Summary
Background
The effects of outsourcing health services to for-profit providers are contested, with some arguing that introducing such providers will improve performance through additional competition while others worry that this will lead to cost cutting and poorer outcomes for patients. We aimed to examine this debate by empirically evaluating the impact of outsourced spending to private providers, following the 2012 Health and Social Care Act, on treatable mortality rates and the quality of health-care services in England.
Methods
For this observational study, we used a novel database composed of parsable procurement contracts between April 1, 2013, and Feb 29, 2020 (n=645 674, value >£25 000, total value £204·1 billion), across 173 clinical commissioning groups (CCGs; regional health boards) in England. Data were compiled from 12 709 heterogenous expenditure files primarily scraped from commissioner websites with supplier names matched to registers identifying them as National Health Service (NHS) organisations, for-profit companies, or charities. We supplemented these data with rates of local mortality from causes that should be treatable by medical intervention, indicating the quality of health-care services. We used multivariate longitudinal regression models with fixed effects at the CCG level to analyse the association of for-profit outsourcing on treatable mortality rates in the following year. We used the average marginal effects to estimate total additional deaths attributable to changes in for-profit outsourcing. We provided alternative model specifications to test the robustness of our findings, match on background characteristics, examine the potential impact of measurement error, and adjust for possible confounding factors such as population demographics, total CCG expenditure, and local authority expenditure.
Findings
We found that an annual increase of one percentage point of outsourcing to the private for-profit sector corresponded with an annual increase in treatable mortality of 0·38% (95% CI 0·22–0·55; p=0·0016) or 0·29 (95% CI 0·09–0·49; p=0·0041) deaths per 100 000 population in the following year. This finding was robust to matching on background characteristics, adjusting for possible confounding factors, and measurement error in our dataset. Changes to for-profit outsourcing since 2014 were associated with an additional 557 (95% CI 153–961) treatable deaths across the 173 CCGs.
Interpretation
The privatisation of the NHS in England, through the outsourcing of services to for-profit companies, consistently increased in 2013–20. Private sector outsourcing corresponded with significantly increased rates of treatable mortality, potentially as a result of a decline in the quality of health-care services.
Funding
Wellcome Trust.
Introduction
,
- Elkomy S
- Cookson G
- Jones S
Meanwhile, in the mid-2000s, reforms centred patient choice by introducing a consumer market that increased the use of private finance and independent sector treatment centres.
,
- Sanderson M
- Allen P
- Osipovic D
this policy made it almost compulsory to outsource certain NHS services, or at least impossible to ensure contracts remained in the NHS.
- Pollock AM
- Price D
- Roderick P
- et al.
,
Mortality rates.
The specifics of these reforms are outlined in secondary legislation, the “Procurement, Patient Choice and Competition Regulations No. 2 (2013)”,
The National Health Service. Procurement, Patient Choice and Competition. No. 2. Regulations 2013 (SI 2004/3554).
which directly ruled against any commissioning priority based on ownership status, meaning NHS providers could not be preferred over for-profit organisations by legal right.
Evidence before this study
We searched Google Scholar, PubMed, and ProQuest for studies published in English from database inception up to Dec 1, 2021, assessing the relationship between health-care privatisation and health outcomes using the terms “healthcare privatisation”, “outsourcing”, “contracting”, “out”, “for-profit”, “healthcare quality”, “mortality”, and “NHS” applied to keywords, abstracts, and titles. We found various studies that largely compared outcomes between different health-care providers on the basis of ownership status. These studies had mixed findings regarding health-care quality and often concluded that different case-mixes of private and public hospitals make firm conclusions difficult. In the UK, studies commonly evaluate the aggregate health outcomes from differing levels of competition between providers, but have not accounted for levels of for-profit outsourcing. Evidence suggests mortality rates rose in Italy following a period of privatisation but whether the National Health Service (NHS) in England has even had substantial levels of privatisation is severely contested. To the best of our knowledge, no studies have been done to investigate the association of for-profit outsourcing at the NHS commissioner level with health outcomes in England.
Added value of this study
This observational study used a novel database with £204·1 billion of expenditure comprising 645 674 individual payments made by 173 NHS clinical commissioning groups in England between 2013 and 2020. We used these data to assess whether changes in the proportion of the expenditure being spent on for-profit companies are associated with changes in treatable mortality rates and, therefore, with the quality of health care. These data allowed us to conduct, to the best of our knowledge, the first empirical evaluation of a controversial health-care reform in England’s recent history. Our findings help advance the debate about health-care privatisation considering that the extent of NHS privatisation in England was previously contested.
Implications of all the available evidence
Our study suggests that increased for-profit outsourcing from clinical commissioning groups in England might have adversely affected the quality of care delivered to patients and resulted in increased mortality rates. Our results provide the first assessment of creeping privatisation in England since controversial reforms were introduced in 2012 to encourage outsourcing of services, and our findings are corroborated in other contexts of health-care privatisation. Our findings suggest that further privatisation of the NHS might lead to worse population health outcomes.
- Honeyford K
- Greaves F
- Aylin P
- Bottle A
Similarly, increases in waiting times and decreased patient satisfaction suggest the NHS is failing to maintain standards of care.
- Green MA
- Dorling D
- Minton J
- Pickett KE
Although austerity measures during this period have almost certainly played a role,
- Watkins J
- Wulaningsih W
- Da Zhou C
- et al.
,
- Alexiou A
- Fahy K
- Mason K
- et al.
in this study we examined whether outsourcing to for-profit companies has contributed to this increase in treatable mortality.
This has occurred in other countries, such as when mortality rates rose in Italy following a period of privatisation, and in other parts of the NHS in England, such as when cleaning services were outsourced.
- Quercioli C
- Messina G
- Basu S
- McKee M
- Nante N
- Stuckler D
,
- Toffolutti V
- Reeves A
- McKee M
- Stuckler D
- Kunz J
- Propper C
- Staub K
- Winkelmann R
Another key dynamic is the different case-mixes often observed between for-profit and public providers—a result of so-called cream-skimming and concentrating the most complicated cases with public providers, which have no extra staff or funding to compensate.
- Chard J
- Kuczawski M
- Black N
- van der Meulen J
,
- Pérotin V
- Zamora B
- Reeves R
- Bartlett W
- Allen P
Moreover, comparisons between for-profit and public providers are often inappropriate because the case-mixes of private and public services are considerably different.
The 2012 reorganisation created new bodies for NHS health procurement, termed clinical commissioning groups (CCGs), which replaced the old primary care trusts, and responsibility for public health services was transferred to local authorities. CCGs were also individually required to publish their expenditure data, which produced discrepancies in the location and availability of commissioning expenditure data; these discrepancies made evaluation of outsourcing previously unfeasible.
We aimed to examine the impact on treatable mortality of increased outsourcing to private for-profit providers from CCGs in England during the period immediately following the implementation of the 2012 Health and Social Care Act.
Methods
Overview
The biggest challenge preventing evaluation of outsourcing from the NHS in England until now has been the absence of a harmonised data resource suitable for analysis. For this observational study, we used a novel database compiling parsable procurement expenditures between April 1, 2013, and Feb 29, 2020 (n=645 674, value >£25 000, total value £204·1 billion). This resource allowed us to analyse the impact of for-profit outsourcing in unprecedented detail, by conducting, to the best of our knowledge, the first robust empirical assessment of the impact of for-profit outsourcing from the NHS following the 2012 Health and Social Care Act on health outcomes.
Data collection
The method builds on recent progress to scrape, parse, and merge disaggregated public payments datasets, making them into accessible data resources with many applications in research and policy.
,
Avoidable mortality in the UK: 2019.
Avoidable mortality in the UK: 2019.
This measure is an age-standardised rate of mortality per 100 000 population for specific causes of death—a full list of causes that are considered treatable is provided in the appendix (p 32). However, CCGs represent registered patients of general practitioners (GPs) through membership, rather than representing a geographical population. Consequently, our treatable mortality measure is an approximation of population outcomes in a given area rather than precise outcomes for patients using CCG services.
The explanatory variable of interest was a measure of outsourcing, defined as commissioning expenditure which is received by for-profit companies as a percentage of total expenditure. This definition excluded expenditure received by private non-profit organisations—all those registered to the central register of charities—as we specifically focused on the aggregate effects of outsourcing to providers that have profit-maximisation incentives.
Statistical analysis
These analyses were all reported with cluster-robust standard errors with small-N adjustments.
Finally, we did a multi-level random intercepts model, clustering mortality rates for local authorities within their geographically overlapping CCGs, allowing the intercept to vary for each cluster to see whether CCG outsourcing explains mortality rates in their relative local authorities. The full models are summarised in the appendix (p 3).
Sensitivity analysis
Our analysis was run on novel data that were produced by web scraping and algorithmic matching of contracts published in non-uniform formats. Despite multiple manual data verification checks, it is probable that a small amount of error existed in our outsourcing observations. To check whether potential error in the contract data influenced our inferences, we synthetically replicated the effect of error on our findings. We ran the linear fixed-effects model 10 000 times but multiplied each observation for outsourcing by a random number with a specified minimum and maximum limit. We then repeated this analysis five times with different maximum error sizes, the largest of which was 50% (replicated by multiplying each value by a random number between 0·5 and 1·5), far larger than we would expect to exist in the data. We then plotted the density of the resulting coefficients for outsourcing in each regression, simulating how random error could have affected the findings.
Role of the funding source
The funder of this study had no influence on data collection, data analysis, data interpretation, writing of the manuscript, or the decision to submit for publication.
Results
UK standard industrial classification (SIC) hierarchy.
Figure 1Levels of CCG outsourcing to for-profit organisations from 2013 to 2020
(A) Rolling percentage of total spending on health care and other for-profits. (B) Percentage change in total spending received by companies in different industrial sectors, based at zero for each sector’s 2013–14 levels. (C) Total for-profit outsourcing over the entire time series for each CCG in England. (D) Levels of for-profit outsourcing across all CCGs each month. CCG=clinical commissioning group. NA=not available.
The table shows the main results from our statistical analysis assessing the relationship between outsourcing and mortality rates.
Figure 2Treatable deaths from 2006 to 2018
Figure 3Synthetic random error
The plot shows the density of the coefficient for outsourcing treatable mortality when running the regression 10 000 times with five different levels of random error. The horizontal dashed line represents the regression coefficient of for-profit outsourcing with observed values.
TableOutsourcing and treatable mortality from multivariate longitudinal regression models
The table shows the estimated annual increase in log-transformed treatable mortality rate against the same dependent variables using five different model specifications. Clinical commissioning group (CCG) fixed effects, time fixed effects, clustered standard errors, and demographic control variables were integrated into all models, with the exception of the multi-level model, which did not include CCG fixed effects. Numbers observed represent the number of observations used in each analysis. Numbers differ for each analysis due to missing data and model specifications. R2 and conditional R2 report the amount of variation in treatable mortality rates accounted for by the explanatory variables. The Akaike information criterion, Bayesian information criterion, and log likelihood function assess the goodness of fit of the models. The intraclass correlation coefficient shows how much of the variation in treatable mortality at the local authority level is explained by their CCG clusters.
* For-profit outsourcing, local authority spending, and CCG spending have a 1-year lag. Treatable mortality, population, and income are log transformed. Full model expressions are available in the appendix (p 4). Robust standard errors are clustered at the CCG level and use a bias-reduced linearisation estimator. Satterthwaite degrees of freedom are used to calculate estimates in the multi-level model. Demographic control variables included degree of education (%), managerial or professional occupation (%), ethnic minority (%), unemployment rate (%), and claimant rate (%).
Discussion
Is the NHS being privatised?.
Using a novel dataset based on procurement contracts for 173 CCGs, we found that increased outsourcing from CCGs in England was associated with an increase in mortality from treatable causes, potentially caused by worsening in the quality of health-care services.
Many have attributed these outcomes to austerity policies, leaving public services underfunded and having direct consequences on the social determinants of health through welfare cuts.
- Watkins J
- Wulaningsih W
- Da Zhou C
- et al.
,
- Alexiou A
- Fahy K
- Mason K
- et al.
We suggest that outsourcing to for-profit companies is another way that the reforms of the post-financial crisis era have affected NHS service quality and mortality rates. However, with outsourcing being used as a mechanism for further austerity in some policy contexts, its relationship to health care deserves further attention.
- Bach-Mortensen A M
- Barlow J
The marketisation of health-care services is underpinned by the beliefs that openness, competition, and management autonomy can improve the efficiency and performance of state-funded services.
- Krachler N
- Greer I
- Umney C
For decades, these principles have dictated the organisation of the NHS in England.
However, our results suggest that these processes, manifesting in the outsourcing of health-care provision, are not associated with improvements in service provision, and instead have been associated with increased deaths among patients.
- Chard J
- Kuczawski M
- Black N
- van der Meulen J
However, recent evidence finds no substantial difference in the rate of deaths from surgeries in private and public hospitals in England, even if selection effects make this estimation difficult.
- Crothers H
- Liaqat A
- Reeves K
- et al.
NHS surgeries might be delivered under more stringent conditions by for-profit companies than by NHS providers; however, differences in health outcomes are yet to be observed for those treated by NHS providers versus those treated by for-profit providers.
Similarly, increased competition for contracts could result in health-care providers prioritising easily quantified outcomes such as waiting times at the expense of quality of care, resulting in higher patient mortality, as was identified in the NHS after the pro-market reforms during the 1990s.
- Propper C
- Burgess S
- Gossage D
The fact that we focused on a measure of health-care service performance and found no association between mortality and outsourcing when using a measure of mortality from causes that are treated by public health interventions suggests the overarching explanation for the increased mortality rates might be an aggregate decline in the quality of care. At the same time, more research is needed to unpack the precise mechanisms of worsening care in England since 2013, including an assessment of how private providers contribute to quality and safety data and systems of accountability. Another future avenue of research is the impact of outsourcing on health inequalities at the neighbourhood level, and the qualitative impact of access to health care.
These results have implications for the NHS privatisation debate, suggesting that for-profit provision of health-care services could be associated with worse population health outcomes. In the case of the NHS in England, our research raises doubts about whether the current extent of private sector use is optimal for the quality of care and suggests that further increases in for-profit provision would be a mistake. However, given the trends in the data, a change in direction and expansion of public sector provision seems unlikely without considerable political intervention.
The findings of this research are timely as new commissioning structures (integrated care boards) are about to replace CCGs entirely and, as in 2013, redraw the NHS market. This is a moment where once again the role of the private sector within the NHS in England must be scrutinised. The current analysis is also important given that, with only 42 integrated care boards replacing CCGs, such an analysis will not be possible in the future as local variation and accountability will be lost.
Limitations of this study include the length of time the data were available for, given the creation of CCGs in 2013, considerable mergers made at the beginning of 2020, and no legal requirement for their predecessors to publish expenditure data, which limited our ability to precisely measure outsourcing before 2013 or conduct before-and-after analyses. The associational nature of our findings cannot rule out the possibility of residual confounding, so our findings should not be interpreted as necessarily showing a causal relationship between outsourcing and mortality rates. Moreover, the expenditure data do not contain information about the specific services provided by the supplier; as such, further research is needed to establish whether some acute services are primarily responsible for the relationship between outsourcing to for-profit providers and increased mortality rates.
Since the passing of the 2012 Health and Social Care Act in England, for-profit companies are providing an increasing share of NHS services. Concerns about the quality of care provided by for-profit companies appear to be justified as our findings show that outsourcing is associated with higher rates of mortality from causes that could be treated by effective medical interventions.
BG and AR conceived the study idea and contributed to research design. BG and AR accessed and verified the data. BG cleaned and interpreted the data. BG did the literature review, analysed the data, and wrote the original draft of the manuscript. AR provided supervision, oversaw the analysis plan, and edited the manuscript.