Background
Increasing levels of obesity, population ageing, and growth in sports-related injuries are all anticipated to manifest in a greater future burden of osteoarthritis (OA) [
1‐
3]. For severe knee or hip OA, total knee (TKR) and hip replacement surgery (THR) have been shown to improve function and quality of life and be cost-effective [
4,
5]. The number of TKR and THR procedures has grown steadily over the last two decades in Australia and other developed countries [
6‐
9]. We recently reported a 105% increase in primary TKR utilisation over a 10-year period (2003–2013) in Australia and a 73% increase in THR surgery for OA over this time [
10,
11].
Demand for joint replacement surgery is expected to increase in many countries. In the United States (US), Kurtz et al. have predicted growth of 673% for TKR and 174% for THR from 2005 to 2030 [
12]. More recently, Inacio et al. have forecast growth in TKR volume in the US of 143 and 855% from 2012 to 2050 using conservative and exponential growth scenarios, respectively [
13]. Studies from the UK, Canada, Sweden, New Zealand and Denmark have also predicted increases in joint replacement surgery over the next two decades, although the estimates vary widely [
14‐
18]. In the US, demand for primary TKR and THR among younger age groups has been predicted to grow markedly, with over half of joint replacement recipients expected to be aged under 65 years by the year 2030 [
19]. Although TKR and THR projections have been recently published for Australia, neither age-specific nor sex-specific projections were reported [
20]. Rising obesity levels are a major driver of TKR rates, as demonstrated by national longitudinal data from the US [
21], and warrant particular consideration given the potential impacts on surgery utilisation. A recent large-scale study (
N = 105,189) also showed that obese individuals in Spain had an at least 2-fold increase (depending on obesity category) in the likelihood of TKR, compared to those of normal weight [
22]. Linked Norwegian Arthroplasty Register data (
N = 225,908) also support a link between weight gain and increased risk of TKR [
23]. Reducing the prevalence of obesity at the population level could have important benefits for healthcare systems with regard to fewer TKR procedures [
24], although this has not yet been evaluated.
An increasing burden of joint replacement surgery has significant cost and health workforce implications. The cost of a TKR or THR procedure in Australia is estimated at $AUD19,000 to $AUD30,000 per patient [
25], with over $AUD1.2 billion spent annually in Australia on OA-related hospital admissions [
26]. There are even greater economic implications relating to surgery for younger patients, given the higher risk of multiple revisions [
27]. From a health workforce perspective, over one-third of active orthopaedic surgeons in Australia are aged 55 or older and likely to retire within the next 10–15 years [
28]. This would undoubtedly impact Australia’s capacity for future provision of joint replacement surgery. An improved understanding of the future national burden of joint replacement in both the public and private health systems is required to ensure that demand can be met and high quality standards can continue. Australia has maintained a validated national joint replacement registry since 2002 (with over 98% coverage of all THR and TKR surgeries performed in public and private hospitals) [
9], and these population-level data offer a unique opportunity to generate well-informed projections of national burden.
This study aimed to forecast the number of primary TKR and THR surgeries likely to be performed for OA in Australia to the year 2030 (including age- and sex-specific estimates), and associated costs. It also aimed to model the impact of two contrasting obesity scenarios on future TKR burden.
Discussion
This study has produced comprehensive estimates of the future burden of TKR and THR for OA in Australia (by age, sex and overall, as well as by healthcare sector), using a well-validated national dataset that includes all joint replacement procedures performed in this country. According to our projections, Australia faces a potentially unsustainable joint replacement burden by 2030 which requires significant investment in public and private health systems and health workforce training. Based on growth in surgery rates over a decade, TKR procedures for OA are expected to increase by 276% (from 42,920 procedures in 2013 to 161,231 in 2030) while THR procedures for OA are predicted to rise by 208% (from 25,945 procedures in 2013 to 79,795 in 2030). The total cost to the health system is forecast to exceed $5.32 billion in 2030. These estimates can be used to facilitate healthcare resource planning and inform health policymakers and public health practitioners about future national demand for joint replacement.
Much of the predicted growth in TKR and THR is driven by population ageing. From 2003 to 2013, the Australian population aged 40 years and over increased from 8.7 million (representing 51% of the total population) to 10.7 million (representing 53.1% of the total population). By 2030, this age group is expected to number 14.8 million, or 55.8% of the total population. Meeting the large growth in surgical demand will prove challenging for Australia, given the ageing surgical workforce [
28]. Although joint replacement should be reserved for severe, end-stage OA, if individuals choose to undergo joint replacement for milder symptoms [
38] this could further augment demand for surgery. We observed an initial sharp increase in TKR procedures from 2003 which appeared to level off by 2013, particularly for the 40–69 age group. This trend was also identified in our state-level analyses [
39], and may relate to ‘catch up’ of previous unmet need for surgery following the introduction of financial incentives designed to improve the uptake of private health insurance. Our projections do not take into account the potential impact of rising rates of lower limb sports-related injuries [
3], and this has specific relevance for TKR as injury has been identified as a major contributor to knee OA [
40]. Further epidemiological data are needed to quantify the increased risk of TKR and THR associated with sports injury, and this research is currently underway.
Although a range of international studies has reported projections of joint replacement burden, comparisons are difficult given differing samples, methodological approaches and timeframes for analysis. Our methods differ substantially to those used for a recent study that projected growth in TKR and THR rates in Australia from 2014 to 2046 [
20]. We limited our data inputs to procedures performed for OA (11% of THR procedures in Australia are performed for non-OA diagnoses [
41] including fractured neck of femur) and we calculated age- and sex-specific rates to examine subgroup trends in joint replacement growth. Despite methodological differences, our 2030 Scenario 2 projections for TKR and THR fall clearly within the 95% prediction intervals reported by Inacio et al. for that year [
20]. Variability in projections between countries could relate to differences in obesity rates, data accuracy, and health system differences, particularly for countries with mixed public-private systems. The study by Kurtz et al. [
12] used discharge records from the US Nationwide Inpatient Sample, representing about 20% of all community hospitals. Based on surgery trends from 1990 to 2003, the researchers predicted a 673% increase in TKR and a 174% increase in THR from 2005 to 2030. The potential impact of changing obesity rates was not examined. Using a general practice database covering approximately 10% of the total population, Culliford et al. estimated a 26% increase in TKR and also for THR in the UK from 2015 to 2030 (assuming that 2010 surgery rates remained constant) [
17]. Using similar methods (assuming the 2013 surgery rate remained constant), we projected a 53% growth for TKR and also for THR in Australia, based on population growth. The differences in our static-rate projections may partly relate to differences in projected growth (and population structure) for the Australian and UK populations over time [
30,
42]. The UK study also predicted that TKR would increase by a further 7% if BMI proportions continue to increase over time. Neither the US nor UK studies was limited to patients with a primary diagnosis of OA. Denmark and Sweden have also predicted increased demand for THR by 2020 and 2030, respectively [
14,
16], while projections of TKR and THR burden in New Zealand to 2026 have also been published [
15]. While this paper focuses on the cost burden associated with joint replacement surgery, the benefits from surgery are substantial, as highlighted by numerous studies assessing the cost-effectiveness of THR and TKR procedures [
4,
43,
44]. Most recently, Elmallah et al. showed that THR and TKR were associated with lifetime quality-adjusted life year (QALY) gains of 2.07 and 1.85, respectively [
44]. Cost-effectiveness was demonstrated by an incremental cost-effectiveness ratio of approximately $US39,000 per QALY for THR and approximately $US43,000 per QALY for TKR [
44], which falls well below the arbitrary cost-effectiveness threshold of $US50,000 to $US150,000 [
45]. Using a discrete-event simulation model, Higashi et al. estimated that population health gains from joint replacement in Australia were equivalent to 115,000 disability-adjusted life years (DALYs) averted for THR and 113,000 DALYs averted for TKR [
46]. These figures highlight the value of joint replacement procedures at the population level.
Quantifying the potential impact of reducing obesity on the projected healthcare costs of TKR provides a strong policy and public health argument for supporting population-level weight loss campaigns and individual-level interventions. ABS data show that the proportion of Australians who are overweight or obese increased from 61.2% in 2007–2008 to 62.8% in 2011–2012 [
37]. We estimated this would exceed 70% by 2030, resulting in a further 15% growth in TKR procedures. TKR surgery for people who are obese can produce substantial improvements in pain and function (comparable in magnitude to improvements experienced by people in the normal weight range [
47]) and may indeed be clinically warranted. However, there are increased surgery and episode of care costs for obese patients undergoing this procedure [
48]. It is also possible that as obesity rates increase over time, TKR costs may accelerate faster than monetary inflation, with significant health budget implications. It is unknown whether a reduction of overweight and obesity by 5% at the population level is achievable, so we modelled the benefits on a sliding scale at varying degrees of weight loss. However, a US randomised controlled trial involving 454 overweight and obese older participants with radiographic knee OA found that the combined diet/exercise intervention group had a mean weight loss of 11.4% of body mass, and the diet-only intervention group reported a mean weight loss of 9.5% of body mass over the 18-month study [
49]. If such dramatic weight loss results could be achieved in ‘real-world’ settings, then a large proportion of overweight people could be transitioned into a normal BMI category. In the present study we did not explore the potential impact of changes in obesity levels on THR, as no consistent association between BMI and risk of hip OA or THR has been demonstrated [
50‐
52].
A key strength of this study was our use of national registry data collected over a 10-year period to project the future incidence of TKR and THR for OA. Given external data validation processes, we are confident the AOANJRR dataset provides a comprehensive picture of joint replacement utilisation in Australia. We also used two different scenarios for projecting future burden, which helps to quantify potential uncertainty in our estimates. However, we acknowledge several limitations to our methods. As patient-level data on BMI were not available, we assumed that overweight and obesity rates matched Australian population rates. As people undergoing TKR may have higher rates of overweight and obesity [
53], this is a conservative assumption. We also assumed linear growth in obesity rates over time, consistent with previous methods [
54]. Administrative data (average costs per TKR or THR admission) were used to impute costs for the index procedure and we did not have access to patient-level costing data relating to post-operative complications or revision surgery. We also conservatively assumed that bilateral procedures were performed simultaneously and recognise that staged procedures are more costly. However, as relatively few bilateral procedures are performed (representing 16.3% of all TKRs and 6.5% of all THRs in 2013) we do not consider this to be problematic. Our study focused on direct costs to the health system and did not include personal (eg out-of-pocket healthcare costs) and societal costs (eg lost productivity and carer time), which are likely to be substantial. Finally, it is possible that the development of new medical interventions for OA could allay some of the future burden of joint replacement although this would be unlikely over the study forecast period, and that initiatives to reduce unit cost per joint replacement episode (for example, lower implant costs, use of day surgery, and home-based rehabilitation) could also reduce the future economic burden but were not the focus of this research.