FP-00027

Section 1 - Basic information about you and your application:

Title of research project
Establishing the feasibility of real world surveillance of orthopaedic devices using routinely collected data

Grant Type
The ORUK Early-career Research Fellowship

Research area
Diagnostic and treatment

Duration
24

Start date
January 1, 2024

Have you previously received funding from ORUK?
No.

Profession
Orthopaedic surgeon

Your current job title/position
Academic Clinical Lecturer in Trauma and Orthopaedic Surgery

Are you an early-career researcher (ECR)? (definition of ECR)
yes

Section 2 - Lay summary

Lay summary:

Doctors are putting thousands of medical implants into people every year, but unfortunately most of these implants are difficult to monitor for their safety. This has led to some people being harmed. 

  

Implants used in bone and joint surgery make up 65% of all devices used in the NHS. Information is collected for some implants like hip and knee replacements, but not the majority used in other operations. This project will test if it is possible to collect information about bone and joint implants so that future surgeries are safer for patients and better value for money for the NHS. The project will also focus upon how it is best to inform patients if an issue is found with a device they have had implanted, and how they would like their data to be used to help them get the best care.  

 

A government led enquiry in the UK was asked to focus on medications and a medical device that have been found to have harmful side effects. Many patients contributed to the enquiry, but unfortunately because routinely collected big data is currently not fully utilised, no one actually knows how many patients were harmed. 

 

Ambition 

My ambition is to identify the implants used in bone and joint surgery through using data already collected within medical treatment. I aim to do this efficiently by converting data into a common data format, to generate a sustainable method of replicating research in other centres in both the UK and internationally. This will enable the research to be reproduced in real time to identify if complications associated with a certain device or population group are seen in multiple hospitals.  I aim to build a centre in the UK that leads surgical big data research using anonymised routinely collected data.  I aim for this centre to be known worldwide for inclusive, collaborative and innovative research, establishing a dynamic international orthopaedic community for future projects and raising the profile of big data research.

 

Impact 

This project is a direct response to the national enquiry into drug and implant safety, designing a method for quickly identifying patients who have been treated with implants that may be causing harm. It aims to offer solutions to better monitor patients who are at risk, and to streamline the work needed in the NHS to monitor devices. Alongside this, I will include patients and members of the public from the beginning to ask them how they would best want to see their data used, and how best they would like to have information about implant safety given to them.

Section 3 - Purpose of research

Purpose of research:

I will use routine data from one large NHS Trust mapped to a common data model. I will establish the feasibility of orthopaedic implant safety surveillance using longitudinal data linked between primary and secondary care, alongside mortality data to identify complications associated with device use. Through this fellowship I will establish the feasibility of data driven decision identification of devices at risk of causing harm, and to develop infrastructure for a UK centre leading international collaborative big data device research 

The objectives of this fellowship are to: 

1. Establish the feasibility of surveillance of orthopaedic devices using routine NHS data  
2. Identify if orthopaedic device surveillance is possible within an international common data model, enabling multi-centre replication 
3. Undertake surgical and device research in underrepresented patient groups 
4. Build an infrastructure upon which to lead future work in national and international post market surveillance of orthopaedic devices, externally validating results internationally to address biases 
5. Establish the patient perspective upon how best to involve them in care pathways should a device appear to have higher levels of complications 
6. If sufficient time, or international replication does not appear feasible, to determine if this data driven approach can be used to identify comparable cost effectiveness of orthopaedic devices  

This fellowship aims to deliver:  

1. A road map for orthopaedic device surveillance in UK routinely collected data 
2. A reproducible analytical pipeline for replication of this study in other UK sites 
3. Feasibility studies of international replication within North America and Europe including A.Temporal and geographical trends in device use B.Demographic features associated with adverse outcome
4. An international collaboration of academic and clinical partners with big data led by an ORUK fellow, engaged in producing real world evidence in orthopaedic surgery
5. Health economic analysis using routinely collected data to estimate comparative cost effectiveness

Section 4 - Background to investigation

Background to investigation:

Literature review 

The Cumberlege report has shown it is vital for us to develop a better understanding of medical devices surveillance, and of how routinely collected data could be used for the benefit of surgical patients, mirroring the pharmaceutical world.[1] The enquiry was unable to evaluate the magnitude of complications resulting from the three interventions investigated due to a lack of available information regarding their use. This emphasises how research focussed upon leading efficient, reproducible post market surveillance of devices in the UK is vital for patients and clinicians alike. Increased agility and efficiency is also required to generate meaningful evidence in the modern healthcare climate, especially as financial constraints upon research emerge and healthcare environments are forced to change at short notice.[2] The ever-increasing generation of big datasets as a reusable result of clinical interactions has increasing benefits for orthopaedic research.[3-7]

Surgical epidemiology using big data, particularly real world data (RWD) is in its relative infancy in comparison to pharmacoepidemiology.[8,9] Big, well organised, RWD data offer an alternative to the idealised environment of the randomised trial (RCT); offering prognostic insights unavailable from RCTs and including whole populations and people who are underrepresented within trials.[10] Big data can also study complications that accumulate late and which are not financially practicable to report from other study designs.[11]

The modern NHS offers the possibility of designing research in one institution that can be replicated across the nation. The digitalisation of perioperative care also offers the possibility of being able to evaluate the impact of pre, inter and post operative factors upon surgical outcomes, and the relative role of the specific device used. 65% of all devices used in surgical care are within orthopaedic surgery, and therefore harnessing post market surveillance of devices in this field offers the opportunity to develop techniques that could be transferred to other areas of healthcare. 

 A growing number of observational data networks have emerged, including those using a federated network where data remains at source.[12-15] International federated networks such as the European Health Data Evidence Network (EHDEN) and the Observational Health Data Sciences and Informatics (OHDSI) network enable studies to be replicated as precisely as possible within data that is mapped to a Common Data Model (CDM).[16,17] Using a common data model offers the possibility of generating reproducible analytical pipelines and harnessing the power of collaborative research within a diverse data science community. They also offer the possibility of replication in other forms of routinely collected data at a national and international level, to compare the potential causes of bias at a patient, surgeon, hospital, or healthcare system level. 

This work therefore aims to identify the feasiblity combining these areas of research need with new epidemiological innovation. It will use routinely collected surgical data, mapped to a common data model to undertake post market surveillance of devices. It also aims to develop the expertise, infrastructure and academic community in this domain in orthopaedic surgery as a legacy of the fellowship.

Preliminary data

The host institution has already mapped the bulk of their electronic healthcare record (EHR) to the Observational Medical Outcomes Partnership (OMOP) CDM within a robust iterative process. Conditions, demographics, observations and mortality data now exist within the CDM, but perioperative and device data has not been mapped. This project would lead mapping the perioperative fields into the data model, including intraoperative events, anaesthetic type, unique device identifier and post-operative analgesia use. Collaborations with 3 international academic centres (Oxford, Barcelona Spain, Maine USA) with expertise in using the OMOP CDM have already begun to determine the feasibility of converting surgical data and generate counts for model procedures. Memoranda of understanding are in progress to be signed within quartile 4 of 2023. Locally, I have developed links within the host institution informatics department, supporting expertise in the OMOP CDM for future national projects in device epidemiology. Local collaborations are in place to provide expertise in health economic analysis, and a local PPI group has already been established to support the fellowship to explore the role of using routine data in patient care. 

Personal track record

I believe I have both the personal and scientific skills to use all the opportunities an ORUK early career fellowship would offer me. I will use this independent funding to drive this new initiative and to build the foundation as a future leader of surgical and device epidemiology. 

 Leadership qualities

I will lead successful post doctoral research in surgical and device epidemiology due to the skills I developed leading pharmacoepidemiological research that changed healthcare policy during the Covid pandemic. Within the OHDSI international community, I led several high impact research projects with multiple international collaborators to generate real world evidence, including over 12 publications. Two publications were first author and led to EU and US regulatory policy change, were featured in the media and gained an altmetric of over 1800.[18] I was awarded the OHDSI titan award in 2021 for clinical application by the community. 

Scientific endeavour

In 2021 I graduated from a PhD in Musculoskeletal Sciences Data Science and have published over 34 peer review articles including 3 in the BMJ, 2 first author Lancet Rheumatology and 1 Nature Communications article, with a H index of 16. I have been invited to speak at the EU IMI Horizon 2020, the BOA, the British and American Hand societies and author of the epidemiology chapter of the textbook for the triennial global hand conference (IFSSH). 

Past grants and use of training

I began my academic clinical career as an NIHR ACF in 2014, during which I developed my PhD project and gained doctoral training fellowships from Versus Arthritis and the Medical Research Council. Alongside my PhD, I was a successful co-applicant on a £100,000 pump priming grant, delivering a project on the national state of UK trauma hand surgery using NHS data. I am a co-investigator in the £1.35M NIHR HTA funded trial ‘FLARE’ investigating flexor tendon injury repair. 

Section 5 - Plan of investigation

Plan of investigation:

This project contains four work packages.

WP1- Feasibility, device mapping and analysis package generation in 1 NHS centre

Study Design

I will establish the feasibility of undertaking epidemiological studies investigating adverse events associated with orthopaedic device use, and the temporal and geographic trends in patient demographics and treatment undertaken. Studies will be developed in UK data, undertaken in anonymised routinely collected data mapped to the OHDSI OMOP CDM.[1-3] 

Setting

I will use 1 NHS dataset of routine healthcare records covering primary, secondary care, intensive care and mortality data. 2.4 million patients are currently included in the host mapped dataset over a ten-year life course; data are refreshed bimonthly. Mapping of the host institution to the CDM is already complete for routine fields due to a successful grant awarded by the EU. In this fellowship I will work with named collaborative partners to map in addition surgical procedure and device data collected at the time of surgery, including unique device identifier, into the OMOP CDM. I will undertake mapping of this data into the CDM within an iterative process with data scientists, with the aim of also developing methods converting reproducible analytical pipelines for drug utilisation already used in the OHDSI community for device utilisation analyses. 

Participants

Patients undergoing orthopaedic surgery; the main cohort will focus on those over 18 years. Due to the volume and rich heterogeneity of patients in the data, subgroup analyses will be undertaken wherever possible to cover those with atypical disease (eg implants used to fix pathological fractures) or from traditionally smaller demographic groups (eg ethnic minority groups, paediatrics, extremes of age, multi-morbidity)

Variables

Studies undertaken will include

Exposures: orthopaedic implant of choice

Covariates: surgical procedure and anaesthesia undertaken

Outcomes: local complications (surgical site infection, bleeding) systemic complications (such as VTE, MI, pneumonia, AKI, stroke, UTI), revision surgery and death. 

Data Quality & Validation studies

Data quality will be checked within every step of the process, including during the conversion of the surgical fields to the CDM to ensure source data is appropriately represented. Prior to study design, data quality dashboards will be used for the overall value of the data source, and open source software used to identify the most appropriate way to define a population during study design.[4-6] 

Study size

No formal a priori sample size will be calculated, including all patients who meet the relevant inclusion criteria for each study. Minimal detectable relative risk (MDRR) will be generated for each analysis to evaluate power prior to unblinding.

Statistical methods

All analysis plans will be set apriori in a published protocol. In descriptive epidemiology studies, the number of patients in each subgroup will be reported, along with chosen covariates of interest. Standardised mean differences will be reported and plotted to compare subgroups. In comparative cohort studies, multivariable regression modelling and large scale data driven propensity score analysis will be undertaken to identify associations with outcome using R software.[7-9]

Ethics & Approvals

The main approval for the study would be registered within the EU via the EU PAS register, in addition to at a local level for each dataset used.[10] Standardised analytics are undertaken at source and therefore do not require data sharing outside an institution.

 

WP2- National and international external validation

Following WP1, I will lead external validation of analyses through replication initially within the established collaborators in this study (Oxford, Spain, USA) where memoranda of understanding will be in place. These comparisons will allow quantification of any potential selection bias in the host centre, but also to establish the feasibility of replicating device studies using national and international real world data.

Data sources

A combination of administrative claims data from private and national insurance providers (IBM MarketScan, Pharmetrics), and nationally collected hospital linked-primary care data (SIDIAP-H, Spain) will be used. If successful, the reproducible analytical study will be offered to other members of the OHDSI community to replicate in their datasets. 

Data processing & management

Data used is de-identified with patient data remaining confidential.Data processing and management of each dataset will take place within each collaborator’s own institution, with only metadata being shared for final analysis.This enables collaborators to fully adhere to local data protection law.

All data received from collaborators will initially be blinded and evaluated for power, and likelihood of observed and unobserved confounding prior to inclusion in the meta-analysis.Empirical calibration using negative control outcomes will be used to identify the risk of residual confounding in unobserved covariates, and covariate balance alongside propensity score distributions after adjustment for observed covariates.[11-13] Meta-analysis will be undertaken using a random effects model, with a pre-set measure of heterogeneity of inclusion, likely I2<0.4.[14]

 

WP3-  Patient and Public Involvement 

Using an established PPI group, I will lead discussions regarding the use of data for device surveillance after surgery. This will build upon feasibility work already begun in establishing the role of non-identifiable routine healthcare data in research. It will focus upon how patients and the public would like to be informed about potential complications associated with implant use, and how their data should be used to improve implant safety surveillance. 

 

WP4- Health economic analysis 

This package is designed as an extension within the host institution if international replication fails due to data heterogeneity or work progresses faster than anticipated locally. Collaboration with our trials unit have been established to lead comparative cost effectiveness analyses of implants using OMOP mapped data in the host institution. This will be based upon costs within the operative episode using HRG coding, and including subsequent interactions associated with complications or routine follow up. 

Opportunities & risks for clinical adoption & commercial exploitation

MHRA, the UK regulatory body have recently engaged with UK OHDSI collaborators to explore the use of OMOP mapped data, emphasising the potential opportunity for us to lead national device surveillance with this project. This research has the potential for NHS adoption due to similarities in peri-operative electronic healthcare record content, in addition to internationally in centres already using the OMOP CDM, particularly the 187 partners within EHDEN.

Section 6 - Research environment and resources

Research environment and resources:

Research, Clinical and Technical Support

This project is supported by a strong heterogenous international community of people with data, computing and epidemiological expertise. As a partner within the collaborations of 3 centres, in addition to within the OHDSI and EHDEN communities, I will develop transferable skills in leading surgical and device epidemiology in routinely collected data.  Building on past experience, OHDSI data scientist collaborators in Oxford and the US will support me in the project, bringing methodological support. Within the university, organised data research and technical support has been baked into the design of the project to ensure success, alongside support from a professor of orthopaedic trauma surgery in a mentoring role.

Space and Equipment 

The project is designed to run using a SafeHaven server environment within the hospital trust enabling remote but secure access, maintaining data privacy. The data is anonymised and never shared. It is designed to develop a local appetite for international big data research and develop data scientist expertise in large scale analytics to leverage in future external funding bids. It aims to build a physical and workforce infrastructure for future collaborative data research, designing and developing the OHDSI software tools to expedite data analysis and management in surgical and device epidemiology. Physically, the department has also developed a hybrid meeting space designed for international meetings, and collaborative working on site. This robustly supports the project within the potential of future pandemics and to enable flexible, international and dynamic workspace befitting of an innovative project. 

Additional fundings

The grant requested will support half of my salary to give time to the project, the other half being provided by my host institution. For the data, this funding builds upon a previous EU grant to map the basic fields of the electronic healthcare dataset in my host institution, and a supplementary Medical Research Council grant for access to Spanish data. Combined together, this will enable this early career fellowship grant to enable me to take that important first independent step as a post doctoral researcher, but also to deliver tangible output, aligning with ORUK’s goal in driving development of big data infrastructure for future endeavours. Salary of the senior advisor is funded by the NIHR, with local charitable funding supporting the infrastructure of the secure data environment. This fellowship will enable much more than an individual grant could achieve, bringing together data access, statistical, analytical and data management skills and infrastructure alongside clinical expertise to deliver a sustainable big data research environment. Full details of costings and other fundings available are attached in a pseudonymised spreadsheet.

Section 7: Research impact

Who will benefit from this research?

Patients and clinicians

This work explores whether routine electronic healthcare records can be used for real-time NHS orthopaedic device surveillance as a proactive response to the Cumberlege enquiry. It is a clinician-led response to improve surgical device safety, to generate better estimates for the risk of adverse events and outcomes after orthopaedic surgery. The work focuses upon traditionally underrepresented patient groups, and will identify areas to reduce adverse events and improve treatment equity. 

Policy makers and Researchers

This work is a proof of concept for orthopaedic device surveillance and regulation in routine big data, mirroring methods used in the pharmaceutical industry. Building on past expertise, it will suggest efficient, reproducible methods of using big data to improve regulation of surgery and devices, and to improve NHS resource use. It aims to promote big data research to drive policy, developing a centre of expertise and associated infrastructure for future projects.

How can your research be translated in real-life?

This research is in data generated from everyday, routine care. It will identify outcomes from surgery and device use in routine practice to directly inform patients of their likely outcome at the point of consent. The project translates into the clinical discussions between clinicians with patients and to build patient focussed interactions should a device be identified associated with excess harm. It aims to engage regulatory bodies, which has been achieved in previous projects. Using evidence generated, it aims to drive change in policy from regulatory and governmental bodies in order to reduce variability, improve outcomes in all patient groups and streamline care. Finally, it aims to generate evidence of how future research may be undertaken using big data through involving patients in the fellowship, and open discussions with patients about how acceptable this is, and how best this data can be harnessed whilst remaining appropriate by society.

How will your research be beneficial for ORUK and its purpose?

The large-scale analytic innovations of OHDSI proved vital in rapid evidence generation during the COVID-19 pandemic using big data. This project will harnesses this power for device surveillance, expertise, and community for orthopaedic research aligned to the ORUK agenda of developing big data research. This research is an exciting opportunity to identify if technological solutions can improve efficiency of future observational research, whilst also reducing the human cost of involvement in clinical trials and the time to evidence generation.

This research focuses on orthopaedic devices and surgical treatment, deliberately choosing common implants for the proof of concept work in big data research to simultaneously generate high impact evidence whilst driving methodological advances. The fellowship aims to develop UK expertise and to put ORUK at the forefront of innovations in device surveillance nationally. The research has patients at its core, with patients leading how they see data improving surgical care.

Section 8: Outreach and engagement

All sections of the project aim to be published in the peer review literature, and presented at international clinical conferences to disseminate results to the orthopaedic community. The international and large scale nature of the data lends itself to robust, high impact evidence generation within an appropriate statistical analysis, and likelihood of adoption. The topic of surgical and device outcomes also lends itself to regulatory adoption in the current UK political climate driven by the Cumberlege report.
The PPI group generated to discuss data use and involvement in clinical care will assist in identifying the best way for communicating our results to patients and the public. Locally within our university and associated hospitals the Centre for Public Engagement is the hub for outreach work. The department within which the work is based also has established links with the local community for communication of research to the public.
Within the OHDSI and EHDEN community, I have also been personally involved in outreach and engagement activities for the public and clinical communities to engage with work that can seem very technical, in the form of podcasts, webinars and invited talks. In the summer of 2023, the UK OHDSI node was created of UK based partners in charities, governmental organisations, as well as clinical and academic institutions. The UK OHDSI node has recently engaged with the MHRA who are particularly interested in how research using routine data mapped to a common data model could be used to improve UK healthcare. This work would be of specific interest to this group, and I would aim to involve this group in identifying how to disseminate the work. I aim to develop these skills further within the community to enable outreach and engagement work with non-clinicians, including governmental and regulatory organisations that can best utilise the results of this work to translate it into practice.

Section 9: Research budget

Requested funding from ORUK

University fees (if any)
£0

Salary
£95395.73

Consumables
£4000

Publications
£2000

Conference attendance
£2000

Other items
£6184

Total 'requested fund'
£109579.73



Other items
Data storage and archiving: £2000.00 Conference travel and accomodation: £2500.00 PPI: £1000.00 Shinyapps.io website for sharing results: £684.00



Other secured funds

Internal funding
£164251.47

Partner (University)
£0

Partner (Commercial)
£0

Partner (Charity)
£9500

Other sources
£0

Total 'other funds)
£173751.47

Section 10: Intellectual property and testing on animal

Is there an IP linked to this research?
No

Who owns and maintains this patent?


Does your research include procedures to be carried out on animals in the UK under the Animals (Scientific Procedures) Act?
No

If yes, have the following necessary approvals been given by:

The Home office(in relation to personal, project and establishment licences)?


Animal Welfare and Ethical Review Body?


Does your research involve the use of animals or animal tissue outside the UK?
No

Does the proposed research involve a protected species? (If yes, state which)



Does the proposed research involve genetically modified animals?


Include details of sample size calculations and statistical advice sought. Please use the ARRIVE guidelines when designing and describing your experiments.


There should be sufficient information to allow for a robust review of any applications involving animals. Further guidance is available from the National Centre for the Replacement, Refinement and Reduction of Animals in Research (NC3Rs), including an online experimental design assistant to guide researchers through the design of animal experiments.

Please provide details of any moderate or severe procedures


Why is animal use necessary, are there any other possible approaches?


Why is the species/model to be used the most appropriate?