Costs and models used in the economic analysis of Total Knee Replacement (TKR): A Systematic Review

Objectives: The major objective of this review was to summarize the evidence on the core modelling specifications and procedures on the cost-effectiveness of TKR compared to non-surgical management. Another objective of this study was to synthesize evidence of TKR cost and compare it across countries using purchasing power parity (PPP). Methodology: The electronic databases used were MEDLINE (PubMed), Cochrane Central Register of Controlled Trials (CENTRAL), HTAIn repository and Cost effectiveness Analysis (CEA) registry. Consolidated Health Economic Evaluation Reporting Standards (CHEERS) was used to assess the validity of the methods and transparency in reporting the results of the included studies. The cost of TKR surgery from high income and low- or middle-income countries were extracted and converted to single USD ($) using purchasing power parities (PPP) method. Result: 29 studies were included in this review, out of which eight studies used Markov model, five studies used regression model, one study each reported Marginal structure model and discrete simulation model and decision tree analysis to assess cost-effectiveness of TKR. For PPP, 23 studies were included in the analysis of TKR cost. The average cost of TKR surgery was lowest in developing country like India ($3457) and highest in USA ($19,645). Conclusion: The findings of this review showed that the Markov model was most widely used in the analysis of the cost effectiveness of TKR. Our review also concluded that the cost of TKR was higher in developed countries as compared to developing countries.

Osteoarthritis is a degenerative joint disease involving the cartilage and surrounding tissues 44 and is the leading cause of disability worldwide among older adults (1). Globally, the 45 prevalence of OA knee was estimated to be around 22.9% (2) whereas In India, it was 46 estimated to be 28.7% (3). The direct cost of OA knee was US$5294 per person per year for those aged over 65 years and $5704 for patients less than 65 years. This was estimated to be 48 twice of non-OA patients. The Indirect costs of OA knee was around US$4603 per person 49 annually, mainly due to work-related losses and home-care costs. The Total knee replacement 50 (TKR) is considered as one of the interventions to overcome the burden of OA knee. The 51 number of TKRs being done to mitigate the burden of OA knee has also been increasing 52 throughout the world. In the United States of America, in the year 2010, approximately 700,000 53 TKR surgeries were performed, and its demand is predicted to grow to 3.8 million per annum 54 by the year 2030 (4). 55 The cost of TKR in developed countries -USA is around $17500 (in 2017) and in European 56 countries -Denmark and UK are €13149 (in 2020) and £7313 (in 2013) respectively, however, 57 in developing countries like India, it is ₹80,000 (in 2021) (5)(6)(7)(8). Many developed countries 58 have considered cost-effectiveness analysis as one of the methods for policy level decision-59 making. However, India, which has multiple health system constraints and limited government 60 investment on health, is progressively preparing to include cost-effective analysis as a tool for 61 decision-making at the policy level. There has been only one study conducted in India, which 62 showed that TKR is cost-effective in the base case scenario with an Incremental Cost 63 Effectiveness Ratio (ICER) of ₹9789 ($132.3) per QALY (9). is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted November 14, 2022. ; https://doi.org/10.1101/2022.11.14.22282318 doi: medRxiv preprint systems. Moreover, given varieties of models-the decision tree, Markov model, and regression 69 analysis (10) used in these types of studies, identifying the suitable model also assume 70 significance. During our literature search, four systematic reviews were identified which 71 compared the cost-effectiveness of TKR to non-surgical management in patients with OA knee.

72
Two reviews compared the cost-effectiveness of TKR through ICER value irrespective of the 73 models used (11,12). A recent review by Kamaraj et al (13) assessed the scope and quality of 74 all current cost-effectiveness analysis (CEA) studies for TKR in order to identify trends, and  However, none of the systematic reviews mentioned above suggested a suitable method for the 79 analysis of cost-effectiveness of TKR. Therefore, we aim to perform a systematic review of the 80 methods used in economic evaluations/cost-effectiveness of TKR compared to non-surgical 81 management of OA knee. The major objective of this study is to summarize the evidence on 82 the core modelling specifications and procedures on the cost-effectiveness of TKR compared 83 to non-surgical management. Another objective of this study is to synthesize evidence of TKR 84 cost and make different countries cost comparable using purchasing power parity (PPP), which 85 has not been done before. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint In this review, all the reports of randomized control trials (RCT's) and cohort studies were 95 included. Studies like cross-sectional, observational and case-control that have the possibility 96 to provide information on cost-effectiveness, cost benefit analysis, and cost utility analysis 97 were also included in our initial search.   is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted November 14, 2022. ; https://doi.org/10.1101/2022.11.14.22282318 doi: medRxiv preprint analysis) registry. The search history was conducted from the earliest possible date till 4 th June 113 2021 and filters were not applied for time period. The computer-based search terms included 114 the combination of keywords and Mesh terms like "Total Knee Arthroplasty" and "Cost-    only QALY comparison were reported. Non-economic evaluations were also excluded. 137 7. Studies such as systematic reviews, letters to the editors, commentaries, and protocols 138 were excluded.

140
All the identified studies from the database were imported to the RAYYAN software and 141 duplicates were removed. During the "first pass", three authors independently reviewed the 142 study titles and abstracts before concluding whether the study should be "included," 143 "excluded," or there is "uncertainty" about it. The consensus was reached in studies that had 144 "uncertainty." The remaining studies were excluded, while those that were "included" were 145 considered for the next stage of selection.

146
Three authors retrieved the full texts of the selected abstracts from the first pass and screened 147 them again. Similar screening methods were applied, but the authors agreed to include papers 148 through an iterative consultation process for those with conflicting findings. Papers with 149 unresolved conflicts were excluded after this process.  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted November 14, 2022. ; https://doi.org/10.1101/2022.11.14.22282318 doi: medRxiv preprint 158 The cost of TKR differs across different metrics such as time and currency. Change in the 159 prices due to inflation have affected the cost of TKR over the years. This study adjusted the if not reported. NA was reported if the item was "not applicable". A score of 1 was assigned if 175 the criteria was "Yes", 0.5 if it was "Part" and 0 if the criteria "No". Hence, the maximum and 176 minimum score for a study were 24 and 0 respectively. Additionally, the percentage was 177 determined under the presumption that each study received equal weighting, with "Not 178 Applicable" item being left out. Studies with a score of at least 75% were defined as high 179 quality, those between 60 and 75% as moderate quality, and those below 60% as low quality. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

Conversion of Costing data on Purchasing Power Parities (PPP)
The copyright holder for this this version posted November 14, 2022. ; https://doi.org/10.1101/2022.11.14.22282318 doi: medRxiv preprint The quality assessment did not influence the inclusion or the exclusion of studies. Studies 181 meeting the inclusion criteria were included for the purpose of this review irrespective of the 182 quality criteria.

184
Due to heterogeneity of the included studies and reported cost data, formal meta-analysis could 185 not be performed. Thus, each study was analysed qualitatively and were primarily presented in 186 tables and graphs.

187
As for PPP, out of the twenty-nine included studies, six studies were excluded from the 188 analysis, of which three studies were excluded due to unavailability of cost data (17)(18)(19) and  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted November 14, 2022.  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted November 14, 2022. ; https://doi.org/10.1101/2022.11.14.22282318 doi: medRxiv preprint time horizon. Two studies (31,37) used more than one time horizon to estimate their functional 227 improvement and cost effectiveness.

228
Nineteen studies reported discount rates, out of which, thirteen studies (6,9,19 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted November 14, 2022. ; https://doi.org/10.1101/2022.11.14.22282318 doi: medRxiv preprint

(17.2)
Time Horizon [2] Lifetime <12 Months is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted November 14, 2022. ; https://doi.org/10.1101/2022.11.14.22282318 doi: medRxiv preprint [1] The studies used more than one comparator and approach. Hence there is an overlapping.

268
The cost of TKR surgery in India was found to be the cheapest at $3457 followed by China at   is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted November 14, 2022. ; https://doi.org/10.1101/2022.11.14.22282318 doi: medRxiv preprint The main objective of this study was to summarize the evidence on the core modelling 284 specifications on the cost-effectiveness of TKR compared to non-surgical management.

285
Another aim of this study was to synthesize evidence of TKR cost and to compare the variations 286 across different countries through purchasing power parity (PPP) by converting the cost in each 287 study to international dollar ($).

288
This systematic review assessed the models used in cost utility or cost-effectiveness of the 29 289 included studies, out of which, sixteen studies reported methodology to determine cost-290 effectiveness. Eight studies used Markov model (5,7,18,23,24,27,29,37) whereas five studies 291 reported regression model (6,33,34,36,38). One study each reported Marginal structure 292 model(28), discrete simulation model (32) and decision tree analysis (8) to assess the cost- is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted November 14, 2022. ; https://doi.org/10.1101/2022.11.14.22282318 doi: medRxiv preprint intervention compared to the old intervention. So, the ICER is very easy to understand and is 308 also useful in the decision of policy. The Markov model can use data from real world as well 309 as from assumed data set to calculate ICER value. So, the real-world data can also be used to  The cost of TKR surgery widely varied across different countries. Another important finding 322 of our review is that the cost of TKR surgery was three to five times higher in developed 323 countries such as USA and UK than developing countries such as India after adjusting cost in 324 PPP terns. These variations in cost were mostly due to the difference in the cost of human 325 resources. Despite the high cost of TKR in developed countries, the TKR surgery was found 326 to be cost effective than non-surgical management for OA knee (23). Similarly, the TKR 327 surgery in developing country like India was also found to be cost effective (7). So, when the 328 economic condition in developing country improves, the cost of TKR surgery may increase but 329 it will still be a cost-effective strategy to treat OA knee. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Since studies included in our review were from various countries, the generalisability of our is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted November 14, 2022. ; https://doi.org/10.1101/2022.11.14.22282318 doi: medRxiv preprint highest in a developed country like USA. More studies with high methodological standards in 355 developing countries are recommended.
. CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted November 14, 2022. ; https://doi.org/10.1101/2022.11.14.22282318 doi: medRxiv preprint