PRediction of Outcome With ED Intervention for Colles Type Wrist Fractures [PREDICT]

Overview

The Colles fracture, a fracture of the wrist, is frequently associated with deformity of the broken part of the bone, particularly in older or frail patients. This can cause long term wrist deformity and problems using the wrist and hand if not corrected. Manipulation under anaesthesia (MUA) is often undertaken in the emergency department (ED) for 'displaced' fractures in an attempt to correct the deformity. The procedure involves a local anaesthetic technique, additional staff, and takes some time to complete. The procedure is not without risk; it can cause bruising, skin tears, complications from the local anaesthetic and can be uncomfortable. Although fracture positions are usually improved by ED manipulation initially; these fractures can slip back to an unacceptable position over the next 1-2 weeks, despite plaster cast immobilisation. For fractures that slip, open surgery is usually required to correct and hold the fracture with metal plates or wires. This is performed in the operating theatre and requires another visit to the hospital. Preliminary work suggests this affects over a quarter of patients, undergoing ED MUA for Colles' wrist fractures. If it were possible to reliably identify patients whose fractures were likely to slip and require open surgery despite ED manipulation, unnecessary procedures and visits to hospital could be avoided. This would ensure patients got the right treatment first time and save patients and the NHS time and money. There are a number of factors that might affect the likelihood of fracture instability and need for surgery. These include patient factors such as age, functional status and presence of osteoporosis (thin weak bones) and the specific position of the fracture. In this study we will be measuring the fracture positions on x-rays of patients with a Colles' fracture to see if we can accurately predict ED MUA failure on the initial x-ray.

Many thousands of patients attend emergency departments (ED) with displaced (deformed) wrist fractures and undergo closed manipulation under local anaesthetic (MUA) in the ED. Local audit suggests between 25-40% of these manipulated fractures slip back into an unacceptable position (unstable fractures) and require a subsequent open operation to reduce and pin or plate the bone (ORIF). There is some evidence that the degree of initial deformity and other factors might predict instability but this is neither well established nor well researched. Reliably predicting which fractures are likely to fail and which are likely to succeed closed ED manipulation could reduce the proportion undergoing a futile ED procedure, save patients' time and visits to hospital and ensure definitive care the first time round. We will identify cases from a previously conducted clinical activity analysis of patients who had undergone MUA for Colles' fracture over a two year period at Royal Devon and Exeter Hospital ED. These cases were retrospectively identified (from electronic attendance logs) and followed up until fracture healing or surgery. In this database, the need for subsequent surgery was recorded from electronic records and review of comments in fracture clinic. X-rays of these patients have been identified and anonymised by a member of radiology staff, who is not a member of the research team. These X-rays and the database cases have been given a unique trial number and all identifiable data removed before trial use. The database will be used to provide outcome data for the development of a model to see if x-ray features can reliably predict ED MUA failure. We will exclude Smiths' (anteriorly displaced) fractures and those whom there is no follow up data in the database eg. Patients who were followed up elsewhere. Anonymised x-rays will be used to develop computer-assisted methods to measure key fracture angles as well as other factors such as degree and position of comminution (fragmentation) and markers of osteoporosis. All x-rays will then by analysed using the system to record these measurements. Depending upon the speed of development, any user facing systems could also be tested on staff to gauge usability and test for inter-rater agreement. Multivariate analysis of these x-ray features measured by computer, together with the database demographic and outcome data will be used to develop the best model /decision rule to predict likelihood of MUA success. A 'leave one out cross validation' technique will be used to validate such a model and assess whether we could have reduced the ED MUA failure rate without increasing the overall number of open operative procedures. This work will be used to inform a larger prospective study where any decision rules can be assessed and further factors explored with the prospect of adding these to any final computer model.