Predicting the risk of needing a total knee arthroplasty.

The COVID-19 pandemic has disrupted health-care services across the world and has resulted in waiting lists for total knee arthroplasty that are unprecedented in the UK National Health Service (NHS). Clinicians have suggested increasing knee arthroplasty productivity in the NHS by 10–20% to address the burden over the coming decade. An alternative or additional strategy would be to address potential reversible factors associated with the need for total knee arthroplasty and therefore reduce the demand for surgery. Even if such a strategy enabled surgery to be delayed by several years, this could also reduce the potential revision burden of total knee arthroplasty, with increasing age being associated with lower revision risk.

In The Lancet Rheumatology, Qiang Liu and colleagues present predictive models for the need for total knee arthroplasty. Unlike in previous models, the authors included radiographic severity as a predictive factor. Despite radiograph results being an obvious predictor of progression to total knee arthroplasty, with worsening radiographic severity of osteoarthritis being associated with increasing risk of total knee arthroplasty, adjusting for this confounding factor in modelling could help to identify other reversible risk factors that might have been overlooked previously. Liu and colleagues used data from the Multicentre Osteoarthritis Study (MOST) to develop two prediction models: with and without radiographic severity. MOST is a prospective observation dataset that was established to identify factors that affect the occurrence and progression of knee osteoarthritis in American adults aged 50–79 years over a 7-year follow-up period. Liu and colleagues then used data from the Osteoarthritis Initiative to validate their prediction models.

The only reversible variable identified to be associated with increased risk of total knee arthroplasty was body-mass index (BMI), with a hazard ratio (HR) of 1·31 (95% CI 1·18–1·46) for each increase in BMI point. For example, an increase in BMI of 4 points was associated with more than twice the risk of needing a total knee arthroplasty. The rest of the variables associated with risk of total knee arthroplasty were either not reversible (age, sex, and race) or were potentially associated with, but were not directly related to, progression of disease (pain, analgesia use, and glucosamine use). However, when radiographic severity of the osteoarthritis was accounted for in the modelling, BMI was no longer a significant factor associated with risk of total knee arthroplasty. One factor that remained significant in both models, which might or might not be reversible, was knee arthroscopy, which was independently associated with total knee arthroplasty (HR 1·26, 95% CI 1·15–1·38). This led the authors to conclude that it is “crucial to follow the guidelines for osteoarthritis management regarding knee arthroscopy”. Although there is a move away from knee arthroscopy in patients with degenerative knee arthritis, patients with less severe radiographic osteoarthritis with mechanical symptoms might still benefit from such an intervention and this could have even helped defer total knee arthroplasty for several years.

The most predictive factors associated with increased risk of total knee arthroplasty were older age, pain, and radiographic severity of osteoarthritis. To most physicians and surgeons managing such patients, these criteria will not come as any surprise and are in line with most surgeons' criteria to go forward with a total knee arthroplasty. However, if a patient's pain was adequately controlled from alternative measures and other ways of dealing with the pain, such as cushioned insoles or the Escape Pain programme, their need for a total knee arthroplasty might be delayed. Furthermore, one reason why BMI might have become a non-significant factor when radiographic severity was included in the model could be that total knee arthroplasty is needed at an earlier age with increasing BMI. There is growing evidence that increasing BMI is associated with earlier need for total knee arthroplasty, and patients with early symptoms and radiographic signs of osteoarthritis should be encouraged and supported to lose weight. This could not only delay the need for total knee arthroplasty but could also decrease the perioperative risks, such as deep infection.

Prediction models are likely to be the future of medicine, with artificial intelligence predicting the best course of management to delay total knee arthroplasty, or even the optimal time for surgery. Ideally, this technology will be able to provide patient-specific predicted outcomes and potential risks of total knee arthroplasty. However, such advancements might not be needed in view of the increasing knowledge of osteoarthritis and the molecular mechanisms behind the development of the disease, and by the time such prediction or artificial intelligence tools exist, it might even be possible to avert the need for a total knee arthroplasty entirely.

I declare no competing interests.