Relationship Between General Practice Capitation Funding and The Quality of Primary Care in England: A Cross-Sectional, 3 - Year Study
Relationship Between General Practice Capitation Funding and The Quality of Primary Care in England: A Cross-Sectional, 3 - Year Study
nurses and other staff) as explanatory variables in the urban practices. STATA V.14 (StataCorp) was used for all
model because staffing is likely to be directly affected by statistical analyses.
practice capitation funding and so inclusion of these vari-
able may lead to an underestimate of the full effect of Patient involvement
capitation funding. Moreover, a major change in the way Funding for this study included funding of a dedicated
in which staffing data were collected in 2015/2016 would patient involvement group. Patients were involved in
have resulted in a large reduction in observation number. developing plans for the study design, approving the
outcome measures and commenting on the potential
Practice capitation impact of outcomes. A lay summary was also provided.
Practice capitation funding depends on the total number
of practice patients adjusted to reflect factors affecting
GP workload (age, gender, patients in nursing and Results
residential homes, small area measures of morbidity), Summary statistics for the main characteristics of the
rurality and an index of local staff costs which affect the general practices are provided in table 2. Mean practice
cost of providing services.10 Data were available for the capitation funding per registered patient increased from
financial years 2014/2015, 2015/2016 and 2016/2017.22 £77.49 in 2014/2015 to £83.17 in 2016/2017 (table 3).
We use the mean capitation payment per patient for the The mean capitation funding per patient across the CQC
year prior to inspection and the year in which the practice inspection period was £79.48. The SD of the mean capita-
was inspected. tion funding per patient was £22.00.
The distribution of practice ratings across each quality
domain is shown in figure 1. A total of 79% (n=5774) of
Sample
practices achieved an overall rating of ‘good’, while only
We linked inspected practices (n=7310) with funding
4% (294) achieved an overall rating of ‘outstanding’.
data for their year of inspection. We excluded atypical
‘Inadequate’ ratings varied across the domains, from 1%
practices with ≤750 registered patients (n=10) or ≤500
(caring domain) to 6% (safety domain) and 4% (overall).
patients per FTE GP (n=8) following a previously used
Figure 2 shows the difference in capitation funding for
method.23 Practices with recorded negative (n=2) or zero
practices with the lowest quality rating compared with
funding (n=52) were excluded. The final analysis sample
those with the highest quality rating. In each domain,
consisted of 7238 practices.
‘inadequate’ practices received less capitation funding.
Using an independent group t-test, this difference was
Data analysis found to be significant for three domains (caring, safe
Analysis was conducted at GP practice level. Since the and well led) and for the overall practice rating.
CQC rating outcomes are ordered categories we used Table 4 reports the ORs on capitation funding per
ordered logistic regression to model the relationship patient estimated from four regression models of overall
between funding and the practice CQC ratings.24 Sepa- practice CQC rating. The OR on capitation funding per
rate models were estimated for each domain. patient is reported in (SD units). In the first model, capita-
The key explanatory variable was capitation funding tion funding is the only explanatory variable (unadjusted
per patient (measured in SD units). We also include model); remaining models are adjusted for inclusion of
patient and practice characteristic covariates, thereby successive additional explanatory variables: year effects,
reducing the risk of bias from the omission of variables patient characteristics and practice characteristics. The
which might affect the CQC rating and are correlated unadjusted model shows an association between higher
with practice capitation funding. The regression models capitation funding and higher overall CQC ratings with
included year effects to allow for inspection year and an OR of 1.09 (95% CI 1.03 to 1.15). Allowing for the year
annual general practice funding uplifts. We accounted of inspection increased the OR slightly to 1.10 (95% CI
for local area effects by adjusting for clustering at clinical 1.04 to 1.16). Additional allowance for patient character-
commissioning group level. Multicollinearity was tested istics (OR 1.13, 95% CI 1.06 to 1.19) and practice charac-
for by calculating the variance inflation factor (VIF) and teristics (OR 1.13, 95% CI 1.06 to 1.19) further increased
variables with a value for VIF >5 were excluded. The the OR. The number of observations in table 4 fell
proportional odds assumption of the ordered logit model from 7168 to 7045 because of missing data. Very similar
was also tested.25 We report the OR from the estimated changes in ORs across the models were observed when all
models. models were restricted to equal sample sizes. A likelihood
We calculated the average marginal effects of funding ratio test demonstrated that the addition of patient and
on the predicted probabilities of achieving overall ratings practice variables create a statistically significant improve-
of ‘outstanding’ and ‘inadequate’ for all practices. We also ment in model fit, confirming that higher ORs were asso-
compared the predicted probabilities of an ‘outstanding’ ciated with the addition of model variables, rather than to
overall rating at different practice capitation funding a change in sample size.
levels for training versus non-training practices, single- The final adjusted model indicates that for a 1 SD
handed versus multihanded practices and rural versus increase in capitation funding, the odds of achieving an
Figure 1 Distribution of CQC ratings by each domain. CQC, Care Quality Commission.
Figure 2 Practice capitation funding by overall practice rating: ‘inadequate’ versus ‘outstanding. Difference in practice
capitation funding between practices rated 'Inadequate' versus 'Outstanding for each domain. Caring Domain £80.84 vs
£91.14, p<0.001. Effective Domain £80.14 vs MAD, No significant difference. Responsive Domain £78.48 vs £83.82, No
significant difference. Safe Domain £77.69 vs 90.11, p<0.05. Well-led Domain £78.48 vs £87.82, P<0.05. Overall Domain £78.47
vs 87.82, P<0.001.
Table 4 Association of capitation funding per patient with overall practice CQC rating: unadjusted and adjusted regression
models
Model 1 Model 2 Model 3 Model 4
Capitation funding OR† 1.09** 1.10** 1.13*** 1.13***
95% CI of OR 1.03, 1.15 1.04, 1.16 1.06, 1.19 1.06, 1.19
Observations 7168 7168 7144 7045
Models contain
Year effects N Y Y Y
Patient characteristics‡ N N Y Y
Practice characteristics§ N N N Y
*P<0.05; **P<0.01; ***P<0.001.
†ORs based on SD units.
‡Patient-adjusted deprivation, proportion of patients aged 0–4 years, proportion of patients aged ≥75 years, proportion patients black or
Asian ethnicity, proportion of nursing home residents
§Region, minimum distance to hospital, contract type, dispensing status, training practice status, singlehanded.
CQC, Care Quality Commission.
with higher quality ratings, representing primary care variables which would otherwise tend to underestimate
structures which support higher quality of care. However, the positive association of funding with the quality rating.
some factors related to the registered practice popu-
lation, such as urban location, social deprivation and Strengths and weaknesses of the study
larger proportions of ethnic minority patients, were nega- This is the first study to explore the relationship between
tively associated with the practice quality of care rating. practice-level capitation funding and practice quality as
Many of these factors are already known to be negatively measured by CQC ratings. The findings are based on a
associated with reported patient satisfaction26 and QOF near-complete sample of general practices across England.
achievement.27 Including them in the model led to a Using data linkages from a wide range of sources and
stronger association of practice capitation funding with multilevel statistical models, this study has been able to
practice quality rating. The likely reason for this is that demonstrate the independent effects of practice funding
practice capitation funding is positively correlated with and practice characteristics on quality ratings, which
patient characteristics, which have negative effects on the might otherwise be confounded in single-level analyses.
quality rating. Thus, including these patient characteris- A variety of sensitivity analyses have confirmed the robust-
tics in the model removes a source of bias from omitted ness of the ordered logistic regression modelling.
Figure 3 Estimated probability of overall practice rating of ‘outstanding’ at various levels of capitation funding per registered
patient. Average predicted probability at each funding level (accounting for year effects, patient & practice characteristics) Mean
values displayed with 95% confidence intervals.
Figure 4 Estimated probability of overall practice rating of ‘inadequate’ at various levels of capitation funding per registered
patient. Average predicted probability at each funding level (accounting for year effects, patient & practice characteristics) Mean
values displayed with 95% confidence intervals.
However, there are several limitations. Routinely In a country-level European analysis, it was found
collected data are subject to coding and recording errors. that systems relying on capitation funding were more
As with all observational studies, significant associations, responsive than those based on fee for service or mixed
even if large, may not be causal. Although a wide range payment.28 However, analysis of Scottish general practices
of potential confounders were included in the models, suggests that capitation funding may contribute to the
confounding by omitted variables cannot be excluded. persistence of the inverse care law with deprived areas
Comparison with existing literature experiencing lower quality of care, as defined by inspec-
These findings complement those of a previous study tion ratings.29 Consistent with our study, others have
which found that increased general practice capitation found that GP practice funding is negatively correlated
funding was associated with reduced emergency hospital with healthcare need predictors such as deprivation
admissions and Accident and Emergency attendances.12 and non- white ethnicity.30 Previous studies have also
Figure 5 Estimated probability of ‘outstanding’ overall practice CQC rating: training versus non-training practices. Adjusted for
year effects, patient characteristics, practice characteristics & funding Mean values displayed with 95% confidence intervals in
shaded areas. CQC, Care Quality Commission.
Figure 6 Estimated probability of ‘outstanding’ overall practice CQC rating: single-handed versus group practices. Adjusted
for year effects, patient characteristics, patient characteristics & funding Mean values displayed with 95% confidence intervals
in shaded areas areas. CQC, Care Quality Commission.
demonstrated that greater GP workload may be associated for overall ‘profit’ per practice are expected to become
with higher levels of social deprivation and with a higher available in due course. Other studies have confirmed
proportion of Asian patients.31 Similarly, practices with a that incentives based on personal income may influence
greater proportion of ethnically diverse patients reported both quality achievement and productivity.34
worse patient experience.32 Our work is also consistent
with a recent study which demonstrated that GPs colo- Implications for policy and practice
cated with other GPs and professionals had improved This work provides further evidence of the association
outcomes compared with single- handed GP practices between general practice capitation funding and the
such as broader provision of technical procedures, wider quality of primary care. A causal association is plausible
coordination with secondary care and increased collabo- and supports the argument that increased quality and
ration among different providers.33 safety of patient care may be achieved through additional
Our study was based on funding data for general prac- investment. The recently published NHS Long Term
tices but was unable to study the relationship between Plan35 outlines proposals to offer increases in capitation
quality ratings and individual GP income. However, values payment together with an emphasis on inter- practice
Figure 7 Estimated probability of ‘outstanding’ overall practice CQC rating: rural and urban practices. Adjusted for year
effects, patient characteristics, practice characteristics & funding Shaded areas demonstrate 95%CI. CQC, Care Quality
Commission.
English quality and outcomes framework and the National GP patient analysis of national patient survey data and prescribing data. Br J
survey. Ann Fam Med 2013;11:467–72. Gen Pract 2016;66:e40–6.
5 Gillam S, Siriwardena AN. Regulation in primary care. Qual Prim Care 21 Office for National Statistics. 2011 census: office for national
2014;22:57–61. statistics, 2011. Available: https://www.ons.gov.uk/census/
6 Emmert M, Eijkenaar F, Kemter H, et al. Economic evaluation of 2011census [Accessed 21 Jan 2016].
pay-for-performance in health care: a systematic review. Eur J Health 22 NHS Digital. Nhs payments to general practice, England, 2013-14:
Econ 2012;13:755–67. experimental statistics – report, 2015. Available: http://www.hscic.
7 Kontopantelis E, Springate DA, Ashworth M, et al. Investigating the gov.uk/catalogue/PUB16847/nhspaymentsgp-13-14-rep.docx
relationship between quality of primary care and premature mortality [Accessed 5th Sept 2015].
in England: a spatial whole-population study. BMJ 2015;350:h904. 23 Ashworth M, Schofield P, Durbaba S, et al. Patient experience and the
8 Campbell SM, Reeves D, Kontopantelis E, et al. Effects of pay for role of postgraduate GP training: a cross-sectional analysis of national
performance on the quality of primary care in England. N Engl J Med patient survey data in England. Br J Gen Pract 2014;64:e168–77.
2009;361:368–78. 24 Lall R, Campbell MJ, Walters SJ, et al. A review of ordinal regression
9 NHS Digital. NHs payments to general practice, England, 2016/17 models applied on health-related quality of life assessments. Stat
2017. Methods Med Res 2002;11:49–67.
10 British Medical Association. Focus on the global sum 25 Brant R. Assessing proportionality in the proportional odds model for
allocationformula (Carr-Hill formula), 2015. Available: https:// ordinal logistic regression. Biometrics 1990;46:1171–8.
www.google.co.uk/url?sa=t&rct=j&q=&esrc=s&source=web&cd= 26 Roberts MJ, Campbell JL, Abel GA, et al. Understanding high and
1&cad=rja&uact=8&ved=0ahUKEwi_sLK47f_OAhWLAMAKHZ2 low patient experience scores in primary care: analysis of patients'
bC8oQFggcMAA&url=https%3A%2F%2Fwww.bma.org.uk% survey data for general practices and individual doctors. BMJ
2F-%2Fmedia%2Ffiles%2Fpdfs%2Fpractical%2520advice% 2014;349:g6034.
2520at%2520work%2Fcontracts%2Fgps%2Ffocus%2520on% 27 Ashworth M, Armstrong D. The relationship between general practice
2520global%2520sum%2520allocation%2520formula.pdf%3Fla% characteristics and quality of care: a national survey of quality
3Den&usg=AFQjCNEfyEp15cpvpLtT77bIobeHfSmGMQ&sig2=ojUD- indicators used in the UK quality and outcomes framework, 2004-5.
oKqav9bcLejniNdtw&bvm=bv.131783435,d.d2s [Accessed 12 Apr BMC Fam Pract 2006;7:68.
2016]. 28 Murante AM, Seghieri C, Vainieri M, et al. Patient-perceived
11 NHS Digital. General and personal medical services, England: 2005- responsiveness of primary care systems across Europe and the
2015, as at 30 September, provisional experimental statistics. NHS relationship with the health expenditure and remuneration systems of
Digital, 2016. primary care doctors. Soc Sci Med 2017;186:139–47.
12 L'Esperance V, Sutton M, Schofield P, et al. Impact of primary care 29 McLean G, Guthrie B, Mercer SW, et al. General practice funding
funding on secondary care utilisation and patient outcomes: a underpins the persistence of the inverse care law: cross-sectional
retrospective cross-sectional study of English general practice. Br J study in Scotland. Br J Gen Pract 2015;65:e799–805.
Gen Pract 2017;67:e792–9. 30 Levene LS, Baker R, Wilson A, et al. Population health needs as
13 Dunham AH, Dunbar JA, Johnson JK, et al. What attributions do predictors of variations in NHS practice payments: a cross-sectional
Australian high-performing general practices make for their success? study of English general practices in 2013-2014 and 2014-2015. Br J
applying the clinical microsystems framework: a qualitative study. Gen Pract 2017;67:e10–19.
BMJ Open 2018;8:e020552. 31 Mukhtar TK, Bankhead C, Stevens S, et al. Factors associated with
14 Care Quality Commission. The state of care in general practice 2014 consultation rates in general practice in England, 2013-2014: a
to 2017. London: Care Quality Commission, 2017. cross-sectional study. Br J Gen Pract 2018;68:e370–7.
15 Care Quality Commission. How CQC monitors, inspects and 32 Lyratzopoulos G, Elliott M, Barbiere JM, et al. Understanding ethnic
regulates NHS GP practices. London: Care Quality Commission, and other socio-demographic differences in patient experience of
2018. primary care: evidence from the English general practice patient
16 NHS Digital. NHS payments to general practice, England: NHS survey. BMJ Qual Saf 2012;21:21–9.
digital, 2015. Available: http://digital.nhs.uk/catalogue/PUB18468/ 33 Bonciani M, Schäfer W, Barsanti S, et al. The benefits of co-location
nhspaymentsgp-14-15-ann1.xlsx in primary care practices: the perspectives of general practitioners
17 NHS Digital. General and personal medical services: NHS digital, and patients in 34 countries. BMC Health Serv Res 2018;18:132.
2016. Available: http://digital.nhs.uk/catalogue/PUB20503 34 Gosden T, Forland F, Kristiansen IS, et al. Impact of payment method
18 Department for Communities and Local Government. English on behaviour of primary care physicians: a systematic review. J
indices of deprivation 2015. Department for Communities and Local Health Serv Res Policy 2001;6:44–55.
Government, 2015. 35 NHS England. The NHS long term plan. NHS England, 2019.
19. NHS England. GP Patient Survey, 2013. Available: https://www. 36 Care Quality Commission. The five key questions we ask: care
england.nhs.uk/statistics/statistical-work-areas/gp-patient-survey/ quality Commission, 2016. Available: http://www.cqc.org.uk/content/
[Accessed 15th July 2017]. five-key-questions-we-ask [Accessed 12 Apr 2017].
20 Ashworth M, White P, Jongsma H, et al. Antibiotic prescribing and
patient satisfaction in primary care in England: cross-sectional