that ‘the

jury is still out on the question as to whether the NHS

provides equal treatment for equal need’ (Le Grand,

1991), still little is known about socio-economic inequal-

ities in the use of NHS services.

Research evidence about inequities in utilisation on

the basis of other population characteristics (e.g.

ethnicity) is scarce. On the basis of available research

evidence, one can therefore only suggest that there is a

case for suggesting that a utilisation-based capitation

methodology may systematically discriminate against

areas serving demographically older populations and

rural areas in estimating need for health care. The

direction and impact of socio-economic and ethnic

biases in the utilisation-based model are far more

uncertain.

Resource allocation and the problem of scale

Until 2002, the weighted capitation formulae have

been applied at Health Authority level (i.e. to popula-

tions ranging from around 250,000–500,000 in size). The

basic unit of resource allocation is now the PCT. In

1999, primary care groups (which have since evolved

into PCTs) had average populations of 100,000, though

this ranged from 44,000 to 277,000 (Bojke, Gravelle, &

Wilkin, 2001). A number of primary care groups and

trusts have since merged, increasing the average

population size. The average population size of PCTs

in the South West Region of England (n ¼ 42) is now

162,000.

Simply as a function of their smaller size, PCTs are

both more internally homogeneous and more socially

and demographically diverse than Health Authorities.

The range of values attached to individual PCTs using

selected needs indicators will thus be wider that the

range attached to Health Authorities and the sensitivity

of different indices to different dimensions of need more

pronounced at the PCT level.

It is therefore likely that, proportionately, distance

from target funding allocations will be greater for PCTs

than for Health Authorities. If the demographic and

socio-economic weightings that make up the current

formulae are adequate proxies for health service need,

then the greater variability in distance between actual

and expected resource use may well be appropriate.

However, if as suggested above, the formulae do not

reflect the legitimate health care needs of all population

groups, any systematic biases against particular demo-

graphic or socio-economic groups will be most strongly

played out where these characteristics are locally

clustered.

ARTICLE IN PRESSS. Asthana et al. / Social Science & Medicine 58 (2004) 539–551 541

The impact of scale on variability in distance from

targets is likely to be exacerbated by changes in the way

in which the allocation of resources can be practically

managed. In the past, large and heterogeneous Health

Authorities could accommodate differences between

localities in their use of health resources. When

resources are allocated directly to PCTs, there will be

reduced scope for a flexible system of redistribution. It is

therefore critical that any systematic biases in the

formulae against particular types of areas are elimi-

nated.

An alternative approach to setting health care capitations:

a morbidity-based model

On the basis of the critiques outlined above, there

appear to be sufficient grounds to propose that

systematic biases against certain population groups

and geographical areas are likely to be introduced by

the English NHS formula. These will be more strongly

played out with the reduction of the scale at which

resources are allocated. One of the perceived merits of

the current resource allocation formula, however, is that

it is empirically based. Until recently, there was general

agreement that, in the absence of adequate data on

direct health needs, the utilisation-based model was the

best available option. Concerns about the legitimacy of

the utilisation-based approach are nevertheless mount-

ing. In response to these, interest is growing in the use of

a radically new approach to health resource allocation,

one that distributes NHS resources on the basis of direct

measures of morbidity rather than indirect proxies such

as health service utilisation or deprivation.

A major review of resource allocation in Scotland

(SHED, 1999) examined the scope for directly measur-

ing the relative distribution of ill-health in the popula-

tion through the use of epidemiological data on the

incidence of disease. However, the review concluded that

although the approach was attractive, more work would

need to be done before it could be practically

implemented. The Welsh Resource Allocation Review

(NHS Wales, 2001) went further in strongly recom-

mending the use of a morbidity-based budgeting

approach. Again, however, this review acknowledged

that the approach was at an early stage of development.

Despite the perceived legitimacy of the morbidity-

based approach, there has thus been little practical

testing of such a model for resource allocation. As part

of an ongoing programme of research on inequalities in

health care utilisation, we have been exploring the

potential of using age, sex and class-adjusted epidemio-

logical estimates (based on the Health Survey for

England) to create more robust needs indicators against

which to monitor activity. We propose that this

approach has a number of important advantages.

First, it directly adjusts for social gradients in disease

prevalence. This means that the association between

socio-economic status and morbidity is captured (a key

aim of the York model). At the same time, uncertainties

about the direction and magnitude of socio-economic

bias in the utilisation model are addressed. This is

because weighted capitation indices that are based on

patterns of utilisation cannot distinguish the effect of

social deprivation on need from the quite separate effect

of social deprivation on use. By contrast, epidemiologi-

cal estimates can be incorporated into models that

explicitly seek to explain variations in use relative to need

(Gibson et al., 2002).

Second, the approach considers the role of age, sex

and class simultaneously when assessing need. It might

be argued that this task is accomplished by combining

age and additional needs indices. However, the ap-

proach taken in constructing the HCHS formula rests

on certain (untested) assumptions about how the

various components that explain variations in need

combine. Unless account is taken of the relative

importance of demography and deprivation in determin-

ing need, a commendable concern for socio-economic

inequality may inadvertently obscure age (and gender)

inequality.

Third, a strength (but also, as we discuss in the

conclusion of this paper, a weakness) of this approach is

that it can be applied to specific clinical areas. With the

exception of the formula for psychiatric services, the

current weighted capitation system offers an approx-

imation of general health service needs. This limits its

utility in specific policy areas such as the National

Service Frameworks. There is growing interest in the

possibility of developing target allocations for selected

disease conditions that exhibit a high social gradient and

are amenable to health care intervention. A morbidity-

based capitation methodology lends itself to this

purpose.

Finally, we have provided an empirical case for the

use of direct estimates rather than indirect proxy

measures including weighted capitation indices. Incor-

porating epidemiological estimates into models that seek

to explain utilisation relative to need, we have found

that a relatively small set of population characteristics

and supply factors appear to play a significant part in

determining both prescribing rates and inpatient activ-

ity. Once these are taken into account, it is possible to

explain a greater proportion of variation in prescribing

and inpatient use than is possible through the use of

both the prescribing and HCHS indices (Gibson et al.,

2002).

To date, we have reported upon the use of needs-

based estimates as a basis for monitoring equity in

health care utilisation. Below, we consider the feasibility

and impact of using direct health estimates as a basis for

setting health care capitations.

ARTICLE IN PRESSS. Asthana et al. / Social Science & Medicine 58 (2004) 539–551542

Method

The aim of the following analysis was to compare how

a morbidity-based capitation methodology and the

current HCHS formula would distribute an identical

budget for the hospital treatment of CHD. By develop-

ing resource allocations for a condition where profound

social gradients are known to exist in disease prevalence

(Davey-Smith, Shipley, & Rose, 1990; Davey-Smith,

Hart, Watt, Hole, & Hawthorne, 1998), we expected the

distribution of resources for the treatment of CHD to be

more skewed towards urban deprived populations than

a general budget. This is because not all conditions are

strongly associated, or indeed positively associated in

their prevalence with deprivation.

Study sample

Our study covers 34 PCTs in 7 Health Authorities in

contrasting locations in England. These include three

rural, one prospering, one maturing and two mining and

industrial areas according to the ONS classification of

health authority types and are located in four NHS

regions. Participating health authorities provided GP

registration data and inpatient minimum datasets for all

589 general practices in the 34 PCTs (n ¼ 3:54 million

patients in April 1999). In the 3 years 1996/1997–1998/

1999, registered patients within the study sample

accounted for 71,426 inpatient episodes with main

diagnosis angina or myocardial infarction (ICD10 I20-

25), The total reference cost of these episodes was

d90,264,143.

Research design

Our method for developing indicative resource

allocations comprises five distinct stages.

Stage 1: Obtaining national prevalence rates: Direct

morbidity estimates for local PCTS are derived by using

age/sex/class prevalence rates recorded in the national

Health Survey for England (HSE) in conjunction with

age, sex and class profiles of PCTs (Gibson et al., 2002).

The decision to base the initial prevalence calculations

on the HSE was informed by the cumulative sample size

of the ongoing survey. It comprises, to date, nine

separate surveys. The first spans 1991 and 1992 and the

rest are annual, from 1993 to 1999 inclusive. The annual

sample size of the HSE has ranged from over 7000 in

1991/1992 to nearly 20,000 in 1996. By summing the

populations of survey years for which relevant questions

have been asked, the HSE provides a sufficiently large

sample to produce robust prevalence rate matrices for

men and women, each comprising 7 age categories and 6

social class categories. We derive these matrices for

social class in addition to age and sex in recognition of

the known variation of conditions with respect to social

class and with the presumption that any consequent

resource allocation based on direct needs would wish to

take account of such inequality.

The HSE provides data for a particularly broad range

of conditions and, for cardiovascular disease in parti-

cular (Lampe, Colhoun, & Dong, 1994; Erens &

Primatesta, 1999), in some detail. Thus prevalence rates

can be described for three-character categories within

the International Statistical Classification of Diseases

(ICD 10), such as angina pectoris (I20) and myocardial

infarction (I21) and blocks of three-character categories

such as hypertensive disease (I10-I15) and arrhythmias

(I47-49). We have chosen to focus on symptom-based

rates of severe angina and myocardial infarction in order

to capture unexpressed need (a problem that may affect

the equivalent doctor-diagnosed variables). The HSE

incorporates the World Health Organisation Rose

Questionnaire (Rose & Blackburn, 1986) to establish

whether respondents have experienced Grade 1 or 2

Angina. Informants are classified as having had

‘possible myocardial infarction’ if they report ever

having had an attack of severe pain across the front of

the chest, lasting for half an hour or more.

Stage 2: Attribution of prevalence rates to PCT

populations: Having determined the demographic and

social distribution of severe angina/myocardial infarc-

tion on the basis of the 36,663 individuals questioned by

the Health Survey for England between 1991 and 1994,

equivalent socio-demographic matrices need to be

established for PCT populations. The demographic

structure of PCTs has been obtained using Patient

Registration data (April 1999). Social class profiles have

been estimated using patient-weighted attribution of

census data (whereby patient postcodes are used to link

patients to the geography of the census). To automate

this process which, to all intents and purposes, mirrors

that used in a variety of previous studies (Ben-Shlomo &

Chaturvedi, 1995; Scrivener & Lloyd, 1995; Ward,

Morton Jones, Pringle, & Chilvers, 1994) and in the

creation of the Attribution Dataset, we have developed a

program called NCP Profiler. This lies at the heart of the

process by which socio-demographic profiles of non-

contiguous populations such as GP or PCT populations

are derived (Gibson & Asthana, 2000).

The age, sex and social class profiles of PCTs are

combined with HSE prevalence rates to obtain a

prevalence matrix for the designated PCT population.

Total rates or counts of individuals who would be

expected to experience a particular condition can then be

taken for the PCT as a whole or, for the purpose of our

methodology, for each of the male and female age

cohorts. Table 1 illustrates the methodology for the male

population of one PCT in our study. The table suggests

that the use of the Rose Questionnaire is likely to have

resulted in an over-estimation of the level of CHD

prevalence amongst younger people. The rates of severe

ARTICLE IN PRESSS. Asthana et al. / Social Science & Medicine 58 (2004) 539–551 543

angina and/or myocardial infarction given for the 16–24

year old age group are significantly higher than other

available morbidity estimates for this age group. This

means that our method is likely to over-estimate CHD

prevalence in demographically younger populations.

Stage 3: Establishing an age/sex resource matrix:

Having derived estimates of the CHD burden within

each age/sex category, the next step is to specify an age/

sex resource matrix for the treatment of CHD. In the

absence of specific clinical guidelines about appropriate

expenditure for each age/sex category, this matrix can

only be based on historical activity. HRG Reference

Costs for inpatient episodes with a main diagnosis code

between ICD-10 I20 and I25 has therefore been totalled

across the entire sample for each age/sex cohort. This

has been divided by the number of men and women in

each age cohort who, on the basis of our age and class-

adjusted epidemiological estimates, would be expected

to have symptoms of severe (grade 2) angina and/or

myocardial infarction. Average per capita resource use

per annum can thus be established for each cohort.

Stage 4: Calculating CHD clinical programme budgets

for PCTs: Although average costs are read from patterns

of use, our indicative allocations are nevertheless

calculated from age, sex and class-adjusted morbidity

estimates. In order to determine a CHD clinical

programme budget for each PCT, we have multiplied

the estimated number of people with symptoms of severe

angina and/or MI in each age/sex cohort by that

cohort’s average ‘per capita in need’ resource use, then

aggregated the totals of all cohorts. The resulting budget

is thus sensitive to the demographic and social

characteristics of PCT populations and reflects the

overall (i.e. symptom-based) level of need rather than

just expressed (e.g. doctor-diagnosed) need.

Stage 5: Calculating an implied standard weighted

capitation allocation for CHD services: In order to

establish how the weighted capitation formula would

distribute the d90,264,143 total reference cost attached

to inpatient episodes with a main diagnosis of angina or

myocardial infarction, we have followed the methodol-

ogy detailed by the NHS Executive (1999) and

established age and additional needs weighted popula-

tions for the 34 PCTs in the study. We have included

neither the Market Forces Factor nor the Emergency

Ambulance Cost Adjustment in the calculation of these

weighted populations as we have not sought to allow for

such unavoidable variations of costs in the calculation of

ARTICLE IN PRESS

Table 1

Estimating the number of male patients in a PCO with symptoms of severe angina and/or myocardial infarction

16–24 25–34 35–44 45–54 55–64 65–74 75+

A: Estimated age-class profile of PCO population; Males 16+

I 312 455 439 481 396 333 287

II 1968 2872 2772 3035 2500 2103 1809

IIIN 903 1318 1273 1393 1148 965 830

IIM 1796 2621 2530 2770 2281 1919 1650

IV 978 1428 1379 1509 1243 1046 899

V 290 424 409 448 369 310 267

Total 6247 9118 8802 9636 7937 6676 5742

B: HSE 1991–1994 derived prevalence rates for symptoms of severe angina and/or myocardial infarction; Males, 16+

16–24 (%) 25–34 (%) 35–44 (%) 45–54 (%) 55–64 (%) 65–74 (%) Over 75 (%)

I 2.60 4.08 3.81 5.47 10.27 12.50 15.07

II 2.72 3.51 6.06 7.51 10.79 15.07 12.42

IIIN 3.47 4.19 4.84 11.42 13.39 15.11 10.34

IIM 2.64 3.25 6.53 9.61 14.90 18.78 17.37

IV 4.69 5.48 8.68 7.49 16.57 12.34 16.56

V 7.56 5.00 10.66 13.22 17.69 16.67 14.29

C: Predicted number of patients in the PCO with symptoms of severe angina and/or myocardial infarction, Males 16+

16–24 25–34 35–44 45–54 55–64 65–74 75+

I 8.10 18.54 16.72 26.33 40.67 41.63 43.25

II 53.62 100.83 168.00 227.86 269.78 316.94 224.65

IIIN 31.38 55.17 61.60 159.04 153.75 145.82 85.86

IIM 47.46 85.24 165.21 266.13 339.94 360.39 286.60

IV 45.84 78.25 119.76 113.07 206.02 129.05 148.84

V 21.93 21.20 43.58 59.24 65.27 51.67 38.14

Total 208.33 359.22 574.87 851.67 1075.44 1045.49 827.34

S. Asthana et al. / Social Science & Medicine 58 (2004) 539–551544

morbidity-based allocations. The total reference cost is

then partitioned between PCTs on the basis of their age

and additional needs weighted populations to provide

an implied HCHS-based clinical programme budget for

CHD.

Results

Morbidity-based resource allocation by age and sex

Table 2 provides an age/sex resource matrix for the

inpatient treatment of angina and myocardial infarction

using 7 age cohorts. These allocations have been

calculated from morbidity-based needs estimates,

though the costs attached to treatment are read from

historical activity (see Stage 3 of the methodology).

In all but the oldest age category, considerably more

resources ‘per capita in need’ are used to treat men than

women. This lends support to previous evidence of

differences in the treatment of men and women

presenting with CHD (Galatius-Jenson, Launberg,

Mortensen, & Hansen, 1996; Sonke, Beaglehole, Stew-

art, Jackson, & Stewart, 1996; Wenger, 1997).

The age–cost relationship given in Table 2 deserves

further comment. At first glance, the fact that our

costing procedure results in such significantly higher per

capita allocations for older patients than younger

patients seems problematic. However, the denominator

used in our calculations is the number of people

expected to have symptoms of severe angina and/or

myocardial infarction in the community. Many of these

patients will be managed within the community and

community/primary management tends to be more

appropriate and feasible amongst younger than older

age groups. Thus, higher rates of average resource use

for older age cohorts do not reflect a tendency to

discriminate in favour of older patients but the need to

hospitalise patients who are too sick to remain at home.

Indeed, we find that although absolute resource use per

capita increases with age up to the 65–74 year old group,

the likelihood of a person with CHD symptoms

receiving surgical intervention (coronary artery bypass

grafts and percutaneous transluminal coronary angio-

plasties) falls off dramatically with age (Asthana,

Gibson, Dicker, Moon, & Brigham, 2001a, b). This is

consistent with the growing body of evidence that

suggests that older people experience age-based discri-

mination in the access and use of health services (Age

Concern, 1999) and suggests that the per capita

allocations given to older age bands in Table 2 could

indeed be conservative. If this is the case, our method

would under-estimate resource needs in areas serving

demographically older populations.

Morbidity-based resource allocation by area: comparison

with HCHS-based indicative allocations

A comparison of morbidity-based and utilisation-

based CHD clinical programme budgets for PCTs

reveals very significant variations. For 16 PCTs in the

sample the adoption of a morbidity-based approach

would result in a drop in revenue relative to the HCHS-

based allocation of between 0.49% and 32.8%. Eighteen

PCTs would gain between 7.6% and 32.9%. d7,641,882

(or 8.5% of the total) would be reallocated if the CHD

clinical programme budget was allocated on the basis of

the morbidity-based methodology as opposed to the

utilisation-based formula. With an overall average

allocation of d31.22 per capita (aged 16+), the adoption

of a morbidity-based approach would result in per

capita reallocations to PCTs of between d7.22 gain and

d13.89 loss.

Differences between morbidity-based and utilisation-

based allocations can be related to the demographic,

socio-economic and geographic character of PCT

populations. Fig. 1 compares percentage differences

between morbidity-based and HCHS-based allocations

(relative to the latter) against (a) Townsend’s Material

Deprivation scores (r ¼ �0:845; po0:001), (b) the

ARTICLE IN PRESS

Table 2

Calculating an age/sex resource matrix for the inpatient treatment of angina and myocardial infarction

16–24 25–34 35–44 45–54 55–64 65–74 75+

A: Total HRG reference cost

Males d10,169 d193,762 d2,243,818 d9,625,578 d16,691,943 d18,958,669 d11,974,252

Females d5,713 d54,972 d489,157 d2,067,633 d5,212,507 d9,728,241 d13,007,729

B: Estimated patients with symptoms of severe angina and/or MI

Males 6451 9959 16,972 21,781 26,089 23,661 15,418

Females 4763 6882 9548 13,031 13,406 13,915 16,637

C: HRG reference cost per patient with symptoms of severe angina and/or MI

Males d1.58 d19.46 d132.21 d441.93 d639.81 d801.27 d776.64

Females d1.20 d7.99 d51.23 d158.67 d388.81 d699.13 d781.85

S. Asthana et al. / Social Science & Medicine 58 (2004) 539–551 545

percentage of population aged 65+ (r ¼ 0:529;p ¼ 0:001), and (c) the DETR’s ‘Access to Services’

scores (r ¼ 0:847; po0:001). This last variable is one of

the eight domains used in the construction of the

DETR’s recently published Index of Multiple Depriva-

tion (DETR, 2000). Available at ward level, this seeks to

summarise accessibility to local services (post-offices,

food shops, GPs and primary schools) and thus provides

ARTICLE IN PRESS

-40%

-30%

-20%

-10%

0%

10%

20%

30%

40%

-3 -2 -1 0 1 2 3 4

Tow nsend's Index of Deprivation

% d

iffer

ence

of

mor

bidi

ty-b

ased

allo

catio

n re

lativ

e to

HC

HS

-bas

ed a

lloca

tion

10% 15% 20% 25% 30%

Percent population aged 65+

% d

iffer

ence

of m

orbi

dity

-bas

ed a

lloca

tion

rela

tive

to H

CH

S-b

ased

allo

catio

n

-1.0 -0.5 0.0 0.5 1.0

DETR Access Domain Score

% d

iffer

ence

of m

orbi

dity

-bas

ed a

lloca

tion

rela

tive

to H

CH

S-b

ased

allo

catio

n

-40%

-30%

-20%

-10%

0%

10%

20%

30%

40%

-40%

-30%

-20%

-10%

0%

10%

20%

30%

40%

Fig. 1. Morbidity-based Resource Allocation for CHD Clinical Programme relative to HCHS allocation against Townsend,

proportion population 65+ and DETR’s Access Domain.

S. Asthana et al. / Social Science & Medicine 58 (2004) 539–551546

a useful proxy measure of rurality. The DETR’s ‘Access’

domain scores and Townsend scores have been attrib-

uted to PCT populations using patient postcodes. High

positive scores indicate high levels of rurality and

material deprivation, respectively. The percentage po-

pulation aged 65+ has been drawn directly from general

practice registers.

The higher the positive score on the y axes in Fig. 1,

the more a PCT would stand to gain from the

introduction of a morbidity-based allocation. Negative

scores indicate that the utilisation-based allocation is the

larger. Fig. 1 clearly illustrates the extent to which a

morbidity-based capitation methodology tends to result

in a significant shift of hospital resources for CHD away

from PCTs serving deprived areas; towards PCTs

serving populations with older demographic profiles;

and towards PCTs in rural areas.

The conclusions concerning deprivation are worthy of

further comment. Because the model underpinning the

standard resource allocation formula takes account of

the impact of deprivation on health service use, a close

fit between utilisation-based resourcing and deprivation

would be expected. However, we have used social class

in the calculation of the prevalence estimates that

underpin our model. Social class is a plausible surrogate

for deprivation. Thus, the resource shift implied in the

comparison of the two approaches reflects the relative

weighting ascribed to deprivation in the two models.

This is greater with respect to the utilisation-

based formula than it is with reference to underlying

morbidity.

It is possible that, due to factors such as co-morbidity

and disease severity (Eachus, Chan, & Pearson, 1999),

deprived groups have a greater need for hospital care

than more affluent groups at a given level of morbidity,

a tendency that would be captured by the utilisation-

based model. At a population level, however, the impact

of this factor will be counterbalanced by the fact that,

within the sample, more affluent areas tend to have older

demographic profiles. If ageism exists, utilisation rates

are likely to be suppressed in such areas.

It is also important to point out that the morbidity-

based model may build in its own biases against

demographically older populations, first by over-esti-

mating CHD prevalence in younger age groups and

second by under-estimating the resource needs of older

age groups.

Finally, the morbidity-based model is designed to

distribute resources for the treatment of a condition—

CHD—that is, subject to profound social gradients in its

prevalence. One would therefore expect it to better target

resources towards socially deprived areas than a formula

that distributes resources for the treatment of general

health needs. The fact that it has just the opposite effect

raises serious questions about the legitimacy of using the

standard resource allocation formula.

The findings suggest that the standard resource

allocation formula allocates resources to urban deprived

areas to a higher level than implied by morbidity alone.

This is because although the prevalence of conditions

such as CHD is characterised by a strong social

gradient, the demographic gradient is even stronger.

This is not to deny the association between poverty and

adverse health outcomes, or to reject the goal of

reducing health inequalities. However, it is equally

important to acknowledge that targeting more health

care resources at areas that have higher relative needs (as

expressed by indicators such as standardised mortality

ratios and premature disease) can shift resources away

from areas that have higher overall rates of morbidity

when the latter have older demographic profiles.

Discussion

Because epidemiological estimates yield direct mea-

sures of health status, there are strong grounds for

proposing that a morbidity-based model provides a

more legitimate basis for allocating health resources

than the use of indirect proxies such as health service

utilisation or deprivation. Despite the perceived legiti-

macy of this approach, there has been little practical

testing of such a model of resource allocation.

In this paper, we have illustrated a methodology by

which symptom-based capitation estimates for the

treatment of CHD can be derived for PCTs. The

prevalence rates that we have extrapolated are drawn

from a national sample comprising more than 36,000

individuals, a sample size that is sufficiently large to

reliably capture demographic and social gradients in

disease. The method is sensitive to the socio-economic

characteristics of PCT populations, addressing the aim

of the York model to reflect the demand for health care

resources produced by deprivation. Because the method

considers the role of age, sex and class simultaneously

when assessing need, the health care demands of elderly

populations are also appropriately reflected.

Despite these important advantages, there are remain-

ing practical issues that would need to be addressed

before a morbidity-based capitation methodology could

be implemented.

Practical issues

First, further research would be necessary to establish

the full range of epidemiological survey work, published

and otherwise, that is sufficiently robust to be used to

derive socially and demographically sensitive prevalence

matrices relative to relevant clinical programme areas.

The Health Survey for England (HSE) reports on a wide

range of diseases, health-related conditions and health

service use and comprises a range of measures

ARTICLE IN PRESSS. Asthana et al. / Social Science & Medicine 58 (2004) 539–551 547

(self-reported, doctor-diagnosed, symptom-based) from

which to derive prevalence rates. Because the HSE

revisits key areas of morbidity, it is also possible to

incorporate data from a number of constituent surveys

and thereby achieve large sample sizes. The HSE

provides particularly detailed data on cardiovascular

disease. For other classes of disease, however, it may not

be the most detailed source of epidemiological data. In

such cases, it may be more fruitful to explore alternative

sources. However, it is also worth noting that because

the HSE is an ongoing project, future rounds of the

survey could be designed to either address specific gaps

or to bolster sample sizes relative to key areas of

morbidity.

Second, the most appropriate mapping(s) between

clinical activities and the descriptions of morbidity

available through epidemiological surveys such as the

Health Survey for England must be established. Our

own analysis suggests that a single mapping is unlikely

to provide a sufficient guide to the variety of treatment

and preventative activities which take place within a

given clinical programme area. In our sample, for

instance, the prevalence of people with Angina and

self-reported ‘very-bad health’ offers a better prediction

of inpatient admissions for Angina (ICD-10 I20) than

does the prevalence of Angina alone. However, it is the

prevalence of Angina alone that provides the better

prediction of Cardivascular disease drug prescribing by

GPs. The need for preventative activities relative to

Angina, meanwhile, is possibly best estimated with

reference to the prevalence of population risk-factors

such as alcohol consumption, smoking, eating habits,

body mass, blood pressure, blood analytes and family

history of CHD mortality (which can also be derived

from the HSE).

This example suggests that the relationship between

diagnosis, problem and need for services is not at all

straightforward. Even in cases where the problem is

fairly clear-cut and identical, different treatments may

be equally appropriate. Against this, there is growing

emphasis within the UK on specifying care pathways for

particular conditions. This move towards the provision

of specific, detailed and fully transparent clinical guide-

lines should facilitate the development of consistent

definitions of ‘clinical need’ across disease categories.

Taken in conjunction, the use of ‘bottom-up’ morbidity

estimates and ‘top–down’ clinical guidelines could allow

the resource algorithm to escape a self-perpetuating

reliance on historic data.

A third issue that must be raised when assessing the

potential of a morbidity-based capitation methodology

concerns its applicability to specific clinical areas. As

part of its drive to improve national standards of health

care, the UK government has identified key service areas

for which explicit standards of quality, effectiveness and

access are expected. The development of specific clinical

programme budgets for CHD, Mental Health and

Diabetes would support the pursuit of improvements

in these ‘National Service Framework’ areas, not least

by bringing together community, primary and secondary

resources into a unified budget. This could promote

better integration between prevention, management and

cure. The fact that a morbidity-based capitation

methodology lends itself to such a purpose is thus a

particular strength. However, the usefulness of the

approach in estimating general health service needs is

open to question.

Fundamental issues

In this paper, we have presented evidence that

suggests that compared to the current English formula

for HCHS, a morbidity-based capitation methodology

would result in a significant shift of hospital resources

away from deprived areas, towards areas with older

demographic profiles and towards rural areas. The

findings not only challenge a widely held assumption

that, because morbidity-based estimates will capture

dimensions of health inequality that traditional proxy

measures cannot, they will result in the better targeting

of health resources towards deprived areas (NHS Wales,

2001). They also raise questions about the desirability of

introducing a morbidity-based approach to resource

allocation. Most health inequalities researchers and

policy makers in Britain are concerned to direct more

health care resources towards the urban poor. Thus, it is

likely that any resource allocation system that would

lead to a shift of resources away from deprived areas

would be regarded as a retrogressive step.

Against this background, it is important to note that

per capita allocations in rural areas are significantly

smaller than those in urban areas. For example, whilst

some parts of Central London receive over d950 per

person, some of the most rural areas receive less than

d600 (White, 2001, p. 16). Despite this, the possibility

that rural areas serving demographically older popula-

tions may have a legitimate need for more resources (on

the basis of their populations’ higher levels of morbidity)

is rarely given serious consideration. This may reflect an

implicit assumption that, because urban deprived

populations suffer profound health inequalities, they

have a greater claim on existing health resources.

However, whilst health inequalities should quite appro-

priately be targeted for action, such an assumption

reflects a lack of clarity about the purpose of health

resource allocation.

The central aim of the current resource allocation

system is to allocate resources to geographical areas in

order to secure equal opportunity of access for equal

needs. The concept of health care equity has thus

underpinned the approach to resource allocation in the

NHS (though it must be acknowledged that equity in

ARTICLE IN PRESSS. Asthana et al. / Social Science & Medicine 58 (2004) 539–551548

resource distribution does not guarantee equity in

access). The Advisory Committee on Resource Alloca-

tion (ACRA) has been exploring the possibility of

introducing a new equity criterion to the system of

resource allocation (Hauck, Shaw, & Smith, 2002). To

this end, it is interested in developing a method of

allocating resources that will contribute towards achiev-

ing equal health outcomes (i.e. reducing health inequal-

ities between the most advantaged and least advantaged

groups). However, this does not imply a departure from

the principle of equal opportunity of access to equal

need. Health care equity is to remain a core objective.

There is recognition, however, that the persistence of

health inequalities demands policy action and that some

resources should also be targeted at the objective of

health equity.

For the foreseeable future, resource allocation for

core services will continue to be based on the traditional

utilisation model. However, if the distribution of core

services should equitably reflect the existing burden of

disease, the current formula appears to be deeply flawed.

It incorporates several elements that result in a shift of

resources away from populations that have higher rates

of overall morbidity toward populations that have

higher rates of relative health need (according to

standardized measures and indicators of premature

morbidity and mortality). The latter populations would

be expected to be targeted by any new budget addressing

health inequalities. However, there is no explicit

rationale for such targeting of core services.

The fact that the current formula allocates resources

to areas that would be expected to eventually benefit

from a health inequalities budget may appear at first

sight to be unproblematic. However, the goals of health

care equity and health equity require very different

policy responses. It is generally agreed that the NHS

(and particularly national hospital services) has relatively

little to contribute towards the reduction of health

inequalities compared to other sources of variation such

as income distribution, education, housing and lifestyle.

Thus, the targeting of core services to urban deprived

populations over and above levels of underlying

morbidity is likely to be an ineffective (and perhaps

co-opting) response to health inequalities. It is one,

moreover, that introduces a new form of inequity by

under-estimating the needs of more elderly but less

deprived populations.

It is, of course, essential that additional health care

needs associated with deprivation are met in allocations

for core services. By attaching weightings to morbidity-

based capitations in a way that reflects how specific

conditions are socially distributed, a morbidity-based

capitation model does just this. It has the potential to do

more. The very transparency of the morbidity-based

approach demands equal transparency in how the

balance is struck between the treatment of existing

disease on the one hand and the reduction of inequalities

in health on the other. If urban areas with younger

demographic profiles lose core resources on the basis of

the fact that they have lower overall rates of morbidity,

an explicit response to their higher relative needs will be

required. The distinction between addressing absolute

and relative needs will have to be explicit in the

allocation of resources to local areas. This, in turn, will

require a real political commitment to the public health

and broader social interventions and initiative that can

address the socio-economic factors that give rise to

health inequalities in the first place.

Acknowledgements

We thank the public health and information depart-

ments of the seven participating health authorities for

providing the study with GP registration and inpatient

data. Health Survey for England data were supplied via

MIMAS with the permission of the ESRC Data

Archive. Socio-economic profiling of PCT populations

was undertaken using 1991 Census data, Crown Copy-

right, ESRC Purchase. This work uses boundary

material which is copyright of the Crown and the ED-

LINE Consortium and has been funded by the

Economic and Social Research Council’s Health Varia-

tions Programme (Reference L128251031).

Thanks too for the detailed and very helpful

comments made by the anonymous reviewers about

the first draft of this paper.

References

ACRA (1999). A brief history of resource allocation in the NHS,

1948–98. Advisory Committee on Resource Allocation,

Department of Health.

Age Concern (1999). Turning Your Back on US. Older People

and the NHS. London: Age Concern.

Asthana, S., Gibson, A., Dicker, J., Moon, G., & Brigham, P.

(2001a). The use of direct health estimates as a basis for

resource allocation: a response to the Welsh Assembly’s NHS

Resource Allocation Review. University of Plymouth,

unpublished report.

Asthana, S., Gibson, A., Moon, G., Brigham, P., & Dicker, J.

(2001b). Inequalities in health care: inverse care in patterns

of service uptake. Paper Presented at the 9th Annual

Conference of the UK Public Health Association, Bourne-

mouth International Centre, 27–29th March.

Beech, R., Bevan, G., & Mays, N. (1990). Spatial equity in the

NHS: the death and re-birth of RAWP. In: A. Harrison, &

Bruscini (Eds.), Health Care UK 1990: An Annual Review of

Health Care Policy. (pp. 44–61) London: Kings Fund

Institute.

Ben-Shlomo, Y., & Chaturvedi, N. (1995). Assessing equity in

access to health care provision in the UK: Does where you

live affect your chances of getting a coronary artery bypass

ARTICLE IN PRESSS. Asthana et al. / Social Science & Medicine 58 (2004) 539–551 549

graft? Journal of Epidemiology and Community Health, 49,

200–204.

Bentham, C. G., & Haynes, R. (1986). A raw deal in remoter

rural areas? Family Practitioner Services, 13, 84–87.

Bevan, G. (1998). Taking equity seriously: A dilemma for

government from allocating resources to primary care

groups. British Medical Journal, 316, 39–42.

Black, N., Langham, S., & Petticrew, M. (1995). Coronary

revascularisation: Why do rates vary geographically in the

UK? Journal of Epidemiology and Community Health, 49,

408–412.

Bojke, C., Gravelle, H., & Wilkin, D. (2001). Is bigger better for

primary care groups and trusts? British Medical Journal,

322, 599–602.

Bowling, A. (1999). Ageism in cardiology. British Medical

Journal, 319, 1353–1355.

Carr-Hill, R., Hardman, G., Martin, S., Peacock, S., Sheldon,

T., & Smith, P. (1994). A formula for distributing NHS

revenues based on small area use of hospital beds. University

of York: Centre of Health Economics.

Carr-Hill, R., Hardiman, G., Martin, S., Peacock, S., Sheldon,

T., & Smith, P. (1997). A new formula for distributing

hospital funds in England. Interfaces, 27, 53–70.

Davey-Smith, G., Hart, C., Watt, G., Hole, D., & Hawthorne,

V. (1998). Individual social class, area-based deprivation,

cardiovascular disease risk factors and mortality: The

Renfrew and Paisley study. Journal of Epidemiology and

Community Health, 52, 399–405.

Davey-Smith, G., Shipley, M. J., & Rose, G. (1990). The

magnitude and causes of socio-economic differentials in

mortality: Further evidence from the Whitehall study.

Journal of Epidemiology and Community Health, 44,

265–270.

DETR (2000). Indices of Deprivation 2000. Regeneration

Research Summary 21. Department of Environment,

Transport and the Regions, August, 2000.

Eachus, J., Chan, P., & Pearson, N., et al. (1999). An additional

dimension to health inequalities: Disease severity and socio-

economic position. Journal of Epidemiology and Community

Health, 53, 603–611.

Erens, B., & Primatesta, P. (1999). Health survey for England,

cardiovascular disease, Vol. 2. London: The Stationary

Office.

Fearn, R. (1987). Rural health care: A British success or a tale

of unmet need? Social Science and Medicine, 24(3), 263–274.

Galatius-Jenson, S., Launberg, J., Mortensen, L. S., & Hansen,

J. F. (1996). Sex related differences in short and long term

prognosis after acute myocardial infarction: 10 year follow

up of 3073 patients in database of first Danish Verapamil

infarction trial. British Medical Journal, 313, 137–140.

Gibson, A., & Asthana, S. (2000). Estimating the socio-

economic characteristics of school populations using pupil

postcodes and census data: An appraisal. Environment and

Planning A, 32, 1267–1285.

Gibson, A., Asthana, S., Brigham, P., Moon, G., & Dicker, J.

(2002). Geographies of Need and the New NHS: Methodo-

logical Issues in the Definition and Measurement of the

Health Needs of Local Populations. Health and Place, 8(1)

Gift, T., & Zastowny, T. (1990). Psychiatric service utilisation

differences by sex and locale. International Journal of Social

Psychiatry, 36(1), 11–17.

Hauck, K., Shaw, R., & Smith, P. C. (2002). Reducing

unavoidable inequalities in health: a new criterion for

setting health care capitation payments. Health Economics.

Haynes, R., & Bentham, C. G. (1982). The effects of

accessibility on GP consultations: Outpatients attendances

and inpatient admissions in Norfolk, England. Social

Science and Medicine, 16, 561–569.

Hippisley-Cox, J., & Pringle, M. (2000). Inequalities in access to

coronary angiography and revascularization: The associa-

tion of deprivation and location of primary care services.

British Journal of General Practice, 50, 449–454.

Jones, A., Bentham, C. G., & Harrison, B., et al. (1998).

Accessibility and health service utilisation for asthma in Norfolk,

England. Journal of Public Health Medicine, 20, 312–317.

Lampe, F., Colhoun, H., & Dong, W. (1994). Cardiovascular

disease and respiratory conditions. Health survey for England,

1994. London: Joint Health Surveys Unit, University College.

Le Grand, J. (1991). The distribution of health care revisited.

Journal of Health Economics, 10, 239–245.

Manson-Siddle, C., & Robinson, M. (1998). Superprofile

analysis of socio-economic variations in coronary investiga-

tion and revacularization rates. Journal of Epidemiology and

Community Health, 52, 507–512.

Mays, N. (1995). Geographical resource allocation in the

English National Health Service, 1974–1994: The tension

between normative and empirical approaches. International

Journal of Epidemiology, 24, S96–102.

NHS Executive (1999). Resource Allocation: Weighted Capita-

tion Formulas. Resource Allocation and Fund Team,

Finance and Performance Directorate, National Health

Service Executive.

NHS Wales (2001). Targeting Poor Health: Professor Town-

send’s Report of the Welsh Assembly’s National Steering

Group on the Allocation of NHS Resources. Report to the

Health and Social Services Committee of the National

Assembly for Wales, 4th July 2001.

O’Donnell, O., & Propper, C. (1991). Equity and the distribu-

tion of NHS resources. Journal of Health Economics, 10(1),

1–19.

Payne, N., & Saul, C. (1997). Variations in the use of cardiology

services in a health authority: comparison of coronary

artery revascularization rates with prevalence of angina and

coronary mortality. British Medical Journal, 257–261.

Powell, M. (1997). Evaluating the national health service.

Buckingham: Open University Press.

Rice, N., Dixon, P., Lloyd, D., & Roberts, D. (2000).

Derivation of a needs based capitation formula for

allocating prescribing budgets to health authorities and

primary care groups in England: Regression analysis. British

Medical Journal, 320, 284–287.

Rice, N., & Smith, P. (1999). Approaches to capitation and risk

adjustment in health care: an international survey. ACRA

paper 09.

Rice, N., & Smith, P. (2001). Ethics and geographical equity in

health care. Journal of Medical Ethics, 27, 256–261.

Rose, G. A., & Blackburn, H. (1986). Cardiovascular survey

methods. World Health Organisation, Monograph, 56, 1–188.

Scrivener, G., & Lloyd, D. (1995). Allocating census data to

general practice populations—implications for study of

prescribing variation at the practice level. British Medical

Journal, 311, 163–165.

ARTICLE IN PRESSS. Asthana et al. / Social Science & Medicine 58 (2004) 539–551550

SHED (Scottish Executive Health Department) (1999). Fair

Shares for All: Report of the National Review of Resource

Allocation for the NHS in Scotland. Edinburgh: Scottish

Executive Health Department.

Sonke, G. S., Beaglehole, R., Stewart, A. W., Jackson, R., &

Stewart, F. M. (1996). Sex differences in case fatality before

and after admission to hospital after acute cardiac events:

Analysis of community based coronary heart register.

British Medical Journal, 313, 853–855.

Sheldon, T., Smith, G., & Bevan, G. (1993). Weighting in the

dark—resource allocation in the New NHS. British Medical

Journal, 306, 5.

Tudor-Hart, J. (1971). The inverse care law. Lancet i (27

February), 405–412.

Ward, P., Morton Jones, A. J., Pringle, M., & Chilvers, C.

(1994). Generating social class data in primary care. Public

Health, 108, 279–287.

Wenger, N. K. (1997). Coronary heart disease: An older

women’s major health risk. British Medical Journal, 315,

1085–1090.

White, C. (2001). Who gets what, where, and why? The NHS

Allocation system in England is failing rural and disadvan-

taged areas. Rural Health Forum and University of St

Andrews.

ARTICLE IN PRESSS. Asthana et al. / Social Science & Medicine 58 (2004) 539–551 551