general practitioner services in primary care groups in england: is there inequity between service...

8
Health & Place 7 (2001) 67–74 General practitioner services in primary care groups in England: is there inequity between service availability and population need? Deborah Baker*, Mark Hann National Primary Care Research and Development Centre, University of Manchester, 5th Floor, Williamson Building, Oxford Road, Manchester M13 9PL, UK Received 30 May 2000; accepted 24 October 2000 Abstract This study examined the coverage of minor surgery, child health surveillance and chronic disease management for asthma and diabetes in relation to population need and key organisational features of general practice in the 481 primary care groups (PCGs) in England. PCG-level summary scores were developed to estimate the relative availability of all four services and their relative importance in discriminating between high and low levels of service provision. The coverage of services was widespread and, in such circumstances, there was no systematic evidence of poorer service availability for PCGs with higher population need (the ‘inverse care’ law). Rather this relation was localised, being most predominant for PCGs covering London and its suburbs. In these PCGs, there was no association between indicators of lack of capacity, such as single-handed practice, and levels of service provision. # 2001 Elsevier Science Ltd. All rights reserved. Keywords: Primary care group; Population need; Chronic disease management; Child health surveillance; Minor surgery; Inverse care law Introduction One of the founding principles of the National Health Service in England on its inception in 1946 was to realise equity in health care provision. This was to be achieved by universal entitlement and by developing a service that was free at the point of use, the first point of contact being the general practitioner’s surgery (Powell, 1995; Glennerster, 1995; Whitehead, 1994). Whilst these ideals embodied the principles of achieving the same high standard of care for all, there has been evidence of variation in the provision and availability of primary health care such that those patients who are poorer and sicker are less likely to have access to the care that will meet their needs (the inverse care law) (Tudor Hart, 1971). This was highlighted in 1990, with the introduc- tion of a new contract for general practitioners in which wide geographical variations in standards of primary care were noted, with particular difficulties being observed in the inner cities (Department of Health and the Welsh Office, 1989). As a consequence, this contract introduced measures to encourage higher standards of care in general practice, including the more extensive coverage of health promotion and preventive medicine. General practitioners are contracted to provide general medical services to patients registered with them, rather than being directly employed by the NHS. In this context, these measures took the form of ‘incentive payments’ for providing specific services, over and above the basic core of health provision that is financed by the payment of capitation fees. Payment was made for achieving specific target levels of provision *Corresponding author. Tel.: +44-161-275-7606; fax: +44- 161-275-7600. E-mail address: [email protected] (D. Baker). 1353-8292/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved. PII:S1353-8292(00)00041-1

Upload: deborah-baker

Post on 19-Sep-2016

214 views

Category:

Documents


1 download

TRANSCRIPT

Health & Place 7 (2001) 67–74

General practitioner services in primary care groups inEngland: is there inequity between service availability and

population need?

Deborah Baker*, Mark Hann

National Primary Care Research and Development Centre, University of Manchester, 5th Floor, Williamson Building, Oxford Road,

Manchester M13 9PL, UK

Received 30 May 2000; accepted 24 October 2000

Abstract

This study examined the coverage of minor surgery, child health surveillance and chronic disease management forasthma and diabetes in relation to population need and key organisational features of general practice in the 481

primary care groups (PCGs) in England. PCG-level summary scores were developed to estimate the relative availabilityof all four services and their relative importance in discriminating between high and low levels of service provision. Thecoverage of services was widespread and, in such circumstances, there was no systematic evidence of poorer service

availability for PCGs with higher population need (the ‘inverse care’ law). Rather this relation was localised, being mostpredominant for PCGs covering London and its suburbs. In these PCGs, there was no association between indicators oflack of capacity, such as single-handed practice, and levels of service provision. # 2001 Elsevier Science Ltd. Allrights reserved.

Keywords: Primary care group; Population need; Chronic disease management; Child health surveillance; Minor surgery; Inverse

care law

Introduction

One of the founding principles of the National Health

Service in England on its inception in 1946 was to realiseequity in health care provision. This was to be achievedby universal entitlement and by developing a service that

was free at the point of use, the first point of contactbeing the general practitioner’s surgery (Powell, 1995;Glennerster, 1995; Whitehead, 1994). Whilst these ideals

embodied the principles of achieving the same highstandard of care for all, there has been evidence ofvariation in the provision and availability of primaryhealth care such that those patients who are poorer and

sicker are less likely to have access to the care that will

meet their needs (the inverse care law) (Tudor Hart,1971). This was highlighted in 1990, with the introduc-tion of a new contract for general practitioners in which

wide geographical variations in standards of primarycare were noted, with particular difficulties beingobserved in the inner cities (Department of Health

and the Welsh Office, 1989). As a consequence, thiscontract introduced measures to encourage higherstandards of care in general practice, including the more

extensive coverage of health promotion and preventivemedicine. General practitioners are contracted toprovide general medical services to patients registeredwith them, rather than being directly employed by the

NHS. In this context, these measures took the form of‘incentive payments’ for providing specific services, overand above the basic core of health provision that is

financed by the payment of capitation fees. Paymentwas made for achieving specific target levels of provision

*Corresponding author. Tel.: +44-161-275-7606; fax: +44-

161-275-7600.

E-mail address: [email protected] (D. Baker).

1353-8292/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.

PII: S 1 3 5 3 - 8 2 9 2 ( 0 0 ) 0 0 0 4 1 - 1

for vaccination and immunisation and for cervicalscreening, and for providing health promotion services

(including chronic disease management for asthma anddiabetes). Payments were also attached to the introduc-tion of services such as child health surveillance and

minor surgery that had previously been provided incommunity and secondary care settings. But researchsince 1990 has suggested that, whilst incentive paymentsincreased coverage, they did not necessarily ensure that

services became more equitably distributed in relation topopulation need. Some studies concluded that the newgeneral practitioner contract accelerated previous

trends, promoting investment in high-earning practices,serving affluent areas, where care is easier; otherssuggested that incentive payments did not break the

existing link between lack of capacity } particularlysingle-handed practices } and poor performance in thedelivery of services (Tudor Hart et al., 1991; Waller et

al., 1990; Baker and Klein, 1991; Gillam, 1992; Readinget al., 1994; Leese and Bosanquet, 1995; Lynch, 1995).Ten years on, a new era has dawned in the provision

of primary care, with the advent of primary care groups

(PCGs) in 1999. These are the 481 general practitionerand nurse led organisations that are responsible forplanning and developing services for their local popula-

tions (approximately 100,000 persons) within eachhealth authority in England. Recent government WhitePapers have set central policy objectives for the new

PCGs/Trusts that explicitly encompass health improve-ment and tackling inequality for their populations(Department of Health, 1998, 1999). The issue of equityin service provision is a core component of this agenda,

with the strong emphasis being placed on ‘fair access tohealth services in relation to people’s needs, irrespectiveof geography, class, ethnicity, age or sex’. It is in this

context that the research reported in this paper is set.We examined the coverage of minor surgery, child

health surveillance and health promotion (chronic

disease management) in relation to population needand key organisational features of general practice forpopulations of PCGs/Trusts. The focus of this study was

on the availability of these services rather than theachievement of target payments for immunisation andcervical screening, because the latter are essentiallymeasures of uptake as well as supply. They are likely to

reflect the health beliefs of patients and populationmobility, which could partially account for poorperformance in deprived inner city areas (Majeed et

al., 1994; Sutton et al., 1994; Jefferies et al., 1991). Theaims of the study were twofold. The first was to test forthe persistence of the inverse care law in circumstances

in which the coverage of chronic disease management,child health surveillance and minor surgery is known tobe widespread. The second was to develop a summary

measure of the distribution of these services relative toone another, so that the range of services available could

be compared between PCGs and between practiceswithin a PCG.

Methods

Aggregation of general practice data into primary care

groups

The main source of information about the activities of

general practitioners and the organisation and resour-cing of general practice is the bi-annually updatedGeneral Medical Services Statistics (GMS statistics),often referred to as the GP Census. Data are collected

for approximately 30,000 GP principals and 10,000practices in England and Wales. For the purpose of thisstudy we used the 1998 data set. The information

required for this study was aggregated into PCGs usingan early version of the National Database for PrimaryCare Groups and Trusts (www.npcrdc.man.ac.uk). This

is a database that has linked population socio-economicand demographic characteristics from the 1991 Censusto GMS statistics for all PCGs in England. Out of 9090practices, 8904 (98%) could be allocated to PCGs using

this database. The 2% loss was likely to be due toinaccuracies in practice postcodes.

Measures of service provision

For each practice, we recorded whether each of fourservices was available to patients registered at thatpractice. These services were: chronic disease manage-

ment for asthma, and similarly for diabetes, child healthsurveillance for the under-fives, and minor surgerysessions.

Practice characteristics

Practice characteristics were selected on the basis thatthey have, in previous analyses, been closely associated

with variations in the outputs of primary care. Thesecharacteristics were: whether the practice was single-handed, whether it was a training practice, whether therewas a female partner at the practice, and the average list

size per whole-time-equivalent (WTE)1 partner. Mea-sures of single-handedness and training status were thenderived for each PCG. These were the percentage of

practices (within the PCG) with these characteristics. Ameasure of the presence of a female general practitionerwas derived as the percentage of WTE principals who

1Information on GP principal time commitment is available

from the GP Census. Each GP is classified as either full-time,

three-quarter time, half-time or job share (assumed to be half-

time).

D. Baker, M. Hann / Health & Place 7 (2001) 67–7468

were female. The average PCG list size per WTE GP wasalso calculated.

Population need

A calculation derived from the Jarman Index (Jar-man, 1983) was used as a measure of population need,

since this index combines indicators of sources of needthat reflect the demographic characteristics of thepopulation (e.g. elderly living alone, children under five)

as well as their socio-economic circumstances. It is alsothe measure of need that is most consistently availablefor the registered populations of general practices. Wecalculated a PCG-level score from the banded

‘deprivation payments’ for each general practice thatare based on the Jarman Index. Payments as of 1October 1998 were £11.40 for heavily deprived, £8.55 for

medium and £6.55 for marginal. PCG-level needs scoreswere calculated as follows:

where summations are over all practices in the PCG.

Measuring levels of service provision in PCGs

All 481 PCGs in England were used in the analysis.Fifty (of 8904) practices were excluded from the

analysis, as their average list size (per WTE principal)was less than 200.In order to compare the levels of overall service

provision within PCGs, a score based on the availabilityof all four services was calculated using an item responsemodel. A one-parameter Rasch model (Rabe-Hesketh et

al., 2000) was fitted to the four binary service indicatorsusing the procedure GLLAMM,2 in the statisticalsoftware package STATA (version 6). A two-parametermodel was then fitted,3 which allowed us in addition to

estimate the relative importance of each service in itsability to discriminate between practices with differentlevels of provision. Using these ‘factor loadings’ or

‘weights’ it was thus possible to discriminate betweenpractices who offer the same number of services, wherethis is at least one, and fewer than all four. The greatest

weightings were assigned to asthma and diabetes,

indicating that the model identified these two servicesas being the most important when discriminating

between practices with the highest levels of serviceprovision, and those with the lowest. This was mostlikely to be because the provision of chronic disease

management for asthma and diabetes is likely to beclosely related, with one of the services only rarely beingoffered without the other. The model also estimated howmany (latent) classes of practice existed. In this case

there were two latent classes containing 93% and 7% ofpractices, which could be described as ‘better thanaverage’ and ‘poor’ respectively. Practice service scores

were estimated from the residuals of the model. Thesepractice scores were re-scaled so that they ranged from 0to 10, and were then averaged over all practices within a

PCG, to give a PCG-level service score out of 10. Thesescores were correlated with measures of practicecharacteristics and population ‘need’, for the whole of

England, and, for each of the eight regional healthauthorities in England, using Spearman’s rank correla-

tion coefficient, adjusted for tied ranks. A number of the

variables of interest were highly skewed, which sug-gested that the use of Pearson’s correlation would havebeen inappropriate.

Results

Table 1 shows that the coverage of the four servicesexamined in this study was widespread, with over 90%

providing chronic disease management for asthma, fordiabetes and child health surveillance. Seven thousandand twelve practices (79.2%) offered all four services,

1063 (12%) offered three, 496 (5.6%) offered two, 177(2%) one and 106 (1.2%) none.

PCG service scores

Table 2 shows the distribution of PCG service scores,

comparing the unweighted one-parameter model, whichjust reflects the level of availability of the four services,and the weighted two-parameter model, that weights theprovision of chronic disease management for asthma

and diabetes more highly than the other two services.One-hundred (21.2%) PCGs had all services provided

in all their practices. The average PCG score in the

unweighted model was 9.0, with 60% of PCGs havingan above- average score. Sixty-five (13.5%) PCGs scoredunder 8, well below average, with 24 (5%) scoring

under 7. The average PCG score in the weighted two-parameter model was 9.3, with, again, 60% of all PCGs

�11:40

Xpatients living in heavily deprived areasþ8:55

Xmediumþ 6:55

Xmarginal

� X.all registered patients;

2GLLAMM fits generalised linear latent and mixed models

(see Rabe-Hesketh et al., 2000).3A likelihood ratio chi-squared statistic of 819.01 (3 degrees

of freedom, p5 0.001) was observed when fitting this model,

suggesting that it is a better fit to the data than the one-

parameter alternative. The two-parameter model is therefore

more appropriate for this data set, and service scores obtained

from this model are likely to be more reliable.

D. Baker, M. Hann / Health & Place 7 (2001) 67–74 69

scoring above average, but a much smaller numberscoring under 8 (30 : 6.4%) and only 4 (0.8%) scoringunder 7. There was a strong regional bias amongst those

PCGs scoring under 8 in the weighted model, with 2 in 3located in London and its suburbs.As this model was found to be more reliable, only

these service scores will be used from this point forward.

Relationship between PCG service scores, practicecharacteristics and population need

Table 3 shows the correlation between PCG servicescores, population need and practice characteristics for

each of the eight English health regions. These resultsshow a stark contrast in patterns of correlation betweenLondon and the rest of the country, particularly thenorthern and northwest regions. In London the stron-

gest correlation was between population need and thePCG service score, indicating lower levels of serviceprovision in PCGs with a higher needs score

(r ¼ �0:41). But there was no relation between practicecharacteristics and PCG service score in London. Incontrast, in other health regions in England and, in

particular, the Northwest region, the strongest correla-tion was between a lower PCG service score and a higher

number of single-handed practices (r ¼ �0:56) and therewas a weak, although statistically significant correlation

with a lower number of training practices (r ¼ 0:26).There was no association between service scores andpopulation need in this region. A similar pattern was

observed in the Northern and Yorkshire region, theWest Midlands region and the Eastern region. Only inthe Trent region was there no association between PCGservice score and either population need or the selected

practice characteristics. These results must, however, beinterpreted with some caution, since despite the sig-nificance of the correlation coefficients, they are, for the

most part, weak.Further analysis of these data using regression

modelling to elaborate the relation between PCG service

scores, practice characteristics and population needwould not be meaningful, due to the skewed nature ofthese measures. We thus attempted to assess the

robustness of these findings using two further descriptiveanalyses. First, we considered the geographical locationof the 30 PCGs receiving the highest average deprivationpayments per capita and examined the relation between

this measure of population need and their correspondingweighted two-parameter service scores. In the second,we used case vignettes to observe patterns of service

provision in PCGs with contrasting ‘needs profiles’ indifferent parts of the country.Table 4 illustrates patterns of association between

levels of population need and PCG service scores. Itshows that high levels of need were not consistentlyassociated with lower levels of service provision and thatthis was predominantly a London-based phenomenon.

PCGs in the Manchester area (there are 3 listed) had(above) average service scores, as did PCGs from theWest Midlands (5). In contrast, London PCGs (13) did

not score as highly on the service scale; all PCGs hadbelow-average service scores.

Case vignettes

The following case vignettes illustrate patterns of

service provision within contrasting PCGs in the northand the south of England.Key: uw=unweighted score; w=weighted score;

£=deprivation payment.

South Camden (uw=6.22, w=8.37, £=6.05)South Camden is a London PCG with high popula-

tion need, serving a registered population of 123,013people. Its service scores are very low relative to themajority of other PCGs. There are 25 practices within

this PCG, 12 (48%) of which offer all four services (4offer 3, 7 offer 2, one each offer 1 and no services).Twenty-one (84%) practices offer a complete CDM

programme, and thus the low two-parameter servicescore is a result of the other four practices, three of

Table 1

Level of service provision in 8854 general practices in England

Practices providing service

Service Number Percentage

On-site care for asthma 8264 93.3

On-site care for diabetes 8245 93.1

Child health surveillance 8176 92.3

Minor surgery sessions 7712 87.1

Table 2

Frequency of unweighted and weighted PCG scores

One-parameter

model

Two-parameter

model

(weighted)

PCG ‘Service’ score Number (%) Number (%)

(all services in

all practices)

Exactly 10

102 (21.2) 102 (21.2)

9.50–9.99 100 (20.8) 162 (33.7)

9.00–9.49 103 (21.4) 104 (21.6)

8.00–8.99 111 (23.1) 83 (17.3)

7.00–7.99 41 (8.5) 26 (5.4)

Less than 7.00 24 (5.0) 4 (0.8)

D. Baker, M. Hann / Health & Place 7 (2001) 67–7470

which do not offer either asthma nor diabetes care, the

other offering only on-site care for diabetes. Of the threeformer practices, one is a multi-handed training practice,with an average list size (per WTE GP) of only 779. The

other two are relatively less well off, both being single-handed, non-training practices, with list sizes of around3000, and very high deprivation payments (£9.82 and

£8.32, respectively).

Manchester East (uw=9.12, w=9.99, £=6.05)Manchester East PCG is in the Northwest region. It is

somewhat of a contrast to South Camden } it too has ahigh level of population need, but also a high level ofservice provision. It serves a registered population of

122,159 people. Of the 25 practices, 16 (64%) offerall four services. Of the remaining 9 practices, all hada uw practice level service score that was belowaverage (mean=7.58) but an above-average w score

(mean=9.97). These different scores arose because thesepractices did not offer minor surgery sessions (8) or childhealth surveillance (1), but they all offered chronic

disease management for asthma and diabetes. All thesepractices had high levels of population need (meandeprivation payment per capita=£7.40); 5 were single-

handed, and of these single-handed practices 3 had listsizes over 2545.

Bradford City (uw=8.21, w=9.89, £=6.99)

Bradford is in the Northern and Yorkshire region. Itis another example of a PCG with high levels ofdeprivation, but good service provision. It is quite a

large PCG, serving a registered population of 141,950and with 41 practices. The characteristics of many of thepractices within this PCG are quite unfavourable; 27

(66%) are single-handed, only 1 is a training practice,and the average PCG list size per WTE GP is nearly

Table 3

Regional service scores and the correlation of PCG ‘service’ scores with practice characteristics and population need by regiona

Correlation of service scores with

Regional

service score

‘Single-

Handedness’

measure

‘Training’

measure

‘Female WTE’

measure Average list size

Population

need

England �0.49 (50.01) 0.36 (50.01) �0.17 (50.01) �0.38 (50.01) �0.31 (50.01)

Northern and

Yorkshire region

94.90 �0.34 (0.01) 0.34 (0.01) 0.00 (0.99) �0.20 (0.14) �0.04 (0.75)

Northwest region 94.00 �0.56 (50.01) 0.26 (0.03) 0.01 (0.92) �0.10 (0.42) �0.12 (0.35)

Trent region 94.91 �0.17 (0.22) 0.07 (0.65) 0.03 (0.84) �0.05 (0.73) �0.08 (0.57)

West Midlands region 94.13 �0.27 (0.04) 0.35 (50.01) �0.10 (0.43) �0.18 (0.16) �0.19 (0.14)

Eastern region 97.03 �0.36 (0.01) 0.28 (0.04) �0.05 (0.75) �0.21 (0.13) �0.10 (0.50)

Southeast region 95.55 �0.39 (50.01) 0.30 (50.01) 0.14 (0.22) �0.24 (0.03) �0.22 (0.05)

Southwest region 97.39 �0.41 (50.01) 0.24 (0.09) �0.24 (0.09) �0.35 (0.01) �0.25 (0.08)

London region 84.72 0.01 (0.92) �0.03 (0.84) 0.05 (0.69) �0.17 (0.17) �0.41 (50.01)

aFigures in parentheses are significance levels.

Table 4

The 30 most deprived PCGs and their service scores

PCG

Deprivation level

(maximum 11.40)

Two-parameter

service score

Tower Hamlets 9.49 7.79

Birmingham-5 8.53 9.99

Southwark North 7.72 7.52

City and Hackney 7.47 7.79

Bradford City 6.99 9.89

Newham 6.99 8.79

Lambeth North 6.91 7.42

Birmingham-8 6.75 9.55

Tottenham 6.28 7.03

Birmingham-4 6.21 8.63

Manchester East 6.05 9.99

South Camden 6.05 8.37

Manchester West 5.99 9.62

South Islington 5.79 6.61

Central West Liverpool 5.50 9.18

Bristol Inner City 5.25 8.54

North Kensington

and Paddington

5.16 8.66

Southwark South 4.82 8.40

Manchester North 4.71 9.71

North Islington 4.43 9.35

Birmingham-6 4.19 8.79

South Walsall 3.87 9.98

Nottingham City

Central

3.79 9.36

North West Liverpool 3.76 9.16

Brent Central 3.76 9.13

City Central Leicester 3.70 7.23

Hammersmith 3.67 7.65

Salford East 3.62 8.24

Nottingham City:

South and East

3.51 8.46

Rochdale 3.34 8.13

D. Baker, M. Hann / Health & Place 7 (2001) 67–74 71

2200. Twenty-two (54%) practices offer all four services,including 11 single handers. Fifteen of the remainder

offer 3 services, with the other four offering 2. The largenumber of practices not offering all services is why theuw score is relatively low. However, the w score is very

high, and this is because 40 of the 41 practices offer acomplete CDM programme; the other practice notoffering on-site care for asthma.

St. Albans (uw=8.25, w=8.18, £=0.00)This PCG is in the Eastern region and has 11 practices

which serve a population of 86,460. The practices havefairly favourable characteristics (no single handers,

average of 4 WTEs per practice, 7 training practices)and none of them receive deprivation payments for theirpatients. Nevertheless, the provision of primary care

services is, on average, very poor. Two practices do notoffer any CDM services, which appears to be a smallnumber, but in such a small PCG, is a relatively high

proportion. The other 9 practices offer all 4 services.

Trafford North (uw=6.51, w=7.20, deprivationpayment=£0.30)

Trafford North is a PCG in the Northwest region inwhich there is no evidence of high levels of populationneed. The provision of services is nevertheless relativelypoor, seemingly because of the high number of single-

handed practitioners. There are 23 practices in TraffordNorth PCG, serving a registered population of 95,158.Fourteen (61%) of these practices are single-handed and

only 1 is a training practice. Of the single-handedpractices, only 3 offered the full range of services and 3did not offer any of the services at all } none of these

practices received deprivation payments. Eight (57%) ofthese 14 single-handed practices failed to providecomplete chronic disease management services.

South Islington (uw=4.13, w=6.61, £=5.79)South Islington is a London PCG serving a popula-

tion of 71,890. There are 15 practices in South Islington,

4 (27%) of which are single-handed. All the practices inSouth Islington receive deprivation payments for theirregistered patients (range=£4.54–£7.47). Only 5 (33%)of the practices deliver the full range of services; of the

remaining 10, 3 practices do not provide any of theservices and 2 practices provide no chronic diseasemanagement. There is no particular relation in this PCG

between the availability of services and list size ornumber of partners. Four of the 10 practices that do notoffer the full range of services are single-handed, but

partnership size in the other 6 practices ranges from 2 to6; none of them are training practices.

Discussion

In the late 1990s, the coverage of chronic diseasemanagement, minor surgery and child health surveil-lance in PCGs in England is widespread. This study

found some evidence of the ‘inverse care’ law in theprovision of these services to the populations of PCGs,but this tended to be geographically specific rather thana systematic pattern across the country. The relation

between high levels of population ‘deprivation’, asmeasured by the Jarman Index, and relatively low levelsof availability for all four services was most apparent in

London PCGs with predominantly disadvantaged po-pulations. In contrast, similarly disadvantaged PCGpopulations in the Midlands (Birmingham) and in the

North (Manchester) had above-average levels of serviceprovision. This confirms the findings of some previouslocality-based studies carried out in Manchester in the

early 1980s and in Glasgow in the early 1990s. Thesestudies found no evidence that general practice indeprived inner city areas was poorer than in moreaffluent parts of the city (Wood, 1983; Wilkin et al.,

1984; Wyke et al., 1992).It appears from the findings in this study that the

geographical differences observed were related to under-

lying variation in the link between practice character-istics, particularly single-handedness, and levels ofservice provision. In London poorer levels of service

provision were not associated with single-handedpractice, but in PCGs outside London, and particularlyin the northwest, there was a weak to moderatecorrelation between them. This suggests that inequity

in the provision of primary care services in LondonPCGs will not necessarily be resolved by increasingpartnership size. Further work is required, using a wider

range of practice characteristics to examine whetherother practice ‘inputs’ are more closely associated withlevels of service provision in these PCGs. For example,

the availability of chronic disease management could bemore closely related to the number of nurses employedby the practice, rather than the number of doctors, since

clinics for the management of asthma and diabetes arecommonly nurse led (Barnes et al., 1994; Neville et al.,1996).

Methodological issues

PCG service score

A central part of this study was to develop a measurethat could summarise the presence or absence of a rangeof services at practice and at PCG level and also be able

to measure the distribution of these services in relationto one another. One- and two-parameter item responsemodels were used for this purpose. In the two-parameter

model, the provision of chronic disease management forasthma and diabetes was identified as discriminating

D. Baker, M. Hann / Health & Place 7 (2001) 67–7472

best between high and low levels of service provision andas a consequence, these services attracted a higher

weighting in the practice and PCG scores. A higherweighting of these two services made conceptual sense inthis study, because they are likely to be of most

relevance for health improvement for locality-basedpopulations of PCGs. But weights determined statisti-cally may not always be the best way of discriminatingbetween levels of provision for different services. There

may be good theoretical justification for weightingaccording to the importance of a service for achievingpolicy objectives, but this method would not have this

degree of flexibility, since weights cannot be forced intothe model.The main advantage of the method is that it produces

a score for PCGs and a score for each practice within aPCG allowing identification of variation both betweenand within PCGs along the same dimensions.

Measuring population needMuch has been written about the conceptual and

methodological limitations of the Jarman Index as a

measure of population need for use in primary care andthese will not be rehearsed again in this paper. The mainpoint of relevance for this study is that this measure does

incorporate both demographic and socio-economicsources of need, thus acknowledging that ‘need’ is notentirely defined by ‘deprivation’. But this also means

that the index cannot be regarded as a precise measureof deprivation (Hutchinson et al., 1987). Moreover,when the indicators forming the index are dis-aggre-

gated, they vary in the strength and even in the directionof the relation they have with measures of theavailability and uptake of primary care services (Bakeret al., 1994). A further disadvantage is that such an

index constrains any allowance for need to a predeter-mined set of variables and a weighting that may not beappropriate for the specific purposes concerned (Carr-

Hill, 1999). More work is required to develop a measurethat can reflect more precisely the balance of differentsources of need in locality-based populations.

Conclusions and implications for future research

This study shows that, in the late 1990s, coverage ofchronic disease management for asthma and diabetes,child health surveillance and minor surgery in general

practice is widespread. In these circumstances, there waslittle systematic evidence of a strong relation betweenservice provision and population need; rather the

‘inverse care law’ appeared to be geographically specificand particularly pronounced for London PCGs. Poorlevels of service provision in these PCGs were however

no longer related to factors indicative of lack ofcapacity, such as single-handed practice, although this

was the pattern in other parts of the country. In the lightof geographical variation in the provision of primary

care observed in this study, policies directed at ensuringthe equitable provision of primary care services may wellbe more effective if they are devised at PCG or at health

authority level, to address local needs. Future researchcould productively focus on more localised studies,learning lessons from PCGs where the supply of primarycare services is meeting high levels of population need.

The ‘service scores’ developed in this study wouldprovide a means for PCGs to examine the balance ofservice provision for practices within their boundaries

and their relation to a range of practice inputs.

References

Baker, D., Klein, R., 1991. Explaining outputs of primary care:

population and practice factors. British Medical Journal 303,

225–229.

Baker, D., Klein, R., Carter, R., 1994. The impact of the 1990

contract for general practitioners on night visiting. British

Medical Journal 44, 68–71.

Barnes, G., Partridge, M.R., Organisation of Care Working

Group of the National Asthma Task Force, 1994. Commu-

nity asthma clinics: 1993 survey of primary care by the

National Asthma Task Force. Quality in Health Care 3,

133–136.

Carr-Hill, R., 1999. The Need for General Medical Services: a

Literature Review. Centre for Health Economics, University

of York, pp. 1–73.

Department of Health and the Welsh Office, 1989. General

Practice in the National Health Service: a new contract.

HMSO, London.

Department of Health, 1998. The New NHS: Modern,

Dependable. HMSO, London.

Department of Health, 1999. Saving Lives: Our Healthier

Nation. HMSO, London.

Gillam, S., 1992. Provision of health promotion clinics in

relation to population need: another example of the inverse

care law? British Journal of General Practitioners 42, 54–56.

Glennerster, H., 1995. British Social Policy since 1945. Black-

well, Oxford.

Hutchinson, A., Foy, C., Smythe, J., 1987. Providing Census

data for general practice: feasibility. Journal of the Royal

College of General Practitioners 37, 448–450.

Jarman, B., 1983. Identification of underprivileged areas.

British Medical Journal 286, 1705–1709.

Jefferies, S., McShane, S., Oerton, J., Victor, C.R., Beardow,

R., 1991. Low immunisation uptake rates in an inner city

health district: fact or fiction? Journal of Public Health

Medicine 13, 312–317.

Leese, B., Bosanquet, N., 1995. Change in general practice and

its effects on service provision in areas with different

socioeconomic characteristics. British Medical Journal 311,

546–550.

Lynch, M., 1995. Effect of practice and patient population

characteristics on the uptake of childhood immunisations.

British Journal of General Practitioners 45, 205–208.

D. Baker, M. Hann / Health & Place 7 (2001) 67–74 73

Majeed, A., Cook, D., Anderson, H.R., Hilton, S., Bunn, S.,

Stones, C., 1994. Using patient and general practice

characteristics to explain variation in cervical smear uptake

rates. British Medical Journal 308, 1272–1276.

Neville, R.A., Hoskins, G., Smith, B., Clark, R.A., 1996.

Observations on the structure, process and clinical outcomes

of asthma care in general practice. British Journal of General

Practitioners 46, 583–587.

Powell, M., 1995. The strategy of equality revisited. Journal of

Social Policy 21, 145–163.

Rabe-Hesketh, S., Pickles, A., Taylor, C., 2000. Generalised

linear latent and mixed models. Stata Technical Bulletin 53,

47–57.

Reading, R., Colver, A., Openshaw, S., Jarvis, S., 1994. Do

interventions that improve immunisation uptake also reduce

social inequalities in uptake. British Medical Journal 308,

1142–1144.

Sutton, S., Bickler, G., Sancho-Aldridge, Saidi, G., 1994.

Prospective study of predictors of attendance for breast

screening in Inner London. Journal of Epidemiology and

Community Health 48, 65–73.

Tudor Hart, J., 1971. The inverse care law. Lancet i, 405–412.

Tudor Hart, J., Thomas, C., Gibbons, B., Edwards, C., Hart,

M., Jones, J., Jones, M., Walton, P., 1991. Twenty five years

of case finding and audit in a socially deprived community.

British Medical Journal 302, 1509–1513.

Waller, D., Agass, M., Mant, D., Coulter, A., Fuller, A., Jones,

L., 1990. Health checks in general practice: another example

of inverse care? British Medical Journal 300, 1115–1118.

Whitehead, M., 1994. Is it fair? Examining the equity

implications of NHS reforms. In: Robinson, R., Le Grand,

J. (Eds.), Evaluating the NHS Reforms. Kings Fund

Institute, London, pp. 208–241.

Wilkin, D., Metcalfe, D., Hallam, L., et al., 1984. Area

variations in the process of care in urban general practice.

British Medical Journal 289, 229–232.

Wood, J., 1983. Are the problems of primary care in inner cities

fact or fiction? British Journal of General Practitioners 42,

271–275.

Wyke, S., Campbell, G., MacIver, S., 1992. Provision of,

and patient satisfaction with, primary care services in a

relatively affluent area and a relatively deprived area

of Glasgow. British Journal of General Practitioners 42,

271–275.

D. Baker, M. Hann / Health & Place 7 (2001) 67–7474