The cost-effectiveness of computer navigation in primary total knee replacement: a scoping review

Search strategy

The results are illustrated in the PRISMA flow diagram (Fig. 1). A total of four cost-effectiveness analyses met eligibility for inclusion.

View Full Size

PRISMA flow diagram.

Citation: EFORT Open Reviews 6, 3; 10.1302/2058-5241.6.200073

• Download Figure
• Download figure as PowerPoint slide

Study characteristics

The data abstraction is presented in Table 1. All studies simulated scenarios over a fixed number of monthly or yearly cycles; none were performed as observational or clinical trials. Modelling was performed using a range of values under sensitivity analyses, and cost-effectiveness was estimated using input data from both published literature and estimates. The payer perspective adopted was universally that of the healthcare system, in contrast to the societal perspective which also includes costs incurred outside of the healthcare system. ¹⁹ All studies nominated a locally applicable cost-effectiveness threshold against which scenarios were compared to determine cost-effectiveness. There were a total of four cost analyses: two studies presented the probabilities of achieving cost-effectiveness across the scenarios in their modelling, and two studies presented suggested reductions in the rate of revision surgery required to achieve cost-effectiveness. ^20–23

Effectiveness of computer navigation

Effectiveness was measured using QALYs. ¹⁰ QALYs are calculated by adjusting utility values for survival time. ²⁴ Utility values are a single index that represents preferences for different states of health, ranging on a scale from 0 (death) to 1 (full health), and are typically derived from health-related quality of life scores of questionnaires such as the Short-Form 6 Dimension (SF-6D) and EuroQol 5 Dimension (EQ-5D). ²⁵ Each of the four studies derived utility values from either the SF-6D or the EQ-5D, and these data are presented in Table 1.

Across the included studies, the cumulative QALY gain following TKR was determined by adding the utility value at each cycle over the full cycle. The cycles started from a pre-operative state of knee arthritis, progressed to primary TKR following either navigated or non-navigated arms, and then various states of transition were modelled including neutral alignment/malalignment, complications, revision TKR, and death. The utility value for each state, survival from revision, and probabilities of transitioning across each state during the cycle were derived from existing arthroplasty literature, the Swedish and Norwegian arthroplasty registries, and United States Medicare cohort data. No differences in the TKR utility values between the navigated and non-navigated arms were modelled in the analyses. The impact of computer navigation on QALYs was primarily due to reductions in the state of malalignment and reductions in transitioning to a revised TKR state. Death with an unrevised TKR resulted in a higher cumulative lifetime QALY gain compared to death after transitioning to a revised TKR state. Navigated TKR, with reductions in malalignment and hence a reduced probability of transitioning to the revised state, therefore resulted in higher cumulative lifetime QALYs.

Utility values attributed to the primary TKR across the four analyses ranged from 0.73 to 0.92 QALYs (with sensitivity analysis values ranging from 0.00026 to 1). Utility values attributed to revision TKR ranged from 0.51 to 0.80 QALYs (0 to 0.99997). Two studies factored in a disutility value for the first year following primary TKR (–0.10 QALYs) or revision TKR (–0.20 QALYs), which is a reduction in the utility value for the first year due to the recovery period following surgery prior to attaining a steady-state utility value. ^22,23 A number of other transitional states between primary TKR in a ‘normal’ state of health and ‘complex’ revision TKR were included as transition states.

The use of computer navigation was estimated to increase QALYs by 0.0192 (95% CI –0.002 to 0.0473) in the United States over a 15-year cycle, and by between 0.0148 to 0.0164 QALYs in the United Kingdom over a 10-year cycle. ^20,21

Cost of computer navigation

Costs in the four studies referred to the price paid for purchasing the navigation system and associated expenses, with one study also pricing the additional consumption of resources related to extra operating time required for use of computer navigation (Table 1). ²⁰ The cost of purchasing equipment (system, software, maintenance, disposables) was typically obtained from the hospital purchasing department or vendors. Upfront costs of purchasing the computer navigation system were averaged over a five-year useable lifespan. Discounting rates of 0% to 5% per year were factored into the modelling to account for the impact of inflation over time. The included studies did not evaluate costs associated with prostheses or training.

For an annual procedural volume of 250 TKRs, the average additional cost per case was estimated at US$192 in the United States, £235 (US$292, 2020) in the United Kingdom, and 952 kr (US$90, 2020) in Norway. At 25 TKRs per year, this was estimated at US$1,920 per case in the United States, and 1,037–1,414 kr (US$98-133, 2020) for 60-year-old patients and 75-year-old patients respectively in Norway.

One analysis in the United States estimated that cost savings could be achieved if the additional cost per case was < US$629, due to reductions in the number of higher-cost revision TKRs over a 15-year cycle of care. ²¹ A British analysis estimated that in some scenarios cost savings of £583 to £637 (US$727 to US$795, 2020) per case could be achieved over a 10-year cycle of care. ²⁰

Cost-effectiveness of computer navigation

The QALY gains from computer navigation, at an additional cost of US$1,500, resulted in an incremental cost-effectiveness ratio of $45,554/QALY in the United States; 54% of simulations were cost-effective below the local threshold of US$50,000. ²¹ In the United Kingdom, computer navigation was suggested to be more effective and less costly in greater than 75% of simulations, with a 92% probability of achieving cost-effectiveness at a £30,000 threshold. ²⁰ Two studies proposed that reductions in the rate of revision surgery were required to attain cost-effective outcomes (Table 1). ^22,23 This requirement ranged from relative reductions of 13.0% at a case volume of 25 TKRs/year to 2.0% at case volumes of 250 TKRs/year in the United States. ²² In Norway, the required reductions were 7.0% (75-year-old cohort) and 7.5% (60-year-old cohort) at 25 TKR/year case volume to 1.0% for both age cohorts at 250 TKR/year case volume. ²³ With larger case volumes, smaller reductions in the percentage of patients transitioning to the revised state and its associated lower utility value were required to achieve QALY gains sufficient to offset the increased costs of computer navigation.

The cost-effectiveness of computer navigation in primary TKR

This review was performed to evaluate the cost-effectiveness of computer navigation in TKR. Computer navigation has not been demonstrated to be a clearly dominant (more effective and less costly) nor dominated (less effective and more costly) strategy; included studies suggested the use of this technology is associated with marginal additional QALYs at additional cost.

Markov analysis was used to model transitions across a range of health states to estimate the cost-effectiveness of computer navigation in primary TKR. ^20–23 Cost-effectiveness analysis requires a balanced assessment of both cost and effectiveness, summarized as an incremental cost-effectiveness ratio (ICER) which is the cost of attaining an additional QALY (Fig. 2). ICERs are compared to a nominated cost-effectiveness threshold, with interventions whose ICERs fall below this threshold deemed cost-effective and those that lie above deemed not cost-effective. The use of ICERs aids decision analysis for interventions that improve QALYs at additional cost, as this requires a trade-off between improved health outcomes but greater investment of scarce health resources. Interventions that both improve QALYs and reduce costs are deemed ‘dominant’, and those that both reduce QALYs and incur greater costs are in contrast considered ‘dominated’ strategies. ²⁶

View Full Size

Formula for the incremental cost-effectiveness ratio (ICER).

Note. QALY, quality-adjusted life year.

Citation: EFORT Open Reviews 6, 3; 10.1302/2058-5241.6.200073

• Download Figure
• Download figure as PowerPoint slide

It remains unclear whether the use of computer navigation in TKR is cost-effective due to variation in underlying assumptions and transition states. The cost-effectiveness of TKR has been extensively evaluated under a range of settings, with a number of outcomes and cost drivers reported. ^27,28 By comparison, the investigation of the cost-effectiveness of computer navigation in primary TKR has been limited. There may be a role for its targeted adoption if clear evidence of improvements in quality-adjusted life years and reductions in the rate of revision surgery can be demonstrated at an appropriate cost. The widespread adoption of newer technologies prior to evidence of benefit may shift limited healthcare resources towards lower-value care. ²⁹ The value analysis of computer navigation is particularly important due to the increasing use of high-cost healthcare technologies which are widely recognized as a significant cost driver of healthcare expenditure. ³⁰

Limitations to practice and policy translation

It has been recognized that numerous studies may be underpowered to unmask potentially small but significant differences in outcomes within certain patient cohorts. ³¹ This is particularly worth noting as the reported QALY gains from computer navigation have been fairly small relative to the range of utility values assumed or estimated in the modelling scenarios. There was a large degree of variation in the utility values assigned to the same health states by different authors, and this was largely a reflection of the broader published literature as stated Gøthesen et al. ²³ Dong and Buxton in their study assigned utility values to certain health states based on assumptions due to a lack of published data. ²⁰ Similarly, utility values from total hip replacement cohorts were adopted for total knee replacement cohorts by Gøthesen et al and Slover et al. ^22,23 Whilst hip and knee replacements are both arthroplasties of large lower extremity joints, they are distinct procedures and differences in utility values have been reported by Konopka et al. ³² Further to this, Schilling et al have demonstrated the impact of timing on utility values and QALYs that are subsequently derived from health-related quality of life scores. ³³ These limitations indicate that even relatively minor changes in the underlying assumptions, and differences in underlying assumptions, may be sufficient to affect the interpretation of the outcomes reported by the modelling despite sensitivity analyses. The underlying assumptions will need to be revisited and further investigated as new data come to light with longer-term follow-up now available.

The effectiveness of computer navigation was generally attributed to reductions in the rate of revision surgery as a result of reductions in the proportion of malaligned TKRs. Two analyses estimated reductions in the rate of revision surgery that were required to meet nominated cost-effectiveness thresholds for a range of surgical volumes ranging from 25 to 250 TKRs/year. ^22,23 However, malalignment is the underlying aetiology for only a proportion of revision TKRs. ³⁴ The effectiveness of computer navigation as measured by reductions in the rate of revision surgery has largely not been borne out by the literature, with limited exceptions. ⁷ Revision rates were derived from a diverse range of sources and based on survival data from surgeries performed in previous decades. With advancements in technique and technology, any narrowing of potential differences in survival between mechanical and computer navigated TKR may have implications for the interpretation of existing cost-effectiveness analyses. It remains unclear whether the proposed reductions in revision surgery are feasible, particularly in lower-volume settings where the required reduction is considerably greater compared to higher-volume settings.

The analyses suggested that the additional cost of computer navigation for each case decreased in association with higher surgical volumes. Lower-volume hospitals operated at a higher unit cost due to the significant capital investment required to procure a computer navigation system. ³⁵ Lower-volume hospitals also had a lower likelihood of meeting nominated cost-effectiveness thresholds as a result of these higher costs. ²² Despite this, the benefits of computer navigation were considered more likely to be realised in lower-volume hospitals due to the possibility of mitigating poorer outcomes. The literature has reported associations between lower hospital volumes and inferior patient outcomes. ³⁵ Further, where the proposed cost-effectiveness of computer navigation is likely improved with higher surgical volumes, the majority of hospitals in the United States do not meet the 250 annual case volume which has informed modelling estimates. ³⁵ In the United States, for example, Katz et al reported that only 25% of TKRs were performed at hospitals with annual case volumes exceeding 200 TKRs. ³⁵ Similar findings may apply to other countries where the use of computer navigation in TKR is also increasingly prevalent. ³⁶ The differences between surgical volumes modelled in the analyses, with direct implications for cost-effectiveness, and the reported surgical volumes undertaken in modern practice settings limits the translation of findings to practice.

Navigating the way forward

The selective use of computer navigation technology has already proven to be advantageous in technically challenging cases where complex deformities or distorted anatomical planes impair the ability to accurately identify landmarks for mechanical referencing. Whilst the definition of accurate alignment is still a matter of active debate, the benefit of computer navigation in enabling greater surgical precision is not disputed. ⁹ The impact of malaligned prostheses on failure rates is markedly higher in the presence of obesity, defined by the World Health Organization as a body mass index ⩾ 30 kg/m². ³⁷ In patients < 65 years old, Australian national registry data demonstrate lower revision rates amongst patients receiving navigated TKRs at 9 years of follow-up. ⁷ Obese patients represent a significant number of TKR recipients, and patients < 65 years of age represent an increasing proportion of TKR recipients. ³⁸

Despite the promising benefits in certain patient cohorts, the additional costs incurred and inconsistencies in reported outcomes have encouraged recommendations against universal adoption. ^8,9,31 However, a considerable period of time has elapsed since the introduction of computer navigation to TKR. Only in recent years have differences in outcomes and rates of revision surgery started to emerge. ^7,39 This offers an opportunity for targeted investigation into the impact of computer navigation on the value and cost-effectiveness of TKR. Computer navigation technology continues to remain an active area of research interest, and long-term follow-up data will help inform future economic decision analyses with a focus on translation to practice and policy settings. This should account for variations in local practice patterns, costs of procuring technologies, and other factors relevant to local cost-effectiveness evaluations. Further research is encouraged to elucidate these differences. The selective use of advanced technology has the potential to improve the cost-effectiveness of surgical practice in an era where the value of high-cost interventions increasingly needs to be justified.