Cost-Utility Analysis of Ligament Reconstruction Tendon Interposition Versus Suture Suspension Arthroplasty for Thumb Osteoarthritis

Abstract

Background:

Thumb carpometacarpal joint osteoarthritis (CMCJ OA) is a common degenerative condition that causes pain, stiffness, and disability, reducing quality of life. Surgery is a well-established treatment option when conservative management fails, but the optimal surgical approach remains debated. This study compared the cost-utility of trapeziectomy with ligament reconstruction and tendon interposition (LRTI + T) versus suture suspension arthroplasty (SSA) for CMCJ OA.

Methods:

A Markov microsimulation model was developed to compare LRTI + T and SSA from a hospital payer perspective. Outcomes included incremental cost-effectiveness ratio, quality-adjusted life years (QALYs), total cost, and net monetary benefit. Clinical outcomes such as complication rates and revision surgery were also evaluated.

Results:

LRTI + T had a higher complication rate (14.6%) than SSA (9.8%), but SSA had a slightly higher revision rate (7.1% versus 5.7%). Over a lifetime, SSA provided an incremental gain of 0.25 QALYs but was marginally more expensive ($2855 versus $2842). SSA yielded an incremental cost-effectiveness ratio of $53.80 per QALY, making it the more cost-effective strategy.

Conclusions:

SSA is a cost-effective alternative to LRTI + T, offering valuable insights for clinicians and policymakers optimizing care for CMCJ OA patients.

Takeaways

Question: What is the most cost-effective surgical option for managing thumb carpometacarpal joint osteoarthritis: ligament reconstruction tendon interposition with trapeziectomy (LRTI + T) or suture suspension arthroplasty (SSA)?

Findings: Using a Markov microsimulation model, SSA provided a small quality-adjusted life year (QALY) gain (0.25 QALYs) over LRTI + T with a marginal cost difference ($2855 versus $2842). Despite slightly higher revision rates, SSA yielded an incremental cost-effectiveness ratio of $53.80 per QALY, indicating cost-effectiveness.

Meaning: SSA is a cost-effective surgical option for thumb carpometacarpal joint osteoarthritis compared with LRTI + T.

INTRODUCTION

Osteoarthritis (OA) affects the thumb carpometacarpal joint (CMCJ), causing pain, stiffness, and disability. In cases where conservative treatment methods are ineffective and patients experience persistent CMCJ pain, surgery can be considered. A recent cost-utility analysis comparing conservative and surgical management of thumb CMCJ OA demonstrated that surgery is cost-effective, with an incremental cost-effectiveness ratio (ICER) of US $25,370 per quality-adjusted life year (QALY).¹

There exists a number of surgical treatment (arthroplasty) options to reduce pain and increase function.² Surgical decision-making is often guided by disease severity, most commonly classified using the Eaton and Littler system, which grades radiographic progression of thumb CMCJ OA.³ The choice of surgical treatment remains controversial. One of the most widely used techniques is the Burton–Pellegrini procedure, which involves trapeziectomy combined with palmar oblique ligament reconstruction and tendon interposition (LRTI + T) using a slip of the flexor carpi radialis tendon.⁴ LRTI + T is one of the most common procedures and is the treatment of choice for 68% of members of the American Society for Surgery of the Hand.⁵ More recently, suture suspension arthroplasty (SSA), which involves using a synthetic construct to maintain trapezial height and prevent subsidence of the thumb metacarpal following trapeziectomy, has been introduced. The technique permits immediate thumb range of motion and hand therapy at the first postoperative visit. Additionally, morbidity associated with tendon autograft harvest (ie, palmar cutaneous branch of the median nerve injury and pulling at the flexor carpi radialis) is avoided with this technique. However, the use of a synthetic implant in SSA increases the procedural cost and potentially leads to hardware-related complications, such as failure. Therefore, it is important to carefully evaluate whether the benefits and risks offered by SSA, when compared with traditional surgical methods, justify its added cost.

Given rising healthcare costs and a shift toward value-based care, economic evaluations are essential. Traditional studies lack a comprehensive cost-effectiveness comparison of surgical options. The aim of this cost-utility analysis is to compare 2 classes of thumb CMCJ arthroplasty: LRTI + T and SSA. By assessing the economic aspects of these 2 interventions and their associated health outcomes in this cost-utility analysis, we seek to provide valuable insights that can inform decision-making processes for healthcare professionals, policymakers, and patients in a strained healthcare system. We hypothesized that SSA would be the more cost-effective surgical option compared with LRTI + T.

MATERIALS AND METHODS

Overview of the Analysis

We conducted a cost-utility analysis comparing the hospital payer perspective for 2 treatment options for patients with thumb CMCJ OA whose medical treatment failed, thus requiring surgery: (1) LRTI + T and (2) SSA, using TreeAge Pro Software (TreeAge Software, Inc., Williamstown, MA). The model followed best practice recommendations by the Panel on Cost-effectiveness in Health and Medicine and the Canadian Agency for Drugs and Technologies in Health.

A Markov model was created to simulate the disease history of a hypothetical cohort of patients with an average age of 67 years⁶ with thumb CMCJ OA. Patients progressed through health states in 1-year time steps. A time horizon of 23 years was chosen, in the context of the median age of the cohort at 67 years at the beginning of the analysis, to reflect a lifetime horizon of 90 years. The model compared health outcomes and costs of the 2 treatment strategies. Cost and effectiveness were discounted at an annual rate of 1.5%.⁷

The primary outcome of interest is the ICER, QALYs, total cost, and net monetary benefit (NMB). Other outcomes of interest included clinical outcomes (complication and revision rates). A successful outcome was defined as resolution of pain following surgery, without the need for additional intervention. An unsuccessful outcome was defined as persistent pain following surgery, with or without the need for revision. Complications were defined as tendon rupture, nerve injury, or infection, as specified by the original study authors. Revision surgery was defined as a procedure performed following an unsuccessful initial surgery, characterized by persistent pain in the thumb.

Model Structure

The Markov microsimulation model is shown in Supplemental Digital Content 1 and Figure 1. (See table, Supplemental Digital Content 1, which displays the tree structure, https://links.lww.com/PRSGO/E385.) A Markov model was selected, as it permits the creation of a cohort of patients with varying individual characteristics and follows them over time, effectively mimicking the real-life progression of patients with thumb CMCJ OA.

Fig. 1.

Transition between states.

In each chosen surgical approach, simulated patients enter the model at the time of their initial surgery. All patients spent a full cycle (1 y) in the postoperative state. In this postoperative state, patients may experience complications secondary to their procedure, which can require operative or nonoperative management. Patients are then transitioned to (1) a nonpainful thumb CMCJ (successful outcome), (2) a painful thumb CMCJ (unsuccessful outcome), or (3) death (Fig. 1). All patients in the model experienced a population-level mortality rate.⁸

If the initial surgery was successful (probability of success of LRTI + T, 0.7202; probability of success of SSA, 0.9342), patients transitioned to the nonpainful thumb CMCJ state. If the initial surgery was unsuccessful, patients transitioned to the painful thumb CMCJ state. In the painful thumb CMCJ state, patients may undergo additional surgical intervention (ie, revision surgery). Rates of revision surgery for each treatment arm were obtained from the literature to reflect the probability of patients who wanted and were offered revision surgery, if appropriate. The probability of revision surgery was 0.0112 for LRTI + T and 0.0135 for SSA. See Table 1^8–40 for further details.

Table 1.

Probabilities Included in the Model

	Probability*	References
LRTI + T
Overall complication	0.1397	9–21
Complication requiring surgery	0.0388	10–20
Complication requiring no surgery	0.9612	10–20
Success of surgery	0.7202	10,13–19,22
Revision surgery	0.0112	10,12–15,17,18,20,21,23
Success of revision surgery	0.3601*	10,13–19,22
SSA
Overall complication	0.0778	9,19,24–40
Complication requiring surgery	0.2446	9,19,24–40
Complication requiring no surgery	0.7554	9,19,24–40
Success of surgery	0.9342	30,33,34,36,37
Revision surgery	0.0135	9,26,27,30,36–39
Success for revision surgery	0.4671†	9,26,27,30,36–39
Death	0.0091	8

^*Probability calculated using means across individual studies.

^†Success of revision surgery estimated to be half that of the initial surgery.

Patients who underwent revision surgery were transitioned back into the postoperative state, where they may experience a procedure-related complication. If the revision surgery was successful, they would be transitioned into the nonpainful thumb CMCJ state, and if the revision surgery was not successful, they would be transitioned back into the painful thumb CMCJ state. Note that revision surgery is only offered once for patients.

Patients in the painful thumb CMCJ state who did not undergo revision surgery may be transitioned back into the nonpainful thumb CMCJ state as the years pass, to reflect patients whose pain improved over time with or without medical treatments such as physiotherapy and steroid injections. Similarly, patients in the nonpainful thumb CMCJ state can be transitioned to the painful thumb CMCJ state to reflect recurrence of pain over time, despite a successful initial surgery (probability of failure = 1 − probability of success).

Model Assumptions

The design of this model involved several assumptions. First, nonoperative treatments were modeled as a unified group, assuming that both patient groups underwent similar nonoperative treatments. Consequently, the likelihood of success of surgery was presumed to be unaffected by the previous nonoperative management. Next, we assumed that patients with a successful surgical outcome were at risk of redeveloping pain at a later time.⁴¹ In addition, based on clinical expertise, we assumed that complications secondary to the surgery would occur in the year following the surgery (ie, during the 1-y postoperative period). Next, based on clinical expertise, in the case of patients whose initial surgery failed, revision surgery was offered only once and was the same technique as the initial surgery. Additionally, our model did not account for crossover in surgical technique at the time of revision. This simplification may not fully capture real-world surgical decision-making and could affect cost and outcome estimates related to revision procedures. Salvage surgery was not offered based on the low revision rates and consultation with hand surgery experts (H.L.B.). Furthermore, the costs of associated complications and revisions, if applicable, were modeled as discrete events, where the cost was applied only if the patient experienced this event. A generalized postoperative cost was assumed for all patients in the postoperative state based on previous literature.⁴² This postoperative incremental cost was applied each time a patient entered the postoperative state and included costs of diagnostic testing, clinic visits, emergency department visits, and hand therapy.⁴² Finally, given the limited supporting literature on postoperative utility, the postoperative utility in this model was assumed to be halfway between the baseline utility and the average utility of a successful surgery (LRTI + T and SSA).

Model Inputs

The literature was reviewed to extract the best available evidence to populate our model with appropriate probabilities and utility values (Tables 1,^8–40 2,^6,43,44 3^42,44). (See table, Supplemental Digital Content 2, which displays the LRTI + T complication, complication requiring surgery, success, and revision surgery probabilities, https://links.lww.com/PRSGO/E386.) (See table, Supplemental Digital Content 3, which displays the SSA complication, complication requiring surgery, success, and revision surgery probabilities, https://links.lww.com/PRSGO/E387.) Costs were determined by a literature search and a review of local hospital data.

Table 2.

Health States and Utility Values Included in the Model

Health State	Utility ± SD	Reference
Painful thumb CMCJ	0.675 ± 0.0675*	Lane et al6
Postoperative period	0.7425 ± 0.07425*†	Lane et al6 Hustedt et al43
Nonpainful thumb CMCJ (LRTI + T)	0.80 ± 0.08*	Lane et al6
Nonpainful thumb CMCJ (SSA)	0.82 ± 0.08	Hustedt et al43
Complication	−0.1 ± 0.01	Retrouvey et al44
Revision	−0.1 ± 0.01	Retrouvey et al44
Dead	0	Retrouvey et al44

^*If SD was not reported, it was assumed to be 10% of the mean.⁴⁴

^†Postoperative period utility was assumed to be between the baseline utility and the average utility of a successful surgery.

Table 3.

Costs Included in the Model

Health State	Total Cost ± SD (CAD$)	Reference
LRTI + T	$2280.42 ± 677.58	Hospital cost data*
SSA	$2346.44 ± 584.37†	Hospital cost data*
Complication	$1535.84 ± 153.58‡	Billig et al42
Postoperative care	$192.06 ± 19.20‡	Billig et al42

^*Average and SD of the most recent cases performed at our tertiary-level center in Toronto, Ontario, Canada.

^†To account for the variation in SSA techniques in the cost analysis, we subtracted the cost of the specific implant used (CAD$; Arthrex FibreWire: $18.50, Tightrope: $525.00, and InternalBrace: $679.50) from the total cost of the SSA procedure at the patient level. Then, the mean cost of the implants was added to the mean total cost of the SSA procedure without the implants.

^‡SD calculated as 10% of the original cost.⁴⁴

Base Case

The base case for this model was extrapolated from a large study using registry data of patients with thumb CMCJ OA undergoing surgery.⁶ The median age of patients with thumb CMCJ OA in this registry was 67 years, with 78% of the patients being women. As per registry data, the base case of this model will be a 67-year-old female patient who experienced failed medical treatment of thumb CMCJ OA.

Probabilities

Probabilities for complications, both those necessitating surgical intervention and those not requiring it, following each surgery were obtained from individual studies in the literature and averaged to obtain a mean complication rate per procedure (Supplemental Digital Content 2, https://links.lww.com/PRSGO/E386; Supplemental Digital Content 3, https://links.lww.com/PRSGO/E387). The mean and SD were then used to generate a lognormal distribution for complication rates of each procedure. Similarly, probabilities of success and revision surgery were determined for each procedure. The probability of success for the revision surgery itself was assumed to be half of the probability of success of the initial surgery. The probability of death was modeled using updated data from Statistics Canada.⁸ All probabilities were modeled as lognormal distributions (Table 1^8–40).

Utility Values

Utility is a numerical value assigned to a health state, ranging from 0 (death) to 1 (representing perfect health), used to quantify a specific health condition.⁴⁵ The baseline utility of thumb CMCJ OA, LRTI + T, and SSA was extracted from published literature and estimated to be 0.675, 0.80, and 0.82, respectively.^6,43,44 Complications and revisions were each associated with a disutility of −0.1, obtained from previous literature.⁴⁴ Utility values were modeled using a beta distribution (Table 2^6,43,44).

Cost

Cost of procedures was obtained from a hospital from a payer perspective, representing costs incurred by a single institution that offers both procedures. The cost of each procedure is denoted in Canadian dollars and was obtained by averaging the hospital costs of cases performed at our tertiary-level care center in Toronto, Ontario, Canada. The costs are as follows: the cost of LRTI + T was $2280.42, and the cost of SSA was $2346.44. The cost of SSA incorporates the average cost of implants for Arthrex FibreWire, Tightrope, and InternalBrace. The cost of revision surgery was estimated to be the same as the initial surgery ($2280.42 for LRTI + T and $2346.44 for SSA). The cost of complications and routine postoperative care was estimated to be $1535.84 and $192.06, respectively. These were obtained from a previous nationwide analysis of costs for patients with hand OA, after accounting for inflation and currency conversion (US dollars to Canadian dollars).⁴² Costs were modeled as a beta distribution (Table 3^42,44).

Analysis

Validation

Validation included review of the model structure, assumptions, and parameters with content experts including multiple hand surgeons (H.R. and H.L.B.). Debugging was performed via 1-way sensitivity analyses of all modifiable variables across value extremes, ensuring that the model behaved as expected for the base case.

Our model was also internally validated by comparing model outputs to those reported in the published literature. (See table, Supplemental Digital Content 4, which displays the results of internal validation of the model, https://links.lww.com/PRSGO/E388.)

Cost-effective Analysis

The Markov microsimulation model was used to identify the surgical procedure that produced the highest QALYs while minimizing costs. A microsimulation of 10,000 iterations was run. Key metrics for cost-effectiveness analysis consisted of cost, QALYs, ICER, NMB, and incremental net monetary benefit (INMB):

$$\begin{array}{r}\mathrm{I}\mathrm{C}\mathrm{E}\mathrm{R}=\frac{\mathrm{\Delta }\mathrm{C}\mathrm{o}\mathrm{s}\mathrm{t}}{\mathrm{\Delta }\mathrm{Q}\mathrm{A}\mathrm{L}\mathrm{Y}},\end{array}$$

$$\begin{array}{r}\mathrm{I}\mathrm{N}\mathrm{M}\mathrm{B}=\left(\mathrm{\Delta }\mathrm{Q}\mathrm{A}\mathrm{L}\mathrm{Y}\right)\lambda -\mathrm{\Delta }\mathrm{C}\mathrm{o}\mathrm{s}\mathrm{t}.\end{array}$$

Lambda (λ) represents the cost-effectiveness (willingness to pay [WTP]) threshold, set at the commonly used threshold of $50,000 per QALY in our analysis.^7,44 The WTP threshold represents the monetary value society is willing to pay for 1 full additional QALY over a lifetime horizon.

Sensitivity Analysis

Sensitivity analysis aims to assess the impact of uncertainty in the model outputs, thereby determining the robustness of the model.⁴⁶ One-way sensitivity analyses are presented in a tornado diagram. A 2-way sensitivity analysis was also performed on the costs of initial surgery and presented in a graphical format.

RESULTS

Clinical Outcomes

The 2 strategies were compared with regard to the patient’s clinical outcomes, specifically, complications and need for revision surgery (Fig. 2). These outcomes were generated by the model using input probabilities derived from the literature (Supplemental Digital Content 2, https://links.lww.com/PRSGO/E386; Supplemental Digital Content 3, https://links.lww.com/PRSGO/E387). The model estimated that complication rates for LRTI + T and SSA are 14.56% and 9.80%, respectively. Over time, 5.65% of patients with LRTI + T required revision surgery over their lifetime, compared with 7.06% of patients with SSA.

Fig. 2.

Clinical outcomes of LRTI + T vs SSA.

Health Economic Evaluation

The lifetime expected values of effectiveness and cost for LRTI + T and SSA are shown in Table 4. Over a lifetime, SSA was found to confer an increase of 0.25 QALYs, which was associated with a cost of $13.45. The lifetime ICER for SSA was therefore $53.80 per QALY. With a WTP threshold of $50,000, SSA yielded a higher NMB than LRTI + T, resulting in an INMB of $12,486.55 over a patient’s lifetime (Fig. 3).

Table 4.

Cost-Utility Analysis Displaying Lifetime Expected Values for a 67-year-old Woman With Thumb CMCJ OA Whose Conservative Management Failed

	Cost (CAD$)	Effectiveness (QALYs)	NMB (CAD$)	ICER (CAD$/QALY)	INMB (CAD$)
LRTI + T	2842.21	13.61	677,428.58	Reference	Reference
SSA	2855.66	13.86	690335.22	53.8	12,486.55

Fig. 3.

Distributions of cost and effectiveness for LRTI + T and SSA. A, Distribution for cost. B, Distribution for effectiveness (QALYs).

Sensitivity Analysis

Numerous 1-way sensitivity analyses were performed across clinically plausible ranges of uncertainty and presented as a tornado diagram. See figure, Supplemental Digital Content 5, which displays the sensitivity analysis. A, Tornado diagram displaying 1-way sensitivity analyses of costs. B, Two-way sensitivity analysis that simultaneously varied the operating room costs for LRTI + T and SSA to assess their impact on cost-effectiveness, https://links.lww.com/PRSGO/E389. In modeling uncertainties of key model parameters (utility, cost, revision, complication, and success), the primary outcome, ICER, was most sensitive to the costs of the operation. Particularly, the cost of the operation for SSA resulted in a spread of the ICER between −3488.79 and 10,958.55. The subsequent 2-way sensitivity analyses compared the costs of the operation for LRTI + T and SSA and demonstrated that SSA remains the favored procedure (Supplemental Digital Content 5, https://links.lww.com/PRSGO/E389).

DISCUSSION

This study compared the cost-effectiveness of 2 surgical procedures (LRTI + T and SSA) for patients with thumb CMCJ OA whose conservative management failed, using a Markov microsimulation model. Our analysis found that patients with CMCJ OA benefited most from SSA compared with LRTI + T, with an ICER of $53.80 per QALY. Not surprisingly, differences in costs of the initial operations were the most influential factor on ICER based on our sensitivity analyses. As our model incorporated a range of implant costs, including higher cost SSA techniques, it is likely that a lower cost, suture-only SSA approach would have further improved the cost-effectiveness of SSA by reducing the ICER even more. Our conclusions are robust, as demonstrated by our sensitivity analyses. Of note, the incremental difference in QALYs between treatment strategies was small (0.25), a characteristic often observed in decision analysis studies related to hand surgery.⁴⁴ Previous studies have explained this via the relatively high utilities seen in hand surgery compared with other clinical scenarios.⁴⁴

In the presented model, SSA has a lower rate of complication compared with LRTI + T (9.14% versus 14.56%, respectively). Lower complication rates translate to reduced patient morbidity and potentially lower costs related to managing complications, including unplanned follow-up care such as emergency visits, additional clinic assessments, or procedures. In a prospective cohort study, Shonuga et al⁹ compared SSA versus LRTI + T and found no difference in complication rates during a 1-year period. In contrast, in a randomized controlled trial by Morais et al,³³ during a 40-month follow-up period, the SSA group had a higher rate of complication compared with the LRTI + T group (11% versus 8%, respectively). This discrepancy demonstrates that complication rates can vary depending on the follow-up duration. However, our study shows that over a lifetime horizon, SSA has a lower rate of complications. Second, although the need for revision surgery over the lifetime is slightly higher for SSA compared with LRTI + T (7.06% versus 5.65%, respectively), this difference is relatively small. The decision to undergo revision surgery may also be influenced by factors such as patient preferences, surgeon expertise, and the severity of the underlying condition, rather than solely being indicative of surgical failure. This may be a conservative estimation, because many studies of both surgical strategies report a 0% revision rate over a long follow-up duration (Supplemental Digital Content 2, https://links.lww.com/PRSGO/E386; Supplemental Digital Content 3, https://links.lww.com/PRSGO/E387). These findings underscore the significance of factoring in long-term outcomes when assessing the efficacy of surgical interventions in thumb CMCJ OA, given its chronic nature.

Our study builds upon existing research that primarily compared operative versus nonoperative management for thumb CMCJ OA. A prospective cost-utility analysis of 151 patients compared the societal perspective of nonoperative and surgical management of thumb CMCJ OA based on EuroQol- 5 dimension, direct medical costs, and productivity losses.¹ Compared with nonoperative management, the ICER for operative management was US $25,370 per QALY, rendering it a cost-effective strategy.¹ Similarly, another study compared nonoperative management, trapeziectomy alone, and LRTI + T from a societal perspective.⁴⁷ They reported that both trapeziectomy alone and LRTI + T were cost-effective compared with nonoperative management (ICER: $2540 and $3511, respectively).⁴⁷ Although trapeziectomy alone eliminates painful, arthritic bony apposition, LRTI is thought to prevent thumb metacarpal subsidence and improve postoperative pinch strength.^11,48 As such, our study is unique in that it compares a relatively new surgical technique to the treatment choice of most American Society for Surgery of the Hand members,⁵ thus building on existing work by delineating the cost-effective surgical strategy.

LIMITATIONS

The results of our study should be interpreted in the context of several limitations. First, the utilities of the study were largely derived from the UK Hand Registry, which may not reflect other countries’ preferences. Next, surgical costs were based on microcase costing data from a single center within Canada’s single-payer system, limiting generalizability to other jurisdictions with differing reimbursement models, labor costs, and implant pricing. Moreover, the model groups various surgical techniques under the umbrella of SSA, averaging complication, success, revision rates, and implant costs. Our institutional cost data include a mix of FibreWire, TightRope, and InternalBrace implants. The inclusion of higher cost implants such as TightRope and InternalBrace may have increased the average SSA cost and could limit generalizability to centers using lower cost techniques. Similarly, variation in how LRTI + T is performed (eg, hemi- versus full trapeziectomy, different tendon grafts) was not accounted for. Eaton stage, a key prognostic factor, was also not included. These sources of heterogeneity may influence complication rates, functional outcomes, and cost, and their omission limits the ability to draw conclusions about the relative cost-effectiveness of specific LRTI techniques or how outcomes may differ based on disease severity at baseline. Hospital cost variations, such as operative time, remain unknown. Despite these limitations, our model is robust, incorporating both Markov modeling and microsimulation to capture clinical complexity and patient-level heterogeneity. It accounts for failed surgery and revision, and uniquely differentiates complications requiring surgical versus nonoperative management. This distinction is important, as, although SSA may have fewer overall complications, a higher proportion may require surgery compared with LRTI + T.

CONCLUSIONS

Suture SSA is a cost-effective alternative to trapeziectomy with ligament reconstruction and tendon interposition for thumb CMCJ OA. Although our findings support its use, treatment decisions should reflect patient and surgeon preferences. Future work should assess the long-term quality of life and societal cost-effectiveness of suture SSA.

DISCLOSURE

The authors have no financial interest to declare in relation to the content of this article.