The findings and methodology of this study will be reported in accordance with the REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement [14].
Source of data
In this analysis data from the National Joint Registry (NJR) will be used. The primary purpose of NJR is to collect data on joint replacements to provide timely warnings of issues relating to patient safety [15]. In doing so, the NJR collects high-quality data that is commonly used in orthopaedic research [15]. The NJR elbow dataset will be the primary data source and it will be linked with data from the NHS England Hospital Episode Statistics-Admitted Patient Care (HES-APC) dataset to incorporate data that are not collected by the NJR.
The NJR started collecting elbow replacement procedures in April 2012. It is currently compulsory for all NHS and independent hospitals in England, Wales, Northern Ireland, the Isle of Man, and the States of Guernsey to submit elbow replacement surgery data, including primary and revision surgery, to the NJR [16]. The NJR collects data using a collection tool called the Minimum Data Set (MDS) form, which is usually filled out by a clinician at the time of surgery. Information regarding the implants used is usually completed by administration staff by including the implant codes (usually stickers that are attached to each implant) in the MDS form. All the information from the MDS forms is entered directly into the NJR data entry system locally by the hospital where the procedure was performed. When the NJR started collecting elbow replacement procedures in April 2012, version 5 of the MDS (MDSv5) was in use. The MDS has been updated twice to version 6 in November 2014 and version 7 in June 2018. Data collection using MDS version 8 started in June 2023, and the changes made in MDSv8 will not impact this project. The changes are summarised in Supplementary File 2.
There is currently a live automated data quality audit process to address any potential missing procedures, which started in June 2020 [17]. A collaborative audit between the NJR, British Orthopaedic Trainee Association (BOTA), British Orthopaedic Association (BOA), BESS and The Royal College of Surgeons of England (RCSEng) showed the completeness of the TER dataset to be 93% and the accuracy to be 98%. This audit included all TER procedures from the start of data collection in April 2012 until the live automated audit started. High completeness of the target population limits selection bias and increases the generalisability of results [18].
The HES-APC dataset contains details about inpatient admissions funded by the NHS, including patients in independent hospitals funded by NHS trusts. All patients undergoing TER surgery are admitted to hospital, and their data will be included in the HES-APC. Each NHS trust collects data while the patient receives treatment. The data are usually collated locally by clinical coders using the patient medical records and then submitted to NHS England. Once collated, NHS England processes and assesses the quality of the data and the yearly HES datasets are released for secondary use, including research. HES includes demographic, clinical, and administrative information which can be used to assess the practices and trends of joint replacement surgery. These include information such as co-morbidities, socioeconomic status, length of hospital stay, and ethnicity [19].
The data linkage of HES-APC to the NJR will be performed as part of an existing agreement between the NJR and NHS England. Access to the data is facilitated under NJR permissions. The data linkage will be performed by applying seven linkage methods using a combination of matched NHS numbers, Local patient ID, date of birth, year of birth, gender, and/or matched care provider (Supplementary File 1). Procedures that are matched based any of those linkage methods will be included, but for a procedure to match, the HES-APC hospital episode start date must be the same or before the NJR operation date whilst the NJR operation date must be before or the same as the HES-APC episode end date.
Patient and public involvement
The Patient and Public Involvement and Engagement (PPIE) members at Wrightington hospital were consulted and involved in the study objectives and methodology.
Ethics
The NJR Research Committee approved this study [20]. The NJR supports public health surveillance and wider clinical decision-making and holds pseudonymised data that are anonymous to the researchers who use it. The NHS Health Research Authority tool guidance dictates that the secondary use of such data for research does not require approval by a research ethics committee [21]. Patients consented to inclusion in the NJR according to standard practice, with permission under the Health Service (Control of Patient Information) Regulations, otherwise referred to as Section 251 support [22].
Participants
All patients aged 18–100 years old with a primary TER on the NJR elbow dataset from the start of data collection on the 1st of April 2012 to the 31st of December 2022 will be included. Patients are excluded if they did not consent for their data to be used for research purposes, if it is impossible to trace them after surgery, if their ID numbers are invalid, or if the surgery was not performed in England. Several steps will be undertaken to confirm that the included cases are primary TER procedures. This will include (1) ensuring all operative patterns are consistent (2) confirming all reported procedures on the NJR matches the implant components submitted. Unconfirmed procedures and procedures with inconsistent operative patterns (i.e. a sequence of operations where the primary operation is not the first operation in the sequence or where there are multiple primary operations recorded for the same joint) will be excluded from the analyses. The data preparation process and exclusion of non-eligible procedures is summarised in Fig. 1. Eligible NJR procedures will then be linked to the available HES-APC data which includes all patients’ hospital episodes from 23rd of October 1996 to 31st of March 2022. Hospital episodes with invalid ID numbers, implausible dates, unknown discharges dates, and duplicate episodes will be excluded (Fig. 2).
Flow diagram for the unlinked National Joint Registry (NJR) elbow replacement dataset
Flow diagram for the linked National Joint Registry (NJR) and NHS England Hospital Episode Statistics-Admitted Patient Care (HES-APC)
Variables
Most of the data from the NJR and HES-APC can be extracted directly but some of the variables used in this study will be derived using the available data and the methods of how this will be performed are highlighted in this section.
Patient related variables
Patient related variables will be extracted from the NJR and HES-APC data. The list of the patient related variables and their data source is displayed in Table 1. Age will be derived from date of birth; socioeconomic status will be derived from postcode as an Index of Multiple Deprivation quintile (2015 version) [23]; and co-morbidities will be derived from the ICD-10 (international classification of diseases, 10th revision) codes reported by HES-APC. The list of co-morbidities for each patient will include all co-morbidities available in the HES-APC hospital episodes up to and including the primary TER. The co-morbidity list will be used to calculate the original Charlson Comorbidity Index [24] and a revised version of Charlson Index using an updated weights which are calibrated using English data due to differences in coding practice and hospital patient population characteristics [25]. If BMI is not reported, then the value will be derived from the height and weight variables if they are reported. The indication for surgery in will be reported as five categories: acute trauma, inflammatory, trauma sequalae, osteoarthritis and other. Procedures that are reported as Essex Lopresti on the NJR will be included in the acute trauma category if the injury is acute or in the trauma sequelae if the injury is chronic. The decision whether the Essex Lopresti injury is chronic or acute will be decided by the admission type and duration of surgery wait on the HES-APC data. Non-elective admissions and elective admissions with less than two weeks from decision for surgery to the date of surgery will be considered as acute injuries. Procedures reported as avascular necrosis will be added to the “other” category. Multiple indications for surgery can be selected on the MDS forms. In this study the likely primary indication will be selected based on a hierarchy agreed by the research team. Acute trauma will be the primary diagnosis if it is selected followed by trauma sequelae, inflammatory arthritis, osteoarthritis, and then “other” category. Ethnicity was categorised into six categories based on the Office of National Statistics (ONS) Ethnic group classification 6a [26]. If ethnicity is missing from the hospital episode admission for TER surgery, ethnicity will be established from other hospital episodes admissions for that patient where available.
Implant related variables
Implant related variables will be derived from the NJR data only. They are based on classifying each implant in the NJR data. In this study, the classification used for the NJR annual report will be used [27]. There are different combinations of the Latitude implant reported by the NJR 20th annual report. In this study Latitude will be classified as Latitude Legacy, Latitude EV, or Latitude mix (i.e., EV and Legacy components). The list of the implant related variables and how they will be measured is shown in Table 2.
Surgeon and hospital related factors
Most of the surgeon and hospital variables used in this study are not directly reported but will be derived from the available data (Table 3). The NJR includes pseudonymised codes to represent the surgeon who performed the surgery and the hospital where the surgery was performed. This data can be used to calculate the surgeon’s and hospital’s volume. In this project, the annual number of TERs will include all TERs from the 1st of January to the 31st of January of that year. The year 2012 will be excluded from this analysis because data is only available from April 2012.
The duration of elective wait and the length of post-operative stay will be derived from the HES-APC data. To establish the duration of elective wait, the difference in days between the date on which it was decided to admit the patient and the date of surgery will be calculated. The length of post-operative stay will be derived from the difference in days between the date of the TER and discharge date. Elective admissions will be classified into general admissions or day case admissions. Day case admission will be derived from the admission method and spell duration variables on HES-APC. For an elective procedure to be classified as day case it must be an elective admission and has a spell duration of 0 days.
Outcome
The primary outcome in this proposed study will be the number and/or the rate of provision of primary TER. Secondary outcomes will include the duration of elective wait and post-operative duration of stay measured in days. Current trends will be described by reporting the outcomes on annual basis.
Serious adverse events (SAE) within 30 days and 90 days from the index TER will also be reported. SAE will be defined as any severe medical complications leading to hospital admission, including pulmonary embolism, myocardial infarction, lower respiratory tract infection, acute kidney injury, urinary tract infection, cerebrovascular events, and all-cause death. SAE will be extracted from the HES-APC data and identified using ICD-10 codes.
Statistical analyses
Descriptive analysis will be performed for all included variables. Frequencies and proportion will be used to summarize categorical variables. The distribution of continuous variables will be assessed using histograms. It is likely that some of the continuous data, such as surgeons’ and hospitals’ volume, will be skewed, therefore, continuous variables will be reported using the median and interquartile range (IQR). The analyses will include summary of all the included population, stratified analysis for elective and acute trauma population, and analysis for each year from 2012 to 2022. The number of procedures performed by surgeons and hospitals will be summarised for the whole population and for each region in England and the results will also be reported on annual basis.
TER rates for different sexes, age categories, socioeconomic status categories and different ethnic groups will be reported. The rates of primary TER per 100,000 persons will be calculated by dividing the number of procedures in the NJR elbow dataset by the corresponding mid-year population estimates published by the Office for National Statistic (ONS). Sensitivity analysis will be performed using the census estimate from 2021. The population estimates by ethnic group reported by ONS will be used to estimate the rates of TER between different ethnicity groups. Age and sex standardised TER rates for each IMD group will be reported. Statistical analyses will be performed using Stata version 18 (StataCorp LP, USA).
Add Comment