Optimising the current model of care for knee osteoarthritis with the implementation of guideline recommended non-surgical treatments: a model-based health economic evaluation

DOI: https://doi.org/https://doi.org/10.57187/smw.2023.40059

Introduction

Knee osteoarthritis is a painful condition that leads to loss of function and is one of the most common reasons for disability worldwide [1]. With an estimated overall prevalence of 8.5% and an attribution to 5,150 disability-adjusted life years in 2019 in Switzerland, knee osteoarthritis places a substantial burden on health and the healthcare system [2]. The economic burden of knee osteoarthritis on the Swiss statutory healthcare system was estimated at 1.12 billion CHF in 2011 [3]. The association of osteoarthritis with reduced physical activity levels may promote frailty and/or cardiovascular disease in the aging population, which negatively affects overall health [4].

Current treatment guidelines recommend non-surgical and surgical treatments for patients suffering from knee osteoarthritis [5, 6]. When non-surgical treatment fails to achieve sufficient symptom relief, total knee replacement (TKR) is endorsed [7]. Even though total knee replacement was shown to be highly effective in patients with moderate to severe knee osteoarthritis [8], up to 30% of the operated patients experienced chronic pain after the procedure [9]. Additionally, total knee replacement is associated with adverse events and implants have a limited lifespan leading to implant failure and subsequent revision surgery, which is associated with high costs and less favourable outcomes [10]. The younger the patient who receives a total knee replacement, the higher the risk of revision surgery is in his/her lifetime [11].

Guideline recommended non-surgical treatments are exercise, education about self-management strategies, weight reduction if indicated, painkiller prescription and insoles [5]. Albeit recommended, uptake of guideline recommended non-surgical treatments remains low in clinical practice. Less than 40% of knee osteoarthritis patients receive appropriate non-surgical care in the U.S. [12]. It is currently unknown what proportion of knee osteoarthritis patients receive guideline recommended non-surgical treatments in Switzerland. As of 2019, physiotherapists can be certified as GLA:D^© therapists, a certification course that trains therapists in the delivery of guideline recommended non-surgical treatments [13]. By the end of 2022, around 700 therapists had completed this course. With more than 11,000 licensed physiotherapists in Switzerland [14], this is a small proportion and the need for such a certification course may be seen as evidence that the uptake of guideline recommended non-surgical treatments is still low in Switzerland.

A recent systematic review by Mazzei et al. [15] including 23 health economic evaluations concluded that guideline recommended non-surgical treatments are cost-effective and implementation should be encouraged. The authors commented that cost-effectiveness, from a healthcare perspective, may be underestimated if guideline recommended non-surgical treatments delay total knee replacement. Even though there is some evidence that guideline recommended non-surgical treatments can delay total knee replacement in eligible patients [16, 17], more research is needed to draw any firm conclusions. Yet, it seems plausible that the delay of total knee replacement through guideline recommended non-surgical treatments is possible and the cost-effectiveness may be even greater when analysed in the context of total knee replacement delay as this likely reduces the amount of revision surgeries needed [11].

Considering the potential effect of guideline recommended non-surgical treatments in delaying total knee replacement and the insufficient implementation of guideline recommended non-surgical treatments in the current model of care for knee osteoarthritis in Switzerland, the aim of this health economic evaluation was to answer the following research questions:

• Is a model of care optimised by the implementation of guideline recommended non-surgical treatments cost-effective from a Swiss statutory healthcare perspective?
• What is the effect of total knee replacement delay by guideline recommended non-surgical treatments on the cost-effectiveness?

Methods

Overview and model

We report this health economic evaluation according to the Consolidated Health Economic Evaluation Reporting Standards [18] and have constructed a Markov model following the recommendations of the International Society for Pharmacoeconomic and Outcome Research [19]. Effectiveness (expressed as quality-adjusted life years [QALYs]), treatment costs, and transition probabilities have been determined for two model of care consisting of surgical and non-surgical treatments for knee osteoarthritis: A) the current model of care in which non-surgical treatment primarily consists of painkillers and written advice and B) an optimised model of care in which guideline recommended non-surgical treatments are implemented.

We coded a state-transition Markov model in RStudio (Version 1.3.1093) [20]. Data manipulation and plots were realised using the “tidyverse” package (Version 1.3.0) [21]. The model represents the various states for patients with knee osteoarthritis in the healthcare system. A major advantage of using Markov modelling is that it encompasses the whole lifetime of each patient allowing us to implement delays with so called “tunnel states” [22]. The structure of the developed model is based on previously published studies [23–25].

Figure 1 depicts a detailed description of the model. We conducted the analysis from the statutory Swiss healthcare perspective. Outcomes were reported as incremental cost-effectiveness ratio which describes the additional costs per quality-adjusted life year gained. The cycle time was defined as one year with a time horizon of 70 years which is considered lifelong.

Figure 1 Conceptualisation of the proposed Markov model visualising the current model of care and the proposed optimised model of care. The difference between current model of care and the optimised model of care is the addition of more extensive guideline recommended non-surgical treatments. By evaluating the cost-effectiveness of the whole model of care the dynamics between the surgical and non-surgical treatments can be investigated. In our scenario analysis, we analysed the effect a delay of total knee replacement (TKR) by guideline recommended non-surgical treatments by two and five years on the cost-effectiveness of the whole model of care. The delay of TKR is modelled by adding tunnel states after “Successful non-surgical care”. For Switzerland a “willingness to pay” threshold of 100,000 CHF was considered.

CHF: Swiss francs; KOA: knee osteoarthritis

Based on a health economic evaluation conducted in the context of Switzerland [26], we decided to apply a discount rate of 3% to costs and health-utilities gained starting from the second year and a willingness to pay threshold of CHF 100,000.

No study protocol was registered or published.

Patient population

The population of interest were individuals at risk of developing symptomatic radiographic-confirmed knee osteoarthritis living in Switzerland aged 40 years in 2019. We chose this starting age because the prevalence of knee osteoarthritis in patients younger than 40 is very low [27]. Absolute population numbers were derived from the Swiss Federal Statistics database from 2019 [28].

Cost data

Because the statutory healthcare perspective was chosen for this health economic evaluation, only direct costs were considered while indirect costs and intangible costs were omitted. Cost data were calculated using the following formula: Cost = unit consumed x unit price (i.e., non-aggregated form). Costs were expressed in 2019 Swiss francs.

Non-surgical treatments for knee osteoarthritis are delivered in an outpatient setting. Outpatient costs are listed by the physician outpatient services catalogue, which is used for the reimbursement of individual statutory healthcare services in Switzerland [29]. Because of Swiss privacy legislation, it is not possible to link individual healthcare services to a diagnosis. Nevertheless, to estimate the costs attributed to outpatient treatments we contacted various clinical experts (orthopaedic surgeons, physicians, physiotherapists). The costs accumulated during the time individuals were successfully treated with non-surgical measures, reflected by the “Successful non-surgical treatment” state, were based on expert opinion.

Surgical treatments (total knee replacement and revision surgery) are delivered in an inpatient setting. Unlike outpatient treatments, inpatient reimbursement is regulated by “diagnosis related groups” (DRG). DRGs are classifications of specific case groups based on diagnoses and treatments. Reimbursements are paid as a flat rate depending on diagnosis, of which publicly accessible databases are available [30]. To separate costs related to total knee replacement versus revision surgery we used insurance claims provided by two health insurance providers, which together hold a market share of 14.4% (i.e., 1,226,535 individuals in Switzerland in 2019 [31]). DRG code I43C was used to calculate costs related to total knee replacement while revision surgery costs were retrieved through DRG codes I04A, I04B, I43A and I43B. We made assumptions regarding the cost incurring after total knee replacement and revision surgery, reflected by “Successful TKR” and “Successful revision” states. Table 1 provides an overview of the costs used in the model. A detailed description of the cost estimation can be found in Appendix 1.

Utility data

To represent utilities for successful non-surgical treatment, we used data from a randomised controlled trial (RCT) that reported data calculated for Denmark [32]. We decided to use data from this trial because it resembled the treatment regimens used in our models and Denmark, like Switzerland, is a western high-income country. In this trial, patients (n = 100) with symptomatic knee osteoarthritis and radiographic confirmed knee OA (Kellgren&Lawrence score ≥1) not eligible for total knee replacement were either randomised to current non-surgical care (painkiller prescription, written/oral advice) or guideline recommended non-surgical treatments (exercise, dietary intervention, insoles, painkillers, education classes). The proportion of women was 51% and mean age of the total sample was 66 years (SD 8.9 years). During the follow-up of 12 months, health utilities were assessed using the EuroQol 5-Dimensions 5-Levels (EQ5D-5L) questionnaire for both groups. At the end of the follow-up, health utility gains of the guideline recommended non-surgical treatments-group were higher (+0.125) compared to the usual care group (+0.091). We used these values for the “Successful non-surgical treatment” state for the current model of care and the optimised model of care, respectively, in our model.

For total knee replacement and revision surgery, another RCT from the same research group was used to extract health utility data for successful total knee replacement [16]. In this trial, patients (n = 100) eligible for total knee replacement (pain and Kellgren & Lawrence score ≥2) were either randomised to immediate total knee replacement or a period of guideline recommended non-surgical treatments with the option of total knee replacement at any time. The proportion of women was 62% and mean age of the total sample was 66.4 years (SD = 8.7). Health utilities were reported for both groups after a follow-up period of 24 months using the EQ5D-5L. Quality-adjusted life year values of the total knee replacement group after 12 months were extracted and imputed in the model as health utility value of the “Successful TKR” state.

The health utilities (using EQ5D-5L) for “Revision surgery” were derived from a study that gathered follow-up data of patients (n = 5,398) receiving revision surgery in a hospital in New York, USA [33]. During the year of surgery, we assumed patients to have baseline utilities as mobility is usually severely restricted during the first months after surgery. Table 2 depicts utility values used in the model.

Transition probabilities

The first state in our model accounts for all males and females living in Switzerland aged 40 years in 2019 without symptomatic knee osteoarthritis. Pooled global incidence for symptomatic radiographically confirmed knee osteoarthritis was reported as 0.02% per year in individuals aged over 20 years [27]. This incidence was used for all ages and for males and females. Using the incidence, the annual number of individuals making the transition from the general population to non-surgical treatment because of the development of knee osteoarthritis was calculated. It was assumed that persons with symptomatic knee osteoarthritis seek help from their physician and receive non-surgical care. The same transition probability for non-surgical treatment was used for current model of care and optimised model of care.

We calculated age-specific probabilities for the transition from the “Successful non-surgical treatment” state towards the “TKR surgery” state as follows:

(1) Define the number of persons living in Switzerland in 2019 by age group and gender.

(2) Establish the prevalence of symptomatic radiographically confirmed knee osteoarthritis by age group and gender for western Europe using data from the Global Burden of Disease Study [1].

(3) Estimate the number of persons living with symptomatic radiographically confirmed knee osteoarthritis by age group and gender by multiplication of population at risk with prevalence.

(4) Assess the number of total knee replacement in Switzerland in 2019 by age group and gender [34].

(5) Assess the probability for undergoing total knee replacement by dividing the number of TKR by estimated persons having symptomatic radiographically confirmed knee osteoarthritis:

number of total knee replacements 2019 / population 2019 × prevalence knee osteoarthritis

Steps 1 to 5 described above allowed us to calculate the age and gender specific proportion of individuals with symptomatic knee osteoarthritis who received total knee replacement in Switzerland in 2019. These proportions were then used as model input. The obtained transition probabilities are in line with incidence rates for total knee replacements published elsewhere [35]. See Appendix 2 for more detail.

We used preliminary data from the “Swiss National Hip and Knee Surgery” database (SIRIS) [34] to calculate age-specific annual probabilities for the transition from “Successful TKR” to “Revision surgery”. Currently, cumulative revision risks for a follow-up period of 8 years are available, which we used for our calculations. The same annual revision probability was used for optimised model of care and current model of care model calculations. See Appendix 3 for more detail.

For the transition probability from any state to death, crude age-specific mortality rates in Switzerland for 2019 were calculated using population data from the Federal Statistical Office [28] according to the following formula:

deaths in age group (2019) / persons in age group (2019)

See Appendix 1 for more detail.

Table 3 provides an overview of all transition probabilities used in the model.

Scenario analysis

In this health economic evaluation three different scenarios were defined. Scenario one is described as “base case”. In this scenario the incremental cost-effectiveness ratio for optimised model of care vs current model of care without delay of total knee replacement has been calculated. The second and third scenario were calculated under the assumption that guideline recommended non-surgical treatments can delay total knee replacement by two or by five years, respectively.

Sensitivity analysis

We evaluated model robustness by probabilistic sensitivity analysis (PSA) with 10,000 iterations on cost and health utility parameters. We used beta distribution for health utilities and transition probabilities because the imputed values were not close to zero and calculated by the method of moments [36]. For cost data we used gamma distributions, as recommended by guidelines [36]. PSA was conducted for all three proposed scenarios.

The developed model has been validated technically and clinically. To ensure technical validity, two differently coded models within RStudio were used to compare the outcomes and the coded model was checked for errors by the last author. Clinical validation was performed by a primary care physician, an orthopaedic surgeon and a physiotherapist specialised in knee osteoarthritis treatments.

Results

Calculations from the model for the current model of care resulted in CHF 4,465 cost per person and 16.92 quality-adjusted life years per person over a period of 70 years. Calculations for optimised model of care with no delay of total knee replacement resulted in CHF 4,806 cost per person and 17.07 QALY per person. Therefore, the “base case” scenario yielded higher costs (+CHF 341) and higher QALY gains (+0.155) over 70 years resulting in an incremental cost-effectiveness ratio of CHF 2,203 per additional QALY gained. This incremental cost-effectiveness ratio falls in the north-eastern quadrant of the cost-effectiveness plane (figure 2).

Figure 2 Cost-effectiveness plane showing the incremental cost-effectiveness ratio for all three scenarios and the results of the probabilistic sensitivity analysis (PSA). The incremental cost-effectiveness ratio for the “base case” scenario equals to 2,203 CHF per quality-adjusted life year (QALY). The other scenarios simulating a “two-year delay” and “five-year delay” yield an incremental cost-effectiveness ratio of –101 CHF per QALY and –4,267 CHF per QALY, respectively. The PSA was run with 10,000 iterations. Cost and utility data was altered according to its standard errors.

CHF: Swiss francs

In the “two-year delay” scenario cost per person was CHF 4,451 and QUALY per person was 17.07 calculated over a period of 70 years. In comparison to the current model of care, cost reduction was CHF –14 and quality-adjusted life year gain per person was +0.13. The incremental cost-effectiveness ratio of the “two-year delay” equals to CHF –101 per additional QALY gained and dominates current model of care (figure 2).

In the “five-year delay” scenario cost per person was CHF 3,964 and quality-adjusted life years per person was 17.03 calculated over a period of 70 years. Comparing the “five-year delay” scenario with current model of care, a cost reduction of CHF –501 and a QALY gain per person of +0.12 was found. The corresponding incremental cost-effectiveness ratio of the “five-year delay scenario” equals extra CHF –4,267 per additional QALY gained when compared to the current model of care scenario (i.e., “dominant”) and falls in the south-eastern quadrant of the cost-effectiveness plane (Figure 2). The computed main results are reported in table 4.

Figure 2 presents the results of the PSA (10,000 iterations for each scenario) in a cost-effectiveness plane for all scenarios. Of the simulated incremental cost-effectiveness ratios of the “base case” scenario 15.8% fell into the south-eastern and 59.2% in the north-eastern quadrant. For the “two-year delay” scenario 27.3% fell into the south-eastern and 45.1% in the north-eastern quadrant, while for the “five-year delay” scenario 79.2% fell in the south-eastern and 0% in the north-eastern quadrant.

The modelled delay of total knee replacement by two and five years resulted in a reduction of revision surgeries by 18% and 36%, respectively. Figure 3 depicts the number of revision surgeries for each scenario.

Figure 3 Number of revision surgeries for each scenario and model cycle. The “two-year delay” scenario and “five-year delay” scenario resulted in a reduction of revision surgeries by 18% and 36% respectively when compared to the “base case scenario”.

In the cost-effectiveness acceptability analysis, we assumed a “willingness to pay” threshold of CHF 100,000 for Switzerland [26]. Based on this threshold, the probability for cost-effectiveness with the “base case” scenario equals 0.75, for the “two-year delay” scenario the probability is 0.72, and the probability is 0.79 for the “five-year delay” scenario (figure 4). All input parameters that were included in the probabilistic sensitivity analysis are reported in table 5.

Figure 4 Cost-effectiveness acceptability curve of the three analysed scenarios.

QALY: quality-adjusted life year

According to our model the lifetime probability to develop radiographically confirmed symptomatic knee osteoarthritis was 57% and the lifetime probability for total knee replacement after diagnosed knee osteoarthritis was 55.7%.

Discussion

This Markov modelling study is, to the best of our knowledge, the first to estimate the cost-effectiveness of a complete model of care involving both non-surgical and surgical treatments for knee osteoarthritis. We conducted model calculations for three different scenarios. The first scenario represented the implementation of guideline recommended non-surgical treatments with no delay on total knee replacement. The second and third scenarios represented the implementation of guideline recommended non-surgical treatments with a delay of total knee replacement by two and five years, respectively. The incremental cost-effectiveness ratio for the “base case” (current model of care versus optimised model of care with no delay of total knee replacement) was extra CHF 2,203 per additional quality-adjusted life year gained, which is highly cost-effective considering a “willingness to pay” threshold of CHF 100,000 per QALY in Switzerland [26]. The “two-year delay” and the “five-year delay” scenario both resulted in dominant incremental cost-effectiveness ratios and had a cost-saving effect. In all three scenarios, the incremental cost-effectiveness ratios were calculated over 70 years from the Swiss statutory healthcare perspective and dominated the current model of care. The achieved QALY gains over the modelling period are in line with other studies using a lifetime horizon [25].

Regarding undiscounted costs per person for each scenario, most costs were due to total knee replacement surgery (figure 5). This is explained by the high procedure costs. Another important share of costs was related to the “Successful non-surgical” state. Despite these costs being low, many individuals remain in this state for a long time accumulating costs. The effect of total knee replacement delay is most prominently reflected in the costs caused by “Revision surgery”. These costs were reduced by 55% with a delay of total knee replacement of five years when compared to current model of care.

Figure 5 Costs per person (undiscounted) for each scenario and the current model of care (CMOC). The optimised model of care calculated under “base case” conditions, meaning with no delay of total knee replacement (TKR), results in the highest overall costs (CHF 11,634). This is due the more intensive non-surgical treatment costs. The optimised model of care calculated with a delay of TKR by two years results in lower costs per person (CHF 10,997) and the optimised model of care calculated with a delay of TKR by five years reduces the total costs even further (CHF 10,044). The reduction of costs by these two scenarios can be explained because of less TKR and revision surgeries performed.

CHF: Swiss francs; TKR: total knee replacement

The observed reduction of revision surgeries by 18% in the “two-year delay” and 36% in the “five-year delay” scenario is due the fact that revision risk is cumulative and consequently becomes higher the longer the implant is in use, which rises in parallel with a younger age of implantation [11]. Our calculated reduction is supported by the results of Gademan et al. [39], who reported a reduction of revision surgeries by 17% when modelling the delay of total knee replacement by five years in individuals under 75 using joint registry data from the Netherlands. The difference to our results may be explained by higher total knee replacement rates (+58% in Switzerland versus the Netherlands [40]) and the assumption in this present health economic evaluation was that total knee replacement was delayed for all ages. There is great consent that the need for revision surgery should be kept at the lowest possible rate as it is considered an expensive and challenging procedure with often unreproducible results [41]. Revision rates are therefore frequently used to determine the outcome of total knee replacement [42].

The modelled lifetime probability to develop radiographically confirmed symptomatic knee osteoarthritis (57%) is somewhat higher than reported in the study of Murphy et al. using data from a cohort of 3,068 U.S. participants reporting a lifetime probability of 44.7% (95% CI 40%–49%) [37]. The probability for total knee replacement after diagnosed knee osteoarthritis in our model was 55.7%. This result is in line with the study of Weinstein et al. who reported 52.2% for males and 50.6% for females using the Osteoarthritis Policy Model, a validated computer simulation model of knee osteoarthritis based on U.S. data [38].

Our findings suggest that an investment in more intensive, guideline recommended non-surgical treatments is likely to be cost- effective in the context of the Swiss statutory healthcare system. If guideline recommended non-surgical treatments lead to a delay of total knee replacement, for which robust empirical evidence is still lacking, it may prevent a significant amount of revision surgeries and therefore has the potential to enhance total knee replacement outcomes and save costs. The developed model shows that non-surgical and surgical care should be delivered in an integrated manner with the aim to select patients for total knee replacement only after appropriate non-surgical care has failed. As clear indications for total knee replacement are still missing and the decision to undergo surgery is mostly influenced by the patient’s willingness [43, 44], it is vital to provide guideline recommended non-surgical treatments before total knee replacement is considered.

The reported results must be interpreted with caution because modelling studies always have limitations. First, we did not account for costs that are associated with inpatient rehabilitation after total knee replacement. Not accounting for these costs may have underestimated the cost-effectiveness of guideline recommended non-surgical treatments, as some studies suggest that up to 50% are referred to inpatient rehabilitation after total knee replacement [45].

Second, we did not adjust our model for patients who had already received total knee replacement and were therefore cured from knee osteoarthritis. This may have led to an overestimation of the performed total knee replacement surgeries in our model and therefore overestimated the cost-effectiveness of guideline recommended non-surgical treatments.

Third, we assumed that total knee replacement is the only available surgical treatment option. Nevertheless, in 2019, 2,908 partial knee replacements were implanted in Switzerland for the treatment of knee osteoarthritis (total knee replacement 15,378, partial knee replacements 2,908) and its usage is currently rising [34]. In addition to partial knee replacements, high tibial osteotomy is also a commonly performed surgical option for which we did not account for.

Fourth, other Markov models included re-revision surgery, for which we did not account for [46]. The assumption that all revisions are successful (meaning no re-revision is needed) is very conservative. In fact, re-revision surgery after revision surgery are six times more likely to be needed than revision surgery after total knee replacement [41]. Again, the lack of accounting for re-revision surgery is likely to have underestimated cost-effectiveness of guideline recommended non-surgical treatments.

Fifth, conclusions drawn from modelling studies need to be interpreted with caution. A narrow perspective was taken in this analysis and utility data from Denmark and the U.S. were used which were not degraded with rising age. This limits the model’s generalisability. To confirm our results, long-term data in the context of the Swiss statutory healthcare system is needed. Like the Swiss Implant Registry, which provides reliable data regarding total knee replacement and revision surgery, a national health record or an extension of this registry would be needed to provide data on how patients move through the healthcare system after they have been diagnosed with knee osteoarthritis.

Conclusion

According to this Markov model, the implementation of guideline recommended non-surgical treatments in Switzerland would likely be cost-effective from a statutory healthcare perspective. If implementing guideline recommended non-surgical treatments leads to delaying total knee replacement by two or five years, we would achieve a cost-saving effect and a reduction of needed revision surgeries by 18% and 36%, respectively.

Acknowledgements

Darren Mazzei and Niklaus Meier provided valuable advice while we developed the protocol and model.

Contributions: TV, NL and JT developed the study concept. To the study protocol inputs were provided by TV, NL and JT. TV and NL coded the model, interpreted results, and prepared the manuscript. All authors contributed to the final manuscript.

References

1. Cross M , Smith E , Hoy D , Nolte S , Ackerman I , Fransen M , et al.  The global burden of hip and knee osteoarthritis: estimates from the global burden of disease 2010 study. Ann Rheum Dis. 2014 Jul;73(7):1323–30. https://doi.org/10.1136/annrheumdis-2013-204763  

2. GHDx. Global Burden of Disease Study 2019 (GBD 2019) results. Published 2020. Accessed March 15, 2021. http://ghdx.healthdata.org/gbd-results-tool  

3. Schneider H , Venetz W.  Cost of Obesity in Switzerland in 2012. Rep Prep Behalf Bundesamt Gesundh. Published online 2012:82. 

4. Wang H , Bai J , He B , Hu X , Liu D . Osteoarthritis and the risk of cardiovascular disease: a meta-analysis of observational studies. Sci Rep. 2016 Dec;6(1):39672. https://doi.org/10.1038/srep39672  

5. Arden NK , Perry TA , Bannuru RR , Bruyère O , Cooper C , Haugen IK , et al.  Non-surgical management of knee osteoarthritis: comparison of ESCEO and OARSI 2019 guidelines. Nat Rev Rheumatol. 2021 Jan;17(1):59–66. https://doi.org/10.1038/s41584-020-00523-9  

6. McGrory BJ , Weber KL , Jevsevar DS , Sevarino K . Surgical Management of Osteoarthritis of the Knee: evidence-based Guideline. J Am Acad Orthop Surg. 2016 Aug;24(8):e87–93. https://doi.org/10.5435/JAAOS-D-16-00159  

7. Carr AJ , Robertsson O , Graves S , Price AJ , Arden NK , Judge A , et al.  Knee replacement. Lancet. 2012 Apr;379(9823):1331–40. https://doi.org/10.1016/S0140-6736(11)60752-6  

8. Skou ST , Roos EM , Laursen MB , Rathleff MS , Arendt-Nielsen L , Simonsen O , et al.  A Randomized, Controlled Trial of Total Knee Replacement. N Engl J Med. 2015 Oct;373(17):1597–606. https://doi.org/10.1056/NEJMoa1505467  

9. Beswick AD , Wylde V , Gooberman-Hill R , Blom A , Dieppe P . What proportion of patients report long-term pain after total hip or knee replacement for osteoarthritis? A systematic review of prospective studies in unselected patients. BMJ Open. 2012 Feb;2(1):e000435. https://doi.org/10.1136/bmjopen-2011-000435  

10. Healy WL , Della Valle CJ , Iorio R , Berend KR , Cushner FD , Dalury DF , et al.  Complications of total knee arthroplasty: standardized list and definitions of the Knee Society. Clin Orthop Relat Res. 2013 Jan;471(1):215–20. https://doi.org/10.1007/s11999-012-2489-y  

11. Bayliss LE , Culliford D , Monk AP , Glyn-Jones S , Prieto-Alhambra D , Judge A , et al.  The effect of patient age at intervention on risk of implant revision after total replacement of the hip or knee: a population-based cohort study. Lancet. 2017 Apr;389(10077):1424–30. https://doi.org/10.1016/S0140-6736(17)30059-4  

12. Hagen KB , Smedslund G , Østerås N , Jamtvedt G . Quality of Community-Based Osteoarthritis Care: A Systematic Review and Meta-Analysis. Arthritis Care Res (Hoboken). 2016 Oct;68(10):1443–52. https://doi.org/10.1002/acr.22891  

13. GLA . D® Schweiz: Arthrose nachhaltig mit Übungen behandeln. GLAD Schweiz. Accessed December 10, 2022. https://gladschweiz.ch/arthrose/  

14. physioswiss - Schweizer Physiotherapie Verband. Statistiken Mitgliederservices. Accessed December 10, 2022. https://www.physioswiss.ch/de/news/2022/statistiken-mitgliederservices  

15. Mazzei DR , Ademola A , Abbott JH , Sajobi T , Hildebrand K , Marshall DA . Are education, exercise and diet interventions a cost-effective treatment to manage hip and knee osteoarthritis? A systematic review. Osteoarthritis Cartilage. Published online November 2020:S1063458420311651. doi:https://doi.org/10.1016/j.joca.2020.10.002  

16. Skou ST , Roos E , Laursen M , Arendt-Nielsen L , Rasmussen S , Simonsen O , et al.  Cost-effectiveness of total knee replacement in addition to non-surgical treatment: a 2-year outcome from a randomised trial in secondary care in Denmark. BMJ Open. 2020 Jan;10(1):e033495. https://doi.org/10.1136/bmjopen-2019-033495  

17. Teoh LS , Eyles JP , Makovey J , Williams M , Kwoh CK , Hunter DJ . Observational study of the impact of an individualized multidisciplinary chronic care program for hip and knee osteoarthritis treatment on willingness for surgery. Int J Rheum Dis. 2017 Oct;20(10):1383–92. https://doi.org/10.1111/1756-185X.12950  

18. Husereau D , Drummond M , Petrou S , Carswell C , Moher D , Greenberg D , et al.; ISPOR Health Economic Evaluation Publication Guidelines-CHEERS Good Reporting Practices Task Force . Consolidated Health Economic Evaluation Reporting Standards (CHEERS)—explanation and elaboration: a report of the ISPOR Health Economic Evaluation Publication Guidelines Good Reporting Practices Task Force. Value Health. 2013;16(2):231–50. https://doi.org/10.1016/j.jval.2013.02.002  

19. Roberts M , Russell LB , Paltiel AD , Chambers M , McEwan P , Krahn M ; ISPOR-SMDM Modeling Good Research Practices Task Force . Conceptualizing a model: a report of the ISPOR-SMDM Modeling Good Research Practices Task Force-2. Med Decis Making. 2012;32(5):678–89. https://doi.org/10.1177/0272989X12454941  

20. RStudio Team . RStudio: Integrated Development Environment for R. Published online 2020. http://www.rstudio.com/  

21. Wickham H , Averick M , Bryan J , Chang W , McGowan L , François R , et al.  Welcome to the tidyverse. J Open Source Softw. 2019;4(43):1686. https://doi.org/10.21105/joss.01686  

22. Carta A , Conversano C . On the Use of Markov Models in Pharmacoeconomics: Pros and Cons and Implications for Policy Makers. Front Public Health. 2020 Oct;8:569500. [cited 2022 Mar 23] Available from: https://www.frontiersin.org/article/10.3389/fpubh.2020.569500 https://doi.org/10.3389/fpubh.2020.569500  

23. Losina E , Usiskin IM , Smith SR , Sullivan JK , Smith KC , Hunter DJ , et al.  Cost-effectiveness of generic celecoxib in knee osteoarthritis for average-risk patients: a model-based evaluation. Osteoarthritis Cartilage. 2018 May;26(5):641–50. https://doi.org/10.1016/j.joca.2018.02.898  

24. Ponnusamy KE , Vasarhelyi EM , Somerville L , McCalden RW , Marsh JD . Cost-Effectiveness of Total Knee Arthroplasty vs Nonoperative Management in Normal, Overweight, Obese, Severely Obese, Morbidly Obese, and Super-Obese Patients: A Markov Model. J Arthroplasty. 2018 Jul;33(7 7S):S32–8. https://doi.org/10.1016/j.arth.2018.02.031  

25. Rajan PV , Ng MK , Klika A , Kamath AF , Muschler GF , Higuera CA , et al.  The Cost-Effectiveness of Platelet-Rich Plasma Injections for Knee Osteoarthritis: A Markov Decision Analysis. J Bone Joint Surg Am. 2020 Sep;102(18):e104. https://doi.org/10.2106/JBJS.19.01446  

26. Ademi Z , Gloy V , Glinz D , Raatz H , Van S , Bucher HC , et al.  Cost-effectiveness of primarily surgical versus primarily conservative treatment of acute and subacute radiculopathies due to intervertebral disc herniation from the Swiss perspective. Swiss Med Wkly. 2016;146:w14382. https://doi.org/10.4414/smw.2016.14382  

27. Cui A , Li H , Wang D , Zhong J , Chen Y , Lu H . Global, regional prevalence, incidence and risk factors of knee osteoarthritis in population-based studies. EClinicalMedicine. 2020 Nov;29-30:100587. https://doi.org/10.1016/j.eclinm.2020.100587  

28. BFS . Struktur der ständigen Wohnbevölkerung nach Kanton, 1999-2019 - 1999-2019 | Tabelle. Bundesamt für Statistik. Published August 27, 2020. Accessed March 15, 2021. /content/bfs/de/home/statistiken/bevoelkerung/stand-entwicklung.assetdetail.13707204.html 

29. TARMED . TARMED Online Browser. TARMED Leistungen und Tarife nach Dignität, Leistungsblöcken oder Leistungsgruppen online abfragen. Published March 15, 2021. Accessed March 15, 2021. https://www.tarmed-browser.ch/de  

30. Swiss DR . SwissDRG. Published March 15, 2021. Accessed March 15, 2021. https://www.swissdrg.org/de  

31. BAG . Statistik der obligatorischen Krankenversicherung. Published online June 1, 2021. Accessed August 27, 2021. https://www.bag.admin.ch/bag/de/home/zahlen-und-statistiken/statistiken-zur-krankenversicherung/statistik-der-obligatorischen-krankenversicherung.html  

32. Skou ST , Roos EM , Laursen M , Arendt-Nielsen L , Rasmussen S , Simonsen O , et al.  Cost-effectiveness of 12 weeks of supervised treatment compared to written advice in patients with knee osteoarthritis: a secondary analysis of the 2-year outcome from a randomized trial. Osteoarthritis Cartilage. 2020 Jul;28(7):907–16. https://doi.org/10.1016/j.joca.2020.03.009  

33. Konopka JF , Lee Y , Su EP , McLawhorn AS . Lee Y yu, Su EP, McLawhorn AS. Quality-Adjusted Life Years After Hip and Knee Arthroplasty. JBJS Open Access. 2018;3(3):e0007. https://doi.org/10.2106/JBJS.OA.18.00007  

34. SIRIS . Swiss National Hip & Knee Joint Registry.; 2020. Accessed December 11, 2020. https://www.siris-implant.ch/media/archive1/201204_SIRIS-Report2020_online.pdf  

35. Bager CL , Karsdal M , Bihlet A , Thudium C , Byrjalsen I , Bay-Jensen AC . Incidence of total hip and total knee replacements from the prospective epidemiologic risk factor study: considerations for event driven clinical trial design. BMC Musculoskelet Disord. 2019 Jun;20(1):303. https://doi.org/10.1186/s12891-019-2680-3  

36. Briggs AH , Claxton K , Sculpher MJ . Decision Modelling for Health Economic Evaluation. Oxford University Press; 2006. 

37. Murphy L , Schwartz TA , Helmick CG , Renner JB , Tudor G , Koch G , et al.  Lifetime risk of symptomatic knee osteoarthritis. Arthritis Rheum. 2008 Sep;59(9):1207–13. https://doi.org/10.1002/art.24021  

38. Weinstein AM , Rome BN , Reichmann WM , Collins JE , Burbine SA , Thornhill TS , et al.  Estimating the burden of total knee replacement in the United States. J Bone Joint Surg Am. 2013 Mar;95(5):385–92. https://doi.org/10.2106/JBJS.L.00206  

39. Gademan MG , Van Steenbergen LN , Cannegieter SC , Nelissen RG , Marang-Van De Mheen PJ . Population-based 10-year cumulative revision risks after hip and knee arthroplasty for osteoarthritis to inform patients in clinical practice: a competing risk analysis from the Dutch Arthroplasty Register. Acta Orthop. 2021 Jun;92(3):280–4. https://doi.org/10.1080/17453674.2021.1876998  

40. OECD . Health at a Glance 2017: OECD Indicators. Published 2017. Accessed December 8, 2020. https://www.oecd-ilibrary.org/social-issues-migration-health/health-at-a-glance-2017_health_glance-2017-en  

41. Ong KL , Lau E , Suggs J , Kurtz SM , Manley MT . Risk of subsequent revision after primary and revision total joint arthroplasty. Clin Orthop Relat Res. 2010 Nov;468(11):3070–6. https://doi.org/10.1007/s11999-010-1399-0  

42. Bae DK , Song SJ , Heo DB , Lee SH , Song WJ . Long-term survival rate of implants and modes of failure after revision total knee arthroplasty by a single surgeon. J Arthroplasty. 2013 Aug;28(7):1130–4. https://doi.org/10.1016/j.arth.2012.08.021  

43. Mandl LA . Determining who should be referred for total hip and knee replacements. Nat Rev Rheumatol. 2013 Jun;9(6):351–7. https://doi.org/10.1038/nrrheum.2013.27  

44. Wright JG , Hawker GA , Hudak PL , Croxford R , Glazier RH , Mahomed NN , et al.; Toronto Arthroplasty Research Group Writing Committee . Variability in physician opinions about the indications for knee arthroplasty. J Arthroplasty. 2011 Jun;26(4):569–575.e1. https://doi.org/10.1016/j.arth.2010.04.028  

45. Ong KL , Lotke PA , Lau E , Manley MT , Kurtz SM . Prevalence and Costs of Rehabilitation and Physical Therapy After Primary TJA. J Arthroplasty. 2015 Jul;30(7):1121–6. https://doi.org/10.1016/j.arth.2015.02.030  

46. McLawhorn AS , Southren D , Wang YC , Marx RG , Dodwell ER . Cost-Effectiveness of Bariatric Surgery Prior to Total Knee Arthroplasty in the Morbidly Obese: A Computer Model-Based Evaluation. J Bone Joint Surg Am. 2016 Jan;98(2):e6. https://doi.org/10.2106/JBJS.N.00416  

Appendices: Supplementary data

The Appendices 1-4 are available for download as separate files at https://doi.org/10.57187/smw.2023.40059.