The role of patient factors, surgical factors and hospital factors upon patient outcomes and NHS costs in the treatment of upper limb musculoskeletal injuries and infections: spatial and longitudinal analysis of routine data

Version 1 extends the Data Sharing Agreement to permit the retention and processing of data. The following provides background on the objective for processing

The University of Oxford requires HES Admitted Patient Care linked to mortality data for the purpose of a retrospective longitudinal study to investigate the trends in surgery undertaken for the treatment of upper limb injuries and infections, and the complications that follow surgery. This study is being undertaken at the Nuffield Department of Orthopaedic, Rheumatology & Musculoskeletal Science (NDORMS) within the University of Oxford.

Upper limb injuries and infections are common and can affect the short and long-term social and occupational function. Emergency conditions (i.e.; hand fractures, dog/cat bites, infections) often need surgery as soon as possible to prevent complications, e.g.; nerve damage, prevent spread of infection to other parts of the body and prevent damage to joint surfaces that lead to post traumatic arthritis.

The complexity of these emergency conditions often requires treatment in dedicated units, especially when the hand is affected. Hand injuries are common, accounting for up to 30% of Emergency Department attendances. Upper limb injuries in general are thought to affect the young, male, working population most commonly, with the majority of injuries sustained either at work or during sport. The prevalence of these injuries among the young, working-age population means they have significant marked economic impact. These costs take into consideration both direct healthcare costs and loss of productivity. Infections of the upper limb can occur in all age groups and are similarly debilitating. Less is known about the incidences and outcomes of upper limb infections. By understanding the burden of surgery needed to treat upper limb injuries and infections on the National Health Service (NHS) and observing temporal and geographic trends, services can be better planned. In addition, understanding injury patterns and risk factors for infections allows targeted preventative strategies to be considered. Lastly, analysis of upper limb injury and infection burden allows clinical research to be directed towards better understanding and managing the most prevalent and impactful conditions.

The research group at NDORMS recently performed and published an analysis of HES data for hand trauma (see https://www.ncbi.nlm.nih.gov/pubmed/?term=Manley+AND+wormald). Researchers used publicly available HES data which is published on line and found that the absolute number of hand injuries per year was 33,948 in 1998–1999 and 59,830 in 2014–2015, a 76% increase. The incidence of hand injuries also increased in this period from 70 to 110 injuries per 100,000 population, a 57% increase. Hand injuries themselves can broadly be divided into fractures and soft tissues injuries. Soft tissue injuries can then be further divided into tendon, muscle, nerve and vessel injuries. Hand injuries also encompass amputation of the hand or digits and soft tissue finger-tip injuries, which usually involve injury to the nail apparatus.

This new study has been founded as a result of the success of this paper.

Despite the commonality of upper limb injuries and infections, there is significant on-going debate and research regarding best management. This is an area of interest to a number of leading research groups in the UK, all of which would benefit from further in-depth analysis of HES data. Additionally, a number of other common hand injuries have been identified as targets for further clinical research by the NIHR, including digital nerve injury (NIHR HTA call for research, 2018) and hand flexor tendon injury (NIHR HTA call for research, 2019). Previous analysis of non-patient level HES data by researchers at NDORMS demonstrates increasing trends in upper limb trauma surgery. NDORMS researchers will use a HES dataset of patients treated for upper limb injuries and infections over the past 20+ years. This pseudonymised, individual case-level data will allow researchers to look carefully at the patient and surgical factors which comprise the scope of upper limb emergencies in England, as well as specific factors that may contribute to differences in outcomes.

The analysis of routinely collected observational data may be limited by the quality of data collection and coding (Kuhn & Mallon 2016). However, the major benefit of HES data is that it represents a comprehensive dataset of NHS Secondary Care in England. It explicitly avoids the problem of selective reporting to a purpose designed research database since reporting of all NHS funded cases is mandated. Whilst technical detail on procedures may be lacking, the research team expect sufficient granularity to inform temporal trends and analysis of patient and hospital factors associated with treatment failure, as identified by repeated need for admitted hospital care.

NDORMS will also focus on variation in outcomes of specific patient groups (e.g. old and frail with comorbidities and obese) and present evidence as to whether the introduction of new surgical innovations (e.g. minimally invasive surgery), and centralisation of services, has led to improved patient outcomes. Civil Registration Mortality data for the patients selected will enable analysis of associations between surgery and death rates.

The researchers require data for processing under General Data Protection Regulation Article 9 (2) (j) and Article 6 (1) (e). The researchers believe that this research is in the public interest as currently there is very little high level evidence available to counsel patients who sustain upper limb injuries and infections on what their outcome is likely to be. The research will investigate geographical trends in injury incidence that will assist in workforce planning and the provision of surgical services. Temporal trends over 22 years will also enable researchers to investigate if any differences in outcome are due to patient factors (past medical history, age) or healthcare factors (admission to hospital).

This study focuses on patients who have sustained upper limb injuries and infections. The data required are limited to only hospital inpatient episodes and mortality records for individuals who had an inpatient care episode since 1997 due to an upper limb information or injury (identified using the ICD code assigned to the hospital episode) which resulted in non-elective surgery (identified using the OPCS code assigned to the hospital episode). Both criteria must apply for an individual to be included in the cohort.

The University of Oxford requires details of all of the data subjects' episodes of inpatient care before the episode due to the upper limb injury in order to gain as full an understanding as possible of their past medical history to see if the individuals have risk factors for infection/complications. All prior and subsequent hospital episodes for the data subjects are required regardless of diagnosis and/or procedure because the study is looking at comorbidities which potentially impact on risks of upper limb injury or adverse outcomes following surgery. It is not feasible to limit the types of previous hospital episodes to specific ICD codes which are presumed to be more likely to be related to risk factors for surgical outcomes as this would bias the findings of the study and could potentially result in failure to identify previously undetected links between certain conditions in past medical history and risk of specific outcomes following surgery. Due to the lack of existing evidence surrounding the role of comorbidities in this research area, it is not advantageous or appropriate to further select a subgroup of patients with certain co-morbidities. Less data than what has been requested would limit the usefulness of the study and be impact the results for informing future patient care.

Details of all inpatient hospital episodes after the individual’s upper limb infection or injury are required to see if they had a complication, or needed further surgery to deal with sequelae of infection or injury (i.e.. develop joint contracture, post traumatic arthritis).

National data is required in order for the results to be generalisable and because one of the main aims of this study is to investigate geographical and temporal trends in the management and outcomes of upper limb injury and infection. For example, the study will investigate links between deprivation and comorbidity and the propensity to incur the injury or infection with a view to recommending targeted prevention programmes in specific areas to reduce the risks of injury and of adverse outcomes post-surgery.

The study team requires both adult and children’s injuries in the extract so will want to include paediatric episodes. Paediatric trauma is as common as adult trauma so excluding this group is non-sensical. The long-term outcome for children with upper limb trauma or infection is potentially even more relevant.

It is specifically desirable not to narrow the cohort by demographics such as age as to do so would introduce bias to the findings of the study. Undertaking a retrospective analysis which includes the very young and very old is advantageous as such groups tend to be excluded from clinical trials. This study will involve focus on variation in outcomes of specific patient groups such as these. Maternal episodes/birth episodes will not be required as they will not be relevant.

In addition to HES APC data, civil registration mortality data would be used to enable risk factor association studies and survival analysis to be undertaken. In this retrospective longitudinal study, the ability to recognise whether a patient is still alive is key to undertaking survival analysis. Civil registration data enables researchers to identify the outcome of patients who sustain polytraumatic injuries (upper limb trauma in association with life threatening injuries elsewhere in the body at the same time) and to identify the true risk of complications following isolated upper limb trauma and infections. This is why knowing both the date of death and the cause of death are key. Requesting the full 22 years of mortality data enables researchers to undertake a long term follow up study that does not occur after clinical trials due to cost and time constraints, and therefore there is a paucity of evidence into long term complications.

The University of Oxford is requesting 22 years of HES and Civil Registry mortality data. Enabling researchers to follow patients up for a long time retrospectively will allow clinicians to be better informed when counselling patients in the future, at the point of surgery, about their options, the risk of them having complications, needing further surgery and long term prognosis. Some complications only occur after many years, and therefore the longer follow up possible the better, as they need to know when and why people died, in order to know if the death was associated with the injury (polytrauma). The date of death will enable researchers to accurately censor the patients within the survival analysis when investigating the true rate of complications (i.e. someone who has died cannot sustain a complication). Researchers want to be able to look at long term complications (post traumatic arthritis, joint contracture etc) which they cannot do if they only have access to mortality data within a certain time period after the event.

National data enables researchers to undertake geographical analysis of the burden of hand trauma and to compare outcomes across England. This enables researchers to identify where greater resources are needed. One of the ways of doing this would be to do an interrupted time series, where statistically one compares the rate of surgery/complications before and after, e.g. introduction of a policy. As this data is observational rather than true research data, further studies, such as interrupted time series’, are needed to identify whether changes in trends over time are due to changes in policy/ hospital factors, or due to patient factors (e.g. ageing population). Having 22 years of data enables researchers to be able to generate a better picture of whether these trends are likely to be due to patient or health system factors.

There is nowhere else in the world that has this length of follow up. There is no current evidence that enables patients to be counselled. Analysis of this data would add a very large body of evidence to counsel patients at the time of consent.

The University of Oxford has considered data minimisation and has taken steps to ensure the data requested is justified and limited to specified upper limb conditions, injuries and procedures. The study team will provide a list of diagnosis and operation codes which relate to upper limb skeletal conditions. Efforts have been made to ensure procedure codes have been combined where possible, e.g. excision of bone/hand, excision of bone/thumb, etc, instead of just excision of bone which may not relate to the upper limb. The research team have produced a tight list of OPCS and ICD fields that restrict the request to only these specific upper limb surgeries.

The interventions that this study wishes to look at have been in use for well over 20 years, and as such the study would like to look specifically at the changes over time in their usage, for example in response to key papers, guidelines, changes in policy.

The University of Oxford is the sole data controller and also processes the data for this study. No other organisations process the data for this purpose. When results are available, an advisory role would be taken by the British Society of Surgery for the Hand BSSH research committee but no data processing would be undertaken by them and all decisions about data analysis would remain with the university. The BSSH would advise about how best to disseminate the results to both healthcare providers and to patients, having recently undertaken a priority setting partnership with patients through the James Lind Alliance (https://www.bssh.ac.uk/patients/bssh_james_lind_alliance_partnership.aspx).

Study Aims

The main aims of this study are:

1. Investigation of the variation in surgical treatments, revision rates and mortality rates in upper limb injuries. Investigation of spatial and retrospective longitudinal trends in mortality will also be studied.

There is substantial debate on an international level regarding how best to treat some of the most common upper limb injuries, including hand, wrist and forearm fractures in adults and children, hand tendon injuries and digital nerve injuries. The key debate for most injuries is whether they should be managed operatively or non-operatively, and within that which interventions are most beneficial. NDORMS wish to investigate these interventions and their outcome in terms of further morbidity and cost.

2. Geographical and temporal trends in management and outcome

Having identified the types of surgical intervention undertaken, NDORMS will investigate whether there are temporal or geographical trends associated with intervention type, compared to geographical and temporal trends in injury prevalence. NDORMS will investigate this comparing patient demographics within regions and over time, and produce maps highlighting these trends. NDORMS will also produce maps highlighting variation in length of stay, readmission complication and re-operation rates across the country as proxy measurements for patient outcome.

3. Assessment of access to care & care costs

Statistical analysis of national data from the HES admissions database will allow identification of hospital organisation and surgical factors that may explain geographical variations in patient outcomes of surgery, after adjustment for patient level case-mix. In this study researchers aim to identify whether the different ways that hospitals organise services for patients presenting with upper limb injuries can lead to improved patient outcomes, and postulate reasons why outcomes may vary between hospitals or regions. NDORMS will investigate whether these differences cause a variation in access to care, due to disease prevalence, or due to variation in management and how these factors influence the cost of patient care. Greater understanding of trends in how these interventions are being used and the outcomes following them will enable NDORMS to propose changes that will influence health service provision and workforce planning.

This dataset has the unique and exciting opportunity of exploring the changes that have occurred in disease presentation, development, treatment and outcome over an extensive time period. Studies that have follow up of this duration have not been previously undertaken in this country, and undertaking research using routinely collected NHS data will enable researchers to better understand how to care for patients with upper limb injuries. Long term follow-up of operative and non-operative interventions allows evaluation of their impact in terms of further morbidity and need for further intervention. Analysis of long term outcomes are vital in order to determine which interventions should continue to be used; evidence that is not available elsewhere.

Alignment with NHS agenda:

In the NHS, patients can choose which hospital they want to have their surgery in. Information on access to treatments and the outcomes of surgery between different hospitals would help patients in making their decision. Outcomes of surgery may vary across different regions and hospitals. Any such differences might be explained by a hospital treating more complex and sicker patients, but could also be explained by the surgical techniques employed in different centres, or centralisation of care into specialist high volume hospitals. Knowledge of this upper limb trauma, injury and complications would inform the development of an NHS Improvement clinical outcome dashboard for trauma. Previous work using elective HES data has informed a similar dashboard for hand conditions, currently being piloted across five sites with national roll-out in 2020. The Chief Investigator has worked on this dashboard with the NHS National Clinical Improvement Programme and has an ongoing collaboration with this group (https://gettingitrightfirsttime.co.uk/ncip/). These dashboards can aid NHS managers and clinicians in changing and optimising service organisation to reduce any variations in outcomes.

The national audit into the orthopaedic surgical procedures called Getting It Right First Time (GIRFT) was launched in 2013. Initial results released in March 2015 found large variations in practice, and have called for better research into the timing and types of procedures undertaken. Better understanding of regional and temporal variations in procedures, and which surgical procedures have the best outcomes would improve the quality of patient care in the UK, reduce costs for the NHS and more importantly provide better patient information to inform shared treatment decision making.

Version 1 extends the Data Sharing Agreement to permit the retention and processing of data. No further data will flow under this Agreement. The following provides background on the processing activities

The University of Oxford is requesting patient level pseudonymised HES Admitted Patient Care and Civil Registry mortality data from NHS Digital. The University of Oxford have requested patient demographic factors (e.g. age, sex, ethnicity, deprivation index); episode factors (e.g. date of admission/surgery/pseudonymised consultant code/HRG); healthcare provision (CCG/waiting times) and data relating to the injury/infection itself (OPCS/ICD codes).

The University of Oxford will not be providing any data to NHS Digital, but just requesting all episodes of care for patients who have an episode containing pre-determined OPCS and ICD codes associated with upper limb injuries and infections. The University of Oxford has requested HES APC data be linked to Civil Registry mortality data to enable survival analysis and will not link this data to any other datasets.

Upon receiving the data from NHS Digital, a senior data manager with experience in managing HES datasets will process the raw data into smaller extracts. Smaller extracts will be made available for analysis based upon disease pathology and surgical intervention undertaken. This will enable the separate research questions defined by the aims of the study to be answered.

NDORMS will provide data at the small area level presented as maps to describe variation in outcomes, before and after accounting for these organisational and surgical factors. All outputs will contain only data that is aggregated with small numbers suppressed in line with the HES Analysis Guide.

Statistical analyses (multilevel regression modelling) of HES data will assess the association of surgical factors on patient outcomes of surgery, adjusting for patient case-mix. Further statistical analyses (random intercept models) will explore geographical variation in outcomes across hospital trusts and Clinical Commissioning Groups. Geographical Information Systems will be used to produce maps depicting variation in outcomes, and graphically display the influence these factors have on explaining such variation.

NDORMS will then use a natural experimental study design to specifically examine the impact that the new treatments have had on NHS resource use, NHS costs and patient outcomes (based upon length of stay, complications, readmission, further surgical intervention including revision surgery). Interrupted time series analysis will examine changes in secular trends in outcomes and NHS costs before and after the introduction of the new treatments. There will be a focus on the benefit of the new treatments to specific patient groups such as frail older people with complex co-morbid conditions. An economic evaluation will describe the hospital NHS costs, patient health related quality of life and cost effectiveness that reflect the new treatments for upper limb conditions. The predominant method of limiting the data requested is through the study of selected upper limb conditions only, and through only selecting certain procedures. Due to the lack of existing evidence surrounding the role of comorbidities in this research area, it is not advantageous or appropriate to further select a subgroup of patients with certain co-morbidities.

A cost-effectiveness analysis will be performed to estimate the economic burden of conservative and surgical care in relation to trends in outcome, adjusting for socioeconomic status and case-mix. This is largely unexplored to date and essential for further studies potentially exploring the cost-effectiveness of the surgical intervention.

Once data has been disseminated to the study;

- The HES and Civil Registration Mortality datasets will be held on a password protected University Computer on an encrypted drive at the Botnar Research Centre, Nuffield Department of Orthopaedic, Rheumatology & Musculoskeletal Science (NDORMS).

- The data will be managed by a researcher based at the Botnar Research Centre Nuffield Department of Orthopaedic, Rheumatology & Musculoskeletal Science (NDORMS) University of Oxford.

- The data will be used exclusively for the purpose of this project.

- 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 would be 24 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.

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 identify any individual from the data being supplied. No data will be onwardly shared.

Throughout all stages of this project, NDORMS will engage with key stakeholders including NHS managers, healthcare professionals, patients and the public for interpretation, dissemination and direct communication of the main findings. This will be facilitated through collaboration with the James Lind Alliance, support of specialist societies, and Patient and Public Involvement (PPI) representation. A Professor of Plastic Surgery at University of Oxford is a named co-applicant on this study and the leader of the BSSH research committee. He will assist in interpreting results and with the national dissemination of findings. This project has also been informed by results from the recent James Lind Alliance (JLA) Priority Setting Partnership (PSP) for surgery for common hand and wrist conditions, carried out in association with the British Society of Surgery for the Hand (BSSH). NDORMS shoulder and elbow, and hand and wrist applicants both have national roles and collaborations that provide excellent access and influence to disseminate the study findings nationally and internationally through the following societies and funded research centres:

1. British Society for Surgery of the Hand (BSSH)- Dissemination to all British hand surgeons and hand therapists. Presentation at the biannual National Congress.

2. British Association for Plastic, Reconstructive and Aesthetic Surgery (BAPRAS) - Dissemination to all British plastic surgeons, hand surgeons and hand therapists. Presentation at the biannual National Congress.

3. British Orthopaedic Association (BOA) - Dissemination to all British orthopaedic surgeons, hand surgeons and hand therapists. Presentation at the biannual National Congress.

4. NIHR Oxford Biomedical Research Unit/Centre – Dissemination to all linked patient and local GP networks

5. Internationally NDORMS will disseminate through peer review publications and via presentations at the Federation of European Societies for Surgery of the Hand (FESSH)

One of the NDORMS professors using this data has written national guidelines for NICE and the specialist societies on managing many shoulder conditions including authoring national commissioning guidelines.

A senior researcher on this study, who is an employee of the University of Oxford will use the aggregated, anonymised results generated from this study to influence practice nationwide. BSSH is in the process of gaining accreditation from NICE for guideline development and NDORMS anticipate this to be in place by the time the results of this work are published.

Working with and informing all stakeholders will remain an important part of NDORMS dissemination plans. NDORMS recognise the importance of meaningful PPI involvement and have worked collaboratively with the PPI Officer at NIHR Research Design Service (RDS) to identify individuals to become involved, and NDORMS Director of Patient Involvement at the Oxford NIHR BRC. NDORMS have identified three lay people who understand the needs and problems of upper limb conditions. Through their involvement and recommendations regarding the dissemination of findings, NDORMS will ensure results are readily available and interpretable to the wider patient and public community.

NDORMS shall disseminate findings in peer-reviewed journals, at national and international conferences, and inform learned societies that include the British Orthopaedic Association, The British Shoulder and Elbow Society (BESS), British Society for Surgery of the Hand (BSSH). NDORMS will work alongside charities and learned societies to disseminate the findings of this study using established platforms that include social media such as Twitter and a study website, as more patients are now turning to these resources for information about planned surgery.

Based on the findings NDORMS will write scientific papers for submission to high quality peer-reviewed journals. NDORMS will also present findings to professionals at conferences and meetings, will develop Plain English summaries of findings for communication to patients and members of the public. All outputs will adhere to the HES analysis guide so that data is only shown in aggregate form with small numbers supressed. NDORMS will publish a full and complete account of that research in the NIHR HS&DR Journal, ensuring the research is reported fully, and publicly available via the NIHR Journals Library website and Europe PubMed Central. A webpage will be developed within the NDORMS website specifically for this study in order to further transmit the results to the public. This study aims to capture the attention of patients and the public by presenting the long term results of surgery for upper limb conditions in the UK not previously undertaken, and to also present the potential reasons why there may be variation in outcomes following surgery. Previous PPI work has shown that variation in disease progression, and outcome following intervention is of particular interest to patients and the public.

The University of Oxford has employed a post-doctoral research fellow, who is a specialist data scientist, to undertake the analyses described above, as well as undertake a separate analysis looking at the epidemiology of surgical site infection following upper limb trauma in adults and children. This will form part of his DPhil at the University of Oxford in collaboration with the Chief Investigator of this research project.

Results of analysis were used to support a NIHR HTA funded trial of flexor tendon surgery techniques . These results have also been published in a text book that was disseminated internationally at the IFSSH conference in June 2022 (Tendon Disorders of the Hand and Wrist; IFSSH/FESSH Instructional Course Book 2022; Thieme June 2022).

The University of Oxford is requesting an extension of 24 months for completion of the data processing, production and dissemination of the outputs due to an inability to access data for over 18 months during the covid-19 pandemic. Whilst good progress in data processing has been made to date (details of outputs already generated in following section), overall production and dissemination of the programme of work is behind schedule due to lack of data access associated with the national lockdowns and inability to use the secure data environment where the data was held. The University of Oxford aim for a further 12 months retention of data after this to respond to changes based on peer-review comments from journals and from funding bodies.

Benefit

The study will inform patients, NHS managers, commissioners and health professionals of the NHS costs and patient outcomes and cost-effectiveness associated with the treatment of upper limb injuries, and the key elements that are most clinically and cost effective. It will provide patients with information on variation in outcomes of surgery to inform patient choice and decision-making. NDORMS will work alongside charities and learned societies to disseminate the findings of this study using established platforms that include social media such as Twitter and a study website, as more patients are now turning to these resources for information about planned surgery. NDORMS will provide evidence of modifiable hospital organisational factors that can explain unwarranted geographical variation in patient outcomes of surgery. This can be used to inform healthcare organisations of factors identified as improving patient outcome and both local and national level, and can be used by clinicians and policy makers to inform healthcare policy.

Impact

The exploration of current trends in procedures undertaken, in temporal and geographical variation in operation numbers and type, mismatch of prevalence and surgery rates, and of outcomes will enable NDORMS to understand more about the current management of upper limb conditions and how to improve NHS services nationwide. NDORMS will explore variations found to identify whether differences in the way hospitals or regions organise their services, such as specialist surgeons, use of new surgical techniques, or centralising care into specialist hospitals, can explain any observed variations. Similarly, understanding the variation in outcomes following a range of procedures enables greater knowledge of which interventions should continue to be funded. Knowledge of these factors would inform changes that can be made to the way services are organised and provided, leading to better access for patients and helping to standardise evidenced based care and patient pathways across the UK. The target date for output and dissemination to produce measurable benefit is now 48 months from receipt of data, incorporating this 24 month extension to make up for the lack of data access due to national lockdowns (Data being received in Spring 2020 right before the time of initial lockdown).

The aim of the rest of the project is to systematically look at other areas of hand and upper limb injuries in the same manner, covering topics such as fractures, infections and burns in order to identify trends in injury presentation and treatment over the past 20 years.

As at November 2022, this data has been used to generate evidence surrounding the treatment of flexor tendon injuries of the hand. This has been presented internationally and results of analysis have been used to support a NIHR HTA funded trial of flexor tendon surgery techniques.