NHS Digital Data Release Register - reformatted

Competition & Markets Authority projects

18 data files in total were disseminated unsafely (information about files used safely is missing for TRE/"system access" projects).


Project 1 — DARS-NIC-32833-M3M9V

Type of data: information not disclosed for TRE projects

Opt outs honoured: No - data flow is not identifiable

Legal basis: Health and Social Care Act 2012, Health and Social Care Act 2012 – s261(1) and s261(2)(b)(ii)

Purposes: ()

Sensitive: Non Sensitive

When:2018.03 — 2017.02.

Access method: Ongoing, One-Off

Data-controller type:

Sublicensing allowed:

Datasets:

  1. Hospital Episode Statistics Accident and Emergency
  2. Hospital Episode Statistics Admitted Patient Care
  3. Hospital Episode Statistics Outpatients

Objectives:

The Competition and Markets Authority (CMA) is required, under the Enterprise Act 2002 (EA02), to monitor, investigate and decide whether mergers may be expected to give rise to a substantial lessening of competition within any markets(s) in the UK for goods or services. If the CMA concludes that there is an anti-competitive outcome, it must decide whether action should be taken to remedy, mitigate, or prevent the identified substantial lessening of competition and/or any adverse effect which may be expected to result from the substantial lessening of competition. The Health and Social Care Act 2012 (HSCA12) confirms this duty with respect to the merger of NHS Foundation Trusts.

The CMA will be conducting reviews of several NHS Trust and Foundation Trust mergers across England. The reviews seek to establish whether the mergers may result in adverse effects for patients by reducing choice and competition. If borne out, such a reduction in competition could have the consequence of removing or dampening the incentives that currently exist for the Trusts to maintain and improve quality for patients, or to retain patients, given that the hospitals currently may seek to attract patient referrals from each other.

The CMA's objective for processing data is to analyse the closeness of competition between the specific merging Trusts at trust and site level to specifically understand:
- the geographic distribution of patients attending Trusts in the merging areas
- the overlap in services provided by the merging Trusts and other trusts in the surrounding area, to the extent that these Trusts provide similar or identical treatments at different levels of data aggregation to understand patients' options in the area, and
- the substitutability of the Trusts, to the extent that the merging Trusts and other trusts in the surrounding area are alternative providers to patients at GP and GP Practice level.

Yielded Benefits:

The CMA has conducted extensive analysis using this data to fulfil their statutory function of 1) deciding whether mergers involving NHS Foundation Trusts would result in a substantial lessening of competition and, if so, 2) determining the scale and scope of the adverse effect from lessening of competition on patients and commissioners that must be outweighed by relevant patient benefits in order for the merger to be allowed. This benefits healthcare by helping to ensure that only mergers that are, on balance, beneficial to patients are allowed to occur. The NHS Digital data underpinned most of the analysis on the impact on patient choice of provider of two high-profile mergers: 1) the merger of Central Manchester University Hospitals NHS FT and University Hospital of South Manchester NHS FT, and 2) the merger of University Hospital Birmingham NHS FT and Heart of England NHS FT. The CMA used the data to analyse the merging Trusts and other local providers’ activity and revenue, and the pattern of referrals from GPs and other referrers. The CMA publishes the full reasoning and analysis behind their decisions on their case pages (https://www.gov.uk/cma-cases). For examples, see this link for CMFT/UHSM: https://www.gov.uk/cma-cases/central-manchester-university-hospitals-university-hospital-of-south-manchester-merger-inquiry and this link for UHB/HEFT: https://www.gov.uk/cma-cases/university-hospitals-birmingham-heart-of-england-merger-inquiry

Expected Benefits:

Economic literature suggests competition between healthcare providers strengthens incentives for hospitals to innovate, therefore maintaining and improving standards of quality. Following a merger between NHS (Foundation) Trusts, if there is a reduction of choice and competition, then this could have the consequence of reducing quality for patients by removing the incentives that currently exist for the Trusts to attract patient referrals from each other.

The CMA must therefore finally decide whether the merger would result in a substantial lessening of competition. If a substantial lessening of competition is found, the CMA needs to consider whether it should take action for the purpose of remedying, mitigating or preventing the substantial lessening of competition concerned. For these purposes the CMA will also have regard to any relevant patient (or consumer) benefits. The HES data requested may also be of material assistance in CMA's weighing up of the potential substantial lessening of competition against any relevant patient benefits put to us by the merging Trusts. Given that many tens of thousands of patients may be affected, the CMA’s decisions as to whether, or on what grounds to allow the Trusts to merge, must be well informed.

Outputs:

The CMA intends to:
- compile a ranking of the relevant Trusts as service providers according to the GP referral analysis
- compare the results to simple choice modelling
- produce maps showing the location of GP practices, where the proposed merger may result in the largest restriction in choice, and
- compile tables showing the extent to which hospitals in the area provide the same or similar treatments.
- produce statistics to understand the distance over which patients in the area travel for care at specialty and provider level

These will be used in CMA's internal decision making process and will inform CMA's internal working papers and reports for eventual publication. Raw data will not be published but instead aggregated to a level where patients are not identifiable, and all small numbers will be removed in line with the HES Analysis Guide.

The CMA publishes the full reasoning and analysis behind their decisions on their case pages (https://www.gov.uk/cma-cases).

For examples, see this link for CMFT/UHSM:
https://www.gov.uk/cma-cases/central-manchester-university-hospitals-university-hospital-of-south-manchester-merger-inquiry
and this link for UHB/HEFT:
https://www.gov.uk/cma-cases/university-hospitals-birmingham-heart-of-england-merger-inquiry

A disclosure room may be required if there is a finding of an substantial lessening of competition which is contested.

Processing:

The CMA considers the following data as necessary in performance of CMA's objectives:
- patient level data with no small number suppression (to use procedure and diagnosis information to cross-check specialty coding in order to know which services both merging Trusts offer, as the merging Trusts have informed us that specialty coding practices can be inconsistent across Trusts)
- pseudonymised data (in order to identify cases where patients have multiple episodes of treatment as part of their care pathway, which may lead to duplicates in the analysis)
- data for only the CCGs proximate to the trusts, or with whom the merging Trusts are contracted to provide services (in order to undertake this analysis for the area over which the patients travel to the merging Trusts)
- data covering several years (in order to increase the sample size in low-volume specialties and thus smooth out unrepresentative figures).

The CMA will then process the data by means of a ‘GP Referral Analysis’ methodology and specifically:
- produce a ranking of Trusts by their credibility as alternative providers of services
- consider which alternative providers are credible for a patient attending a given hospital through examining other referrals at their GP practice
- produce rankings at a GP practice level, which CMA will then aggregate to develop a picture for the whole Trust
- calculate and map referral ratios to providers from individual GP practices and
- tabulate numbers of patients treated by the merging trusts at specialty level.
- compare the results to simple choice modelling.

The CMA will combine the HES data with other publically available information in order to test the impact of specific factors and to adequately control for factors which might otherwise bias CMA's results. CMA plan to match the following publically available data with the postcodes in HES data
(a) ODS characteristics for GP practices, Branch Surgeries, NHS Trusts and Trust Sites.
(b) ONS postcode lookup information including rural/urban indicators and coordinates (for GP practices and care providers), in order to calculate approximations of travel distances, and to control for the impact of rurality on these
(c) ONS population density measures by GP practice, provider postcode and output areas to control for the impact of rurality on willingness to travel
(d) Hospital Estates and Facilities Statistics data by hospital to control for characteristics of each hospital
(e) Hospital performance data from the CQC, NHS choices, NHS Digital, NHS England, and My NHS BETA to test the relationship between quality of hospitals and key variables of interest
(f) Data from NHS Digital, NHS Improvement and the Department of Health in relation to provider tariffs.

If the CMA’s proposed decision is adverse to the Trusts’ anticipated merger, the CMA is likely to use a disclosure room as a way of enabling parties’ external advisers to review the data so as to verify the analyses that the CMA has conducted, thus ensuring that the CMA acts in a procedurally fair way without compromising the security of very sensitive data. The parties’ external advisers who are admitted to the disclosure room will be required to give strict, written and legally enforceable undertakings not to disclose any of the data more widely (including to their client(s)). Under the disclosure room rules, the data will only be accessible to external advisers for a limited period of time via computer terminals on the CMA’s premises, and the parties’ advisers will not be able to transmit or otherwise remove data from the data room other than in aggregated form with small numbers suppressed in line with the HES Analysis Guide. The CMA has operated such disclosure rooms in other cases and has found them to be an effective way of accommodating both procedural fairness and data security. Whilst there is a need to consult on the exact nature of the undertakings related to a disclosure room in this context, and the CMA cannot therefore upload the exact wording, the CMA have uploaded the undertakings from a recent market investigation which used similar pseudonymised data as an example.

With the exception of the disclosure room scenario, access to data will be restricted to authorised substantive CMA employees (with system permissions granted by the Data Access Team within the Corporate Services Department) and will only be used for the fulfilment of the project’s objectives as described above. The Data Access Team within CMA grant permissions to CMA employees only when access is required and approved by the Deputy Chief Economic Adviser at CMA. The CMA Data Access Team are only involved in facilitating access and not in analysing the data. The CMA Deputy Chief Economic Adviser (the data recipient) must authorise any person using the data, and the Data Access Team in IT (also authorised by Deputy Chief Economic Advisor) facilitate giving that person access. If the CMA uses a disclosure room, access – subject to specific controls - may be given to external parties who have signed written undertakings, within the secure environment of the disclosure room.


Project 2 — DARS-NIC-38368-V3S5C

Type of data: information not disclosed for TRE projects

Opt outs honoured: N

Legal basis: Health and Social Care Act 2012

Purposes: ()

Sensitive: Non Sensitive

When:2016.09 — 2016.11.

Access method: One-Off

Data-controller type:

Sublicensing allowed:

Datasets:

  1. Hospital Episode Statistics Admitted Patient Care
  2. Hospital Episode Statistics Outpatients

Objectives:

The Competition and Markets Authority (CMA) is required, under the Enterprise Act (EA02), to monitor, investigate and decide whether mergers may be expected to give rise to a substantial lessening of competition within any markets(s) in the UK for goods or services. If the CMA concludes that there is an anti-competitive outcome, it must decide whether action should be taken to remedy, mitigate, or prevent the identified substantial lessening of competition and/or any adverse effect which may be expected to result from the substantial lessening of competition. The Health and Social Care Act 2012 (HSCA12) confirms this duty with respect to the merger of NHS Foundation Trusts.

In order to gain a deeper understanding about patient choice, which will allow the CMA to assess the effects of mergers involving NHS Foundation Trusts with greater sophistication, the CMA is undertaking a research programme with the following broad themes:
(a) Theme A: Patient responses to changes in the structure of local health economies, significant local health events, and / or provider failings.
(b) Theme B: The extent to which choice is exercised at various nodes along the patient pathway and the implications of this for merger assessment.

The following information explains CMA’s research plans under each of these themes.

Theme A:

Understanding how patients respond to events that may change the quality of a provider is central to CMA’s work to make the healthcare market work for consumers.

CMA intends to conduct an analysis to test how patients respond to events impacting upon the quality of healthcare providers. This analysis will allow CMA to assess the performance of CMA’s existing simple choice models (which are generally used to predict future patient flows) against the choices of patients actually observed in cases where an event has incentivised or forced patients to go elsewhere. This will mean that CMA can make more precise decisions in CMA’s legally mandated reviews in the healthcare sector.

Conducting an event analysis will also help CMA to predict better which patients are more likely to choose a different hospital if there is a decrease in quality at their previously chosen. This would be informative of:
(i) The extent to which patient choice is responsive to changes in quality in any given area, and so provide an indicator as to the extent of the scope of current competition in a given local area;
(ii) The specialties which would be most affected by any given merger.

Theme B:

Whilst patient choice is only formally exercised at the outpatient level, some GPs/patients who anticipate that they will need to be admitted as part of their care may also consider the quality of inpatient and day-case services as part of their initial choice of provider.

CMA seeks to understand more fully how patient choice with respect to outpatient and admitted patient factors works in practice. Specifically, CMA plans to model patient choice factors specific to each treatment setting and factors common across all treatment settings using regression analysis. This will allow CMA to understand the extent to which competition for admitted patients can drive up quality of services, quantify this extent and understand the circumstances in which this effect is largest. This will inform CMA’s judgement in individual merger assessments whether a particular merger may harm patient welfare, or whether the benefits outweigh the costs.

CMA further plans to test its existing models of patient choice for admitted patients against the results of such analysis. Using data with a wide scope will allow CMA to develop a robust understanding of its reliability across the range of specialties and provider types which are relevant to CMA’s statutory duties. It will also help CMA tailor the model to develop more accurate results, if CMA finds that there is scope for improvement, or suggest areas where future research is needed.

Expected Benefits:

Research literature suggests competition between healthcare providers strengthens incentives for hospitals to innovate, therefore maintaining and improving standards of quality. CMA hopes to build on this body of literature. Through developing and refining CMA’s methodology, and gaining a deeper understanding about patient choice, this will allow the CMA to assess the effects of mergers involving NHS Foundation Trusts with greater sophistication. This will help to maintain quality standards in the NHS, ultimately increasing patient welfare and reducing mortality rates.

Outputs:

CMA aims to complete the research programme within a year of initiation. In the event that this does not prove possible, CMA would apply to extend its access to the data.

The details of CMA’s research (including statistical result tables, graphs and maps) will take the form of working papers to be shared inside CMA’s organisation, aggregating small numbers in line with HSCIC policy where these are shared outside of the case team.

CMA’s overall findings will be compiled into academic articles for journal and conference submission, and in research papers which CMA will disseminate within and outside the organisation. The reason for publishing CMA’s findings is to ensure that CMA’s findings are open to critical scrutiny and to preserve transparency of CMA’s work. In all cases of publication, CMA will aggregate the results and suppress small numbers in line with HES Analysis Guide to ensure that patient anonymity is not compromised.

Processing:

The CMA considers the following data as necessary in performance of the CMA research programme objectives:

o Patient episode level data with no small number suppression, to use procedure and diagnosis information to cross-check specialty coding, and to reassign episode information where CMA finds coding inconsistencies. This is necessary for the research to analyse robustly common treatments across providers (in essence, to ensure the CMA are comparing like with like). CMA also requires episode level data to re-compute HRG codes and provider revenues based on such reallocation.

o Pseudonymised data in order to identify cases where patients have multiple episodes of treatment as part of their care pathway, which if unidentified could lead to duplicates in the analysis.

o Using national data (as opposed to local/regional data) would enable CMA to:
(a) In relation to Theme A, have access to a sufficiently large cross section of events within different subcategories, for example: mergers, quality decreases, temporary closures, refurbishments etc). This will allow CMA to conduct robust analysis which is not constrained by sample size and which does not mix together different effects. It will also allow CMA to test whether patient responses vary across different categories of events. CMA will be able to draw out factors which are predictive of larger or smaller responses, and apply these to future local assessments.
(b) In relation to Theme B, capture a sufficient number of data points to ascertain which specialties generally contain patients which are more likely to respond to quality changes. Given that some specialties contain very low numbers of patients, having a large sample size is important. This will help CMA inform the appropriate starting temperature on each specialty in any given local assessment

o CMA will need data covering 3 years (2012/13 – 2014/15) in order to test the stability of patient/GP choices following events in the health economy using the time dimension, and to increase the sample size in low-volume specialties and thus smooth out unrepresentative figures.

CMA will combine the HES data with other publically available information in order to test the impact of specific factors and to adequately control for factors which might otherwise bias CMA’s results. CMA plans to match the following publically available data with the postcodes in HES data:
(a) ODS characteristics for GP practices, Branch Surgeries, NHS Trusts and Trust Sites.
(b) ONS postcode lookup information including rural/urban indicators and coordinates (for GP practices and care providers), in order to calculate approximations of travel distances, and to control for the impact of rurality on these
(c) ONS population density measures by GP practice and provider postcode to control for the impact of rurality on willingness to travel
(d) ONS Health deprivation and income distribution measures by GP practice and provider postcode to control for unequal distributions of health and wealth
(e) Hospital Estates and Facilities Statistics data by hospital to control for characteristics of each hospital
(f) Hospital performance data from the CQC, NHS choices and My NHS BETA to test the relationship between quality of hospitals and key variables of interest
(g) Data from the GP patient survey, NHS choices and My NHS BETA by GP practice to control for characteristics and the quality of each GP practice
(h) Data from the HSCIC, Monitor, NHS Improvement and the Department of Health in relation to provider tariffs.

Under both of the aforementioned themes, CMA will use regression analysis to produce statistical result tables. CMA will also produce graphs, maps and tables to summarise key features of the data and to make CMA’s research more accessible. Outputs will only contain data that is aggregated with small numbers suppressed in line with the HES Analysis Guide.

Access to data will be restricted to authorised substantive CMA employees from the Data Access Team within the Corporate Services Department and will only be used for the fulfilment of the project’s objectives as described above.