One in three adults over 65 fall annually, with one in five remaining on the floor for greater than one hour, which is referred to as a long lie. Pressure on the National Health Service has resulted in extended stays in the Emergency Department (ED) (sometimes longer than 12 hours) and prolonged ambulance response times. This impacts the older adults who have fallen and remain on the floor.

This project aims to develop a risk prediction model (RPM) for use within the ED to understand which older adults (60 years or older) who fall over and remain on the floor for longer than one hour ("long lie") and develop rhabdomyolysis (a serious condition where muscle breaks down and releases substances into the blood that can damage the kidneys) will develop poor outcomes and need admission to hospital for treatment and which patients can be safely discharged home.

Aim:

To develop a RPM to identify which older adults who have a fall and a long lie and attend the ED develop poor outcomes such as Acute kidney Injury (AKI) [kidneys suddenly stop working properly], needing kidney replacement therapy (KRT) [a treatment that helps kidneys that aren't working properly do their job of cleaning the blood] and mortality [death].

Objectives:

1. Abstract patient level data (e.g. biochemical, demographic, situational, medical history, medication history) from medical records combined with outcomes to understand which variables lead to poor outcomes such as AKI, needing KRT and mortality.
2. Analyse the data using a statistical package (Statistical Package for Social Sciences [SPSS]) to develop a RPM with good discriminative abilities [how well the score can tell high-risk from low-risk patients].
3. Demonstrate the ability of the RPM to identify which patients need admission to hospital with treatment and which patients can be safely discharged home.

There is no RPM in existence to understand which older adult patients who fall and develop rhabdomyolysis secondary to a "long lie" will have a poor outcome (AKI, KRT, death). Therefore, it will be beneficial to understand which patients need admission for in-patient care to reduce sequelae [bad outcomes] and which patients can avoid hospital associated issues (e.g. deconditioning, infections) and be safely discharged home. This will be achieved with a retrospective longitudinal cohort study, specifically a retrospective chart review where electronic medical/health records will be reviewed by one researcher (the Principal Investigator [PI]) to abstract key data to identify the risk factors associated with poor outcomes after having a fall and a long lie. The medical/health records are what the PI would use in their normal clinical and non-clinical role when working for the NHS Trust. The main electronic patient records that will be utilised is the 'Acute Care Portal' which has emergency department, prehospital ambulance data, community/primary care and in patient speciality teams/departments records available. This is likely where the majority of the data required will be abstracted from but other clinical systems such as 'SystmOne' may be utilised to check if a patient died since being back in the community (outside the acute hospital Trust). Additionally, the medical/health records for patients who attend the critical care unit (CCU) are within a system called 'Metavision', this will be reviewed for patients that attended the CCU.

The data will be input into a data abstraction tool (DAT) which is an Excel Spreadsheet (XLS). This will include a study ID (a number ascending from 1) and not patient identifiable information such as the NHS number. This ensures that the DAT will be a pseudonymised dataset. This will be saved on the University of Salford (UoS) password protected and two factor authenticated OneDrive of the PI. The participants will be identified through a dataset produced by the NHS laboratory data team which shall be emailed to the NHSMail account of the PI. This will be provided to the NHS Trust Healthcare Analytics team who will upload the dataset to the NHS England Digital Message Exchange for Social Care and Health (MESH) portal to assess for any patients who has opted out using the National Data Opt Out (NDOO). Once this is returned to the PI via NHSMail this will act as the data linkage file (DLF). The pseudonymised dataset can be viewed (but not edited) by the three co-investigators (research supervisors) via UoS OneDrive. The DLF can only be accessed by the PI via the password protected, two factor authenticated NHSMail OneDrive.

Once the study has completed (defined as when the study is published in a peer reviewed journal) the DLF will be permanently deleted by the NHS IT team at which point the pseudonymised DAT dataset becomes anonymised as it can no longer be linked back to individual patients. The anonymised dataset will be changed to a comma-separated values (CSV) format for data preservation and uploaded to the UoS Research Repository, Figshare for 10 years. After the 10 years it will be permanently deleted by the UoS IT Team. Once the anonymised dataset is uploaded to Figshare the version on UoS OneDrive will be permanently deleted by the UoS IT team.

A waiver of consent is being requested to the Confidentiality Advisory Group (CAG) to minimise any harm/distress to the participants or their families. This is because the data is routinely collected and will not negatively affect the participants through the access of routinely collected data being utilised for another purpose (research). Additionally, some of the participants will have subsequently died or may have cognitive issues which are important participants to include so the research is representative of the real world. Excluding these vulnerable groups or avoiding recruiting these would not represent the real-world issues and could disadvantage future patients with similar conditions. Participants that have opted out of their data being used for research purposes will not be included but other participants will be included to ensure that future patients can be managed appropriately, safely and based on contemporaneous evidence.

Direct risk to patients is minimised by seeking CAG approval opposed to directly seeking consent from every individual patient/family member. Indirect harm from data breaches will be minimised by having a robust data management plan (DMP) in place. This clearly sates how the DAT and DLF will be stored, on what platform and how it will be shared as previously stated.

This study will derive the RPM and internally validate it but will not externally validate it. If the RPM has good discriminative abilities and clinical utility it is likely that the RPM will be externally validated in future studies by the PI. This may be in other EDs or within community response teams or ambulance services as appropriate. The participants will be representative of the local population from which they are drawn it is hoped given that the inclusion period is roughly five years and participants are not being excluded based on ethnicity or socio-economic status. Data on ethnicity will be collected as this may be a risk factor for poor outcomes so the final data can be reviewed to see if it is representative of the local population. A future research study would need to validate the RPM in the real world and assess its safety in clinical practice.

Once the study has been completed it will be disseminated via conferences and peer reviewed journals where possible regardless of the results. If the RPM has good discriminative abilities, it will be useful so future research can be built upon this but equally if the RPM does not reach such significance it is important to share this so future research studies do not repeat this wasting time and research funding.

The outcomes that will be reviewed are AKI, KRT or death. Participants that have an AKI can be clearly identified from the laboratory data regarding the creatinine levels and if available the clinical notes where urine output is recorded. Similarly, if a patient receives KRT this automatically classifies the participant as AKI stage 3 according to the Kidney Disease: Improving Global Outcomes (KDIGO) AKI guidelines. The majority of the data will be available within one electronic patient record system, Acute Care Portal (ACP). Patients that are admitted to the CCU have their clinical notes recorded on Metavision which will need to be reviewed in order to determine if the patient had certain interventions such as KRT. Finally, to determine if the participant died within 30 days of the admission (defined as 30 days from the attendance to the ED) community/primary care records may need to be reviewed on a platform called SystmOne. If the patient died while in the hospital this will be noted within ACP.

The proposed statistical analysis is a multivariable logistic regression. This is where a number of different variables (e.g. age, gender, creatine kinase levels and time on floor) are reviewed to see if they increase or decrease the odds of a poor outcome (AKI, KRT or death). Other statistical techniques will be used to see the number of events recorded (e.g. 21% of the participants had an AKI) or average participant data (mean age, ethnicities etc). Missing data will be accounted for using multiple imputation with chained equations (MICE) and the threshold for including the variable will be set at 20% (meaning that if >20% of data for a variable is missing it will be excluded from analysis). Internal validity will be assessed using cross validation once the data has been analysed and RPM developed following face validity and clinical utility of an expert advisory group. The RPM will then be evaluated on data over the preceding data collection period (likely 6-12 months).

Inclusion Criteria:

The participants need to fulfil all three parts of the inclusion criteria:

1. 60 years old or older at time of ED presentation
2. Developed rhabdomyolysis (defined as CK >999U/L)
3. Have had a fall and been / suspected to have been on the floor / immobilised in one position for > 59 minutes.

Exclusion Criteria:

1. Patient opted out of research studies via the National Data Opt Out (NDOO) Service.
2. Principal Investigator (PI) had clinical input into patient care.
3. Other causes of elevated creatine kinase (e.g. seizures, acute coronary syndromes, burns, myositis, muscular dystrophy, cardiac arrest)

NHS Emergency Department in England

A fully anonymised dataset will be made available for other researchers via the University of Salford Research Repository (Figshare)

Likely available September 2027 - stored for 10 years, until Sept 2037

It will be anonymised and is open access to use for research purposes