Improving outcomes for patients having shoulder replacements: guiding patient selection, evaluating cost-effectiveness and informing NHS provision

In this agreement, the University of Oxford requires Hospital Episode Statistics (HES) Admitted Patient Care (APC) and HES Outpatients (OP) data and linked Civil Registration (mortality) Secondary Care data for the purpose of a longitudinal, retrospective study to investigate temporal trends, geographic trends and variations in access for shoulder replacement surgery in the NHS. The study team wish to study the complications that follow surgery, and to predict the projected future burden of shoulder replacement surgery on the NHS which is crucial for service provision planning across the country for the next 25 years.

This study will be undertaken at the Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal Sciences (NDORMS) at the University of Oxford. This study has been funded by the National Institute for Health Research (NIHR) through a Doctoral Fellowship Award and will attempt to answer key questions raised by patients, carers and researchers through the national 2015 James Lind Alliance Priority Setting Partnership. The study has also been endorsed by the National Director of Clinical Improvement for the NHS and the British Elbow and Shoulder Society (BESS). Further patient and public involvement through the Shoulder Research User Group (SHRUG) has reiterated the importance of this study to patients.

Shoulder pain is associated with increased health care utilisation and accounts for 20% of disability claims for musculoskeletal disorders. Degenerative shoulder osteoarthritis causes pain, functional limitation and disability and has an estimated prevalence between 4% and 26%. Patients with bilateral shoulder arthritis can rapidly lose function and be unable to self-care. Shoulder arthritis therefore leads to significant morbidity, particularly in an ageing population. Over 45,000 shoulder replacements were undertaken in the UK between 2012 and 2020.

Despite its high prevalence, there is ongoing treatment uncertainty with no high-quality evidence to guide the choice of the different shoulder replacements that are marketed. An evidence review conducted in 2009 (American Academy of Orthopaedic Surgeons) was published but unable to identify any strong evidence to support any of their 16 recommendations on shoulder replacement surgery. A 2010 Cochrane review reporting the effectiveness of different surgeries for shoulder arthritis determined that the overall lack of evidence precluded any conclusions being drawn about the benefit and safety of shoulder replacement surgery. In 2020, another Cochrane Review confirmed an ongoing lack of high-quality evidence on the topic of shoulder replacements. The 2020 NICE guideline on hip, knee and shoulder replacements further highlights the need for high-quality evidence to inform the surgical management of shoulder osteoarthritis and the need for clinical and cost-effectiveness comparisons between different types of shoulder replacements. With improving shoulder replacement outcomes being a priority for patients, there is an urgent need for better evidence and service planning in the NHS.

This project is aligned with the National Health Service (NHS) agenda. The NHS Improvement (NHSI) Getting It Right First Time (GIRFT) programme is dedicated to improving patient outcomes at a reduced cost to the NHS, by reducing regional variation in practice guided by evidence-based care. While other joint replacement procedures are undergoing service provision changes, much better evidence is now needed before any acceptable and effective national guidance for shoulder replacements can be made. Through this proposed project, researchers at the University of Oxford will gain an improved understanding of the burden of shoulder replacement surgery on the NHS now and in the future by analysing temporal trends, geographic trends, and any variations in access to shoulder replacement surgery based on population demographics.

This proposed project will form part of a larger research agenda on improving the outcomes of patients having shoulder replacements that will also include analysis of data from the National Joint Registry (NJR). This agreement containing the request for HES and Mortality data will specifically address one Work Package of the larger research project, namely the current and future burden shoulder replacement surgery on the NHS, including costs of shoulder replacement surgery and an analysis of any geographical variations or inequity of access.

While data from the National Joint Registry (NJR) collected as a part of the larger research project will provide useful surgery-specific information over a limited timescale, the HES and mortality data requested in this agreement will represent a more comprehensive dataset of NHS secondary care in England and enable a long-term follow up of patients having shoulder replacements allowing an assessment of complications, further surgery, as well as changes in practice, regional provision per population and any variations in access per population demographic. This will all inform future NHS service provision of shoulder replacements for maximum benefit to all patients and society.

The requested pseudonymised, record-level HES and mortality data will enable the research team to fulfil its aims and address patient and healthcare information needs. The NDORMS study team will explore how the change in service provision (number of surgeons and hospitals) corresponds to the demand for shoulder replacements. Trends in associated healthcare costs will be estimated through HES Admitted Patient Care as well as HES Outpatients data. Civil Registration (Mortality) data will enable analysis of the association between surgery and mortality and its temporal and geographic variations, including equality of access across England and across different patient groups.

NDORM’s lawful basis for processing data under GDPR has been reviewed and been assessed as acceptable. The University of Oxford process data under Article 6(1)(e): "processing is necessary for the performance of a task in the public interest or in the exercise of official authority vested in the controller" as they are a Public Authority.

Additionally, the University of Oxford process the Special Category Health Data under Article 9(2)(j): "processing is necessary for archiving purposes in the public interest, scientific or historical research purposes or statistical purposes in accordance with Article 89(1) based on Union or Member State law which shall be proportionate to the aim pursued, respect the essence of the right to data protection and provide for suitable and specific measures to safeguard the fundamental rights and the interests of the data subject" as the data are required for historical research purposes in the public interest. This research is in the public interest as there is very little high-quality evidence available to inform shoulder replacement service provision and workforce planning.

The University of Oxford is the sole data controller who also processes data. When the results of the study are made available, the study group will work closely with British Elbow and Shoulder Society (BESS) and NHSI GIRFT to improve translation of the study results into national policy. No record level data processing will be undertaken by these groups. All outputs will be aggregated with small number suppression applied as per the HES analysis guide.

This study will be funded by the NIHR who will publish summary details of the study on their website but will not be involved in data processing or decisions about data analysis.

The University of Oxford did not provide any patient data to NHS England but did provide NHS England with a list of OPCS codes for data minimisation. NHS England linked HES APC and HES OP data to Civil Registration Mortality data, and prepared a pseudonymised, case-level dataset. This dataset was transferred to the University of Oxford via Secure Electronic File Transfer Service (SEFT) and securely stored and processed as described below. Based on historical data and the projected number of shoulder replacements over the year (2021) the study team are expecting the resultant data extract cohort size to contain approximately 75,000 procedures (1998-2021).

The data provided is as detailed below – one drop of record level pseudonymised data:

HES APC data spanning the years 1998/1999 to 2020/2021,

HES OP data spanning the years 2003/2004 to 2020/2021,

Civil Registration (Deaths) Secondary Care Cut linked to the HES data spanning 1998/1999 to 2020/2021

DATA MINIMISATION

NDORMS has considered data minimisation and has taken the necessary measures to ensure the requested data is fully justified and limited to the required cohort. Details of these are outlined below:

- Datasets and linkages: The study is focused on patients who have had a shoulder replacement. The data required are limited to only hospital inpatient and outpatient episodes and mortality records for adult patients aged 18 years and over who have had a shoulder replacement since 1998 identified using the OPCS procedure codes code assigned to the hospital episode. The study team provided a list of operation codes (OPCS) to ensure a minimised, yet sufficient cohort of patients were identified. The datasets provided were pseudonymised as this will enable the research questions to be answered least intrusively.

Civil Registration Mortality data linkage was required for two reasons. First, it is crucial to know whether a patient is still alive when undertaking survival analysis in such a longitudinal study. Second, it is important to identify the mortality rate following shoulder replacement or following a direct complication of shoulder replacement surgery. Both the cause of death as well as the date of death was therefore provided, and the 23-year follow-up will enable researchers to produce the most accurate estimate of shoulder replacement burden and projected future burden to the NHS.

- Years and filtering: Complications after shoulder surgery including revision surgery for failing implants can occur many years after the index procedure, and the 23-year follow-up data requested will allow these events to be accurately captured. In order to most accurately analyse the temporal trends in shoulder replacement surgery, The University of Oxford requested long-term data covering 23 years. This enabled researchers to identify trends that occurred over a number of years which may correspond to changes in practice or the introduction of national guidelines. Having long term data is also critical to the production of the most accurate estimate of shoulder replacement burden and projected future burden to the NHS over the next 25 years.

Temporal trends spanning a number of years will also allow researchers to best forecast future costs and burden on the NHS. National (England only) data is required to identify geographic variation in the demand and provision of shoulder replacement surgery and to investigate whether deprivation has an influence on patients’ outcomes following shoulder replacement. It also helps to target the expansion of healthcare resources to the areas of greatest need in future.

- Episodes and fields: The study team require details of all the data subjects’ episodes of inpatient and outpatient care before the episode where they had their shoulder replacement (excepting Maternal episodes/birth episodes are not required as they are not relevant to this study) in order to gain a better understanding about their past medical history which may contain important information about risk factors for complications and revision surgery. Limiting previous episodes for the data subjects based on ICD codes may introduce bias and prevent the identification of important associations between risk factors and outcomes, so it is necessary to include all prior episodes. There is limited high-quality evidence to be able to confidently identify which risk factors are important predictors of outcome following shoulder replacement. It follows therefore that all previous patient episodes and code sets are required in order to ensure all true risk factors and associations are highlighted, and to prevent inadvertently discarding risk factors due to assumption. Furthermore, certain comorbidities may have been identified in patient episodes that may not be directly related to their shoulder replacement episode (e.g. medical comorbidities identified during a patient’s attendance for appendectomy). All subsequent hospital episodes for the data subjects are also required to ensure all inpatient and outpatient events that may be associated with the shoulder replacement procedure including revision surgery, hospital appointments and other treatments (however long after the index operation) are accurately captured and their costs accounted for to reflect the true burden of shoulder replacements.

DATA ACCESS

Once the pseudonymised data has been received by the University of Oxford from NHS England:

- The HES and Civil Registration Mortality datasets is held on a password protected University of Oxford Computer on an encrypted drive in the Secure Data Room at the NDORMS Botnar Research Centre. NDORMS holds an up to date NHS Data Security and Protection Toolkit which provides data security assurance for processing and storing NHS data. Data will be encrypted to AES-256 standard as per IG07 NDORMS Confidential Data Storage and Destruction Policy.

All data will be processed only by substantive employees of University of Oxford. Access to the data will be restricted to named individuals working on the project and based at the NDORMS Botnar Research Centre who have received suitable training and will only access the data for the purposes described in this agreement. Those accessing NHS England data at NDORMS are required to take the University wide annual information security awareness module which is mandatory for all staff and students, and undergo an induction to the secure data room with the Information Governance Manager.

DATA ANALYSIS:

NDORMS will carry out statistical analysis to model the change in volume and secondary care costs of shoulder replacements over the 23-year period provided by the NHS England HES data. Correlations between variables will also be investigated such as disease prevalence and patient demographics. Geographic variations in outcome will be explored across hospital trusts and Clinical Commissioning Groups and the influence of these factors on a variety of outcomes (such as length of hospital stay, waiting times, revision and complications) will be highlighted. Variations in access to shoulder replacement surgery across different geographic area, different patient cohorts including levels of deprivation will be reported. Surgical volume will be projected for the following 25 years by applying age- and sex-standardised shoulder replacement rates to national (England only) population forecasts, a method previously applied to the projection of cataract surgery growth in Canada. This will be contrasted with different scenarios of projected growth in the service provision in England based on the analysis of observed trends from the NHS England HES data.

At the end of the study, the data will be safely held in a password protected University Computer at the Botnar Research Centre for 60 months and, in that time, it will be assessed only to answer questions arising from the publication and other publicity. The interim expected timeframe for completion of the data processing, production and dissemination of the outputs is 36 months, with a further 36 months retention of data after this to respond to changes based on peer review comments from journals and from funding bodies. An active data sharing agreement will be held with NHS England during this full data retention period.

All data will be processed only by substantive employees of University of Oxford who have been appropriately trained in data protection and confidentiality.

Data will not be linked to any other record level data. No attempts will be made to re-identify any individual from the data being supplied.

In order to protect patient confidentiality, when presenting results calculated from HES record level data, outputs will contain only aggregate level data with small numbers suppressed in line with HES Analysis Guide.

Throughout this project the University of Oxford aims to actively engage with key stakeholders including NHS managers, professional bodies and patients and the public for results interpretation and dissemination. The research team at NDORMS has strong national collaborations that expect to facilitate effective dissemination of the research outputs to researchers, scientists and healthcare policymakers:

1. NHSI/Getting It Right First Time (GIRFT): A Professor of Orthopaedics (co-applicant on the study) advises GIRFT and has previously authored the new recommendations for provision of elbow replacement surgery in the NHS. The NHS National Director of Clinical Improvement fully supports this study so it is hoped it will gain high-level attention, enabling the results to directly translate to service provision changes on a national scale.

2. British Elbow and Shoulder Society (BESS): A Professor of Orthopaedic (co-applicant on the study) is President Elect for BESS and hopes to be able to facilitate the updating and authorship of BESS national guidelines depending on the results of the study.

3. National Institute for Health and Care Excellence (NICE): NDORMS expects to update NICE on any important findings which have the potential to translate to national change in clinical practice.

4. National Joint Registry (NJR): Two co-applicants on this study (a Professor of Orthopaedics and a Senior Research Fellow in Medical Statistics and lead data analyst for the NJR) work closely with the NJR. Dissemination to surgeons and the public will be supported by the NJR who reference important research projects in their annual report, NJR symposium at the British Orthopaedic Association, and through NJR patient information.

5. NIHR Oxford Biomedical Research Centre: Dissemination through the press, media and charities.

Researchers also hope to present the study findings at the below national and international conferences in the form of podium and poster presentations:

1. British Orthopaedic Association (BOA): This is the largest Orthopaedic conference in the UK and it is expected provide NDORMS with a platform to present the study results to all British Orthopaedic surgeons. This meeting is particularly well attended by Orthopaedic trainees and offers an excellent opportunity to create maximum impact on future consultant surgeons.

2. British Elbow and Shoulder Society (BESS): NDORMS hopes to disseminate the results of the study through BESS to all British shoulder surgeons in the form of conference presentations at the annual scientific meeting.

3. European Shoulder and Elbow Society (SECEC): Dissemination to shoulder surgeons on an international level

All outputs will adhere to the HES analysis guide with data being shown in aggregate form only with small numbers suppressed as per the HES analysis guidance.

Patient and Public Involvement (PPI):

NDORMS recognises the importance of meaningful PPI involvement and patients have been involved at all stages of this project. The importance of this study was identified and prioritised by patients and the public through their participation at the 2015 James Lind Alliance Priority Setting Partnership on shoulder surgery. NDORMS has already worked with patients from both the North (Shoulder Research User Group) and South (Oxford) of England to formulate this study plan including the larger research agenda on shoulder replacements. The study team have worked with the PPI Manager at the NIHR Oxford Biomedical Research Centre to identify patient representatives. The study team are setting up a patient advisory group for this study and a patient expert from the NICE Guidelines Joint Replacement Committee has already joined. The study team have collaborated with the PPI Senior Research Officer at the NIHR Research Design Services to make plans for effective dissemination of the study results so that they are easily accessible and interpretable to the wider patient community and to the public.

The study team plan to disseminate the study findings through scientific publications at peer-reviewed journals. These publications are expected to be open-access to maximise dissemination to the target audience of researchers, scientists, patients and the public. Open-access fees have been secured in the NIHR funding for this project. NDORMS hope to work with the new BESS Expert Patient Group to co-produce patient-friendly summaries including patient perspectives of the project results and infographics. NDORMS expect to publish a full and complete account of that research in the NIHR Health Services and Delivery Research Journal, ensuring the research is reported fully, and publicly available via the NIHR Journals Library website and Europe PubMed Central. A webpage is also expected to be developed within the NDORMs website for this study in order to improve dissemination of the results to the public.

The University of Oxford hopes that this study informs NHS commissioners and healthcare professionals about the likely future burden of shoulder replacements on the NHS both in terms of cost and workforce planning. Key information about the geographic variation in surgical outcomes should help inform patient-choice and decision making as well as highlight modifiable organisational factors that will be useful to NHS commissioners to improve the service provision of shoulder replacements. NDORMS hope to work closely with NHSI so that the evidence from the temporal and geographic analysis of trends using HES data translates to improved provision.

This study forms part of a larger research agenda to improve the outcomes of UK patients having shoulder replacements as part of a surgical trainee’s doctoral thesis at NDORMS. Together with the current study, this overarching project aims to provide patients and clinicians with long overdue evidence to inform shared decision-making around shoulder replacement surgery with the aim of improving patient outcomes. It hopes to provide answers to key research questions raised by patients and the public through the James Lind Alliance Priority Setting Partnership (JLA PSP) and recent Cochrane and NICE evidence reviews.

The target date for the above outputs and dissemination that is expected to translate to measurable benefit is 36 months from receipt of the data.

Updated Outputs 24/04/2024.

Papers

Valsamis EM, Pinedo-Villanueva R, Sayers A, Collins GS, Rees JL. Shoulder replacement surgery’s

rising demand, inequality of provision and variation in outcomes: cohort study using Hospital Episode

Statistics for England. BMC Medicine. 2023; 21(1):406. PMID: 37880689

Presentations

Valsamis EM, Pinedo-Villanueva R, Sayers A, Collins GS, Rees JL. Shoulder replacement surgery’s

rising demand, inequality of provision and variation in outcomes: cohort study using hospital

episode statistics for England. EFORT, Hamburg, Germany, 2024 (Speaker: EMV) (Pending)

Valsamis EM, Pinedo-Villanueva R, Sayers A, Collins GS, Rees JL. Temporal trends and future

healthcare burden of elective shoulder replacement surgery: exploring geographic and

socioeconomic variation in service provision and patient outcomes using Hospital Episode Statistics

for England. BOA Annual Congress, Liverpool 2023