The NHS deserves better use of hospital medicines dataBMJ 2020; 370 doi: https://doi.org/10.1136/bmj.m2607 (Published 17 July 2020) Cite this as: BMJ 2020;370:m2607
All rapid responses
Goldacre and McKenna “The NHS deserves better use of hospital medicines data” assert that NHS medicines data could be put to better use than is currently the case.
The authors’ thesis is built on analyses of primary care prescribing. Current restrictions on hospital data deprive them of the data they want. The principal result of changing the status quo would be to give them (and others) the access they suggest would benefit (deserve) the NHS. Circumstances frustrate their proving the specific benefits of data release.
Unintended consequences are not satisfactorily explored. For example, unpredicted intrusion into medical practice by ministers, interest from US organisations and an expansion of NHS “Head Office” head-count seem inevitable, and need to be weighed against the putative benefits the authors assert. The completed information exercise will be irreversible.
The accumulation of large datasets is an exciting prospect for researchers and computer scientists, but useful practical benefits for patients have only so far occurred in restricted clinical domains, in particular image analysis.
A valuable topic has been opened up. The ramifications, details and consequences need further critical attention so that benefits outweigh harms.
Competing interests: No competing interests
We thank Colin E Richman for his response.
We have been pursuing access to this dataset for over two years. We therefore welcome Rx-Info stating that hospital medicines data will soon be made openly available. It would be a great help if Rx-Info could share a date by which they currently anticipate this will happen?
We commend Rx-Info on their work to date on NHS medicines data. Decisions about procurement and access to data services are made by the NHS. Our paper is not intended to be critical of Rx-Info themselves, and we do not think it is. We simply think the underlying data should be made openly available to support innovation and quality improvement through data science, from a wide community of analysts who are here to help the NHS.
We hope this data will soon be made openly available.
Competing interests: No competing interests
Rx-info was founded in 2003 with the sole intention of providing the NHS with current and relevant information to enable medicines usage to be closely monitored for clinical quality, safety and cost.
The company, based in Exeter, employs a team of eight people and works exclusively for the NHS.
We were disappointed not to be asked to provide any information to assist in the preparation of the above article, and we would like to make the following observations:
NHSE has ensured that all NHS Trusts now have access to the Define data and all Trusts (100%) submit data to the system.
Define defaults (but is not limited) to benchmarking within similar NHS regions and similar hospital types as this is the most relevant to the users of the system.
Define allows for detailed, in-depth analysis across a complete benchmark group - not just the viewing Trust. Define reports on the standardised numbers and has the tools within to overlay the data with practical aggregations and also denominate with admissions and activity data.
This is user controlled.
Secondary care data does not come from a single, simple data set. There are multiple data sources, from different pharmacy stock control systems which have differing concepts in managing for example; stock issues, return of stock, homecare supply and other outsourced supply, manufacture of medicines and VAT. None of these are considerations in the Primary Care dataset.
On top of this are service level agreements which provide medicines and services to other NHS organisations. This can all, as argued by Ben Goldacre and Brian MacKenna, be modelled, though it has been a cumulative 15-year programme of research and development aligning with new structures and demands within the NHS.
NHSBSA receives central funding and its ePACT data is a fortuitous by-product of the reimbursement of GPs and Pharmacy contractors. Our system was developed from scratch supported by our colleagues in NHS hospital Trusts. Training, support and development are included in the licence fee.
We have close working relationships with Public Health England, who receive full, up to date, detailed level of data to support ESPAUR and other antimicrobial stewardship measures. The detail provided in the Carter review for the medicines section would not have been achievable without broad access to the data and tools we produce.
We have ongoing relationships with academia notably Professor Ray Fitzpatrick, Professor of Clinical Pharmacy at Keele University, whose support over the last decade has seen Define become the primary NHS data source for secondary care data. Data has been provided to him by agreement with NHS Trusts to support his specific research projects.
There is no paucity of data provided to the NHS and research establishments granted permission from Trusts (whom we treat as the data owners). In fact, during the early weeks of the Covid-19 pandemic we redeveloped our systems to load data on a daily basis for medicines issues providing the NHS, DHSC and government information on the position of stock and consumption of medicines giving essential early warning of critical stock shortages.
We also responded immediately to requests from NHSE by supplying Covid-related medicines data to the EBM DataLab team on the OpenSafely project led by Dr Goldacre.
There has been no additional cost to the NHS.
Confidentiality of commercial NHS contract prices is a highly pertinent and a topical concern. The NHS Commercial Medicines Unit (CMU) and regional procurement teams make considerable savings against the NHS list price of medicines in secondary care.
More broadly our information helps the NHS manage the risk of monopolies and helps to ensure that the pharmaceutical supply system is robust.
There are also important cases where pricing, subject to Patient Access Schemes, will result in inappropriate analyses or illegal disclosures. Consider also the consequences during negotiations of showing commercial prices.
We treat NHS data confidentially and are mindful of contractual and personal confidentialities. We do not load or process any patient-identifiable data in the system. This is not about secrecy, but about legal constraints.
There are commercial and intellectual property issues which dissuade us from provision of Open Source Code. This would be in line with most software and hardware suppliers to all healthcare settings.
The publication of a comparable, publicly available open dataset via NHSBSA within commercial and legal considerations is imminent, having been delayed for a matter of months while Covid-19 support work was prioritised.
Colin Richman, BPharm (Hons) MPharmS FFCI
Director – Rx-info
Competing interests: Company director. Rx-info Ltd