Similar cost-utility for double- and single-bundle techniques in ACL reconstruction

Purpose The aim was to estimate the cost-utility of the DB technique (n = 53) compared with the SB (n = 50) technique 2 years after ACL reconstruction. Methods One hundred and five patients with an ACL injury were randomised to either the Double-bundle (DB) or the Single-bundle (SB) technique. One hundred and three patients (SBG n = 50, DBG n = 53) attended the 2-year follow-up examination. The mean age was 27.5 (8.4) years in the SBG and 30.1 (9.1) years in the DBG. The cost per quality-adjusted life years (QALYs) was used as the primary outcome. Direct costs were the cost of health care, in this case outpatient procedures. Indirect costs are costs related to reduce work ability for health reasons. The cost-utility analysis was measured in terms of QALY gained. Results The groups were comparable in terms of clinical outcome. Operating room time was statistically significantly longer in the DBG (p = 0.001), making the direct costs statistically significantly higher in the DBG (p = 0.005). There was no significant difference in QALYs between groups. In the cost-effectiveness plane, the mean difference in costs and QALYs from the trial data using 1000 bootstrap replicates in order to visualise the uncertainty associated with the mean incremental cost-effectiveness ratio (ICER) estimate showed that the ICERs were spread out over all quadrants. The cost-effectiveness acceptability curve showed that there was a 50% probability of the DB being cost-effective at a threshold of Euro 50,000. Conclusion The principal findings are that the DB is more expensive from a health-care perspective. This suggests that the physician may choose individualised treatment to match the patients’ expectations and requirements.


Introduction
Anterior cruciate ligament (ACL) rupture is an injury that is mostly related to sports, especially contact sports. However, the consequences affect not only the person's opportunity to continue with his/her sporting activities, perhaps resulting in a future career being ruined. It also affects the whole life, as several professions, such as heavy engineering work and construction work, require stable knees. At a time when health-care costs and the effectiveness of health treatments are major topics for discussion, every orthopaedic department has to consider the most cost-effective surgery in order to prioritise and provide the maximum additional effects per additional unit of resource consumed. According to the Swedish ACL Register, an injury frequency of approximately 80 per 100,000 inhabitants in Sweden would mean that some 5800 individuals suffer anterior cruciate ligament injuries every year and that some 3500 undergo surgery [16].
Finding the optimal surgical technique is of great interest to every anterior cruciate ligament (ACL) surgeon. In the last decade, the double-bundle (DB) technique was introduced by restoring both the antero-medial and posterolateral bundles.
The question of whether the DB technique is able to restore rotational instability to a higher degree compared with the single-bundle (SB) technique is still the subject of debate. The findings in recent systematic reviews are in favour of the DB technique in terms of rotational laxity and IKDC grade [22,37,38].
However, the DB technique is also more demanding and time-consuming and requires two fixation devices, thereby increasing the cost of the whole procedure. According to a meta-analysis, treatment with the DB technique is in favour in terms of knee joint stability and knee joint function [27] but, at least from a health-care perspective, more costly than the SB technique. This is in line with Paxton et al., who concluded that the double-bundle technique may be costeffective in terms of IKDC grade [33]. In 2009, Brophy et al. [9] compared the cost of the SB and DB techniques and concluded that DB added considerable costs to the health-care system with small clinical benefits to patients. Comparing health-related quality of life, Nunez et al. [31] reported no significant difference between the SB and DB techniques, but concluded that the SB technique was more cost-effective in terms of knee joint stability and knee joint function.
Another aspect that should be considered when comparing the two techniques is whether the new technique could shorten sick leave-in other words, whether the new technique will offer a more stable knee earlier in the rehabilitation process, resulting in a shorter period of sick leave and thereby reducing the societal costs [18]. The social insurance system in Sweden is financed through social insurance contributions, and the individual has a legal right to compensation due to lost income (http://www.riksdagen. se/sv/Dokument-Lagar/Lagar/Svenskforfattningssamling/ Socialforsakringsbalk-201011_sfs-2010-110/).
The available knowledge could form the basis of a costutility analysis (CUA), where the expected benefits are related to the costs. The result of a CUA is presented as cost per quality-adjusted life year (QALY). The CUA result can be used by decision-makers when deciding which healthcare interventions should be made available in collectively funded health-care systems [7].
Clinical research often involves comparing the clinical outcome of the new and old technique. However, it is important to relate the efficacy of the new treatment to the costs [10], and many studies evaluate the cost-effectiveness from a health-care perspective (only including costs of health care); in the present study a societal perspective was used, including all costs involved after an ACL reconstruction.
The aim of the present study was to estimate the costutility of the DB technique when compared with the SB technique in patients after ACL reconstruction. The hypothesis was that the choice of surgical technique affects the cost-utility.

Materials and methods
Data for the economic evaluation were collected alongside a randomised controlled study comparing the DB and SB techniques in 105 patients with ACL injuries 2 years after surgery. A CUA was conducted based on a comparison of the two surgical techniques. Prevalence-based direct and indirect costs were calculated from a societal perspective and compared to quality-adjusted life years estimated using the Euro QoL group five dimensions questionnaire for health-related quality of life (EQ-5D).

Design
The study was conducted between 2008 and 2009 at two different hospitals in the western part of Sweden (Fig. 1). Patient-specific resources used and preference-based health status data were collected on an individual basis. A complete description of the study design and results of the clinical intervention have been published previously [1].
The cost difference was calculated by subtracting the mean cost of DB from the mean cost of SB. The cost per quality-adjusted life year (QALYs) was used as the primary outcome. The difference in QALYs was calculated by subtracting the mean QALYs for DB from the mean QALYs for SB. The economic evaluation was performed from a societal perspective.

Study cohort
The inclusion criterion was unilateral ACL injury in subjects over 18 years of age. The exclusion criteria were posterior cruciate ligament injury, more than 1+ medial or lateral collateral ligament laxity after injury or major previous knee surgery. Patients who fulfilled the inclusion criterion were consecutively asked to participate in the study. The participants were randomised to either the DB or the SB technique using closed envelopes administered by the same study coordinator for both hospitals.
At the 2-year follow-up, 103 patients answered the EQ-5D questionnaire on health-related quality of life. The singlebundle group (SBG) consisted of 50 patients, 35  The overall clinical outcome presented in Ahldén et al. [1] was that the clinical assessments at follow-up revealed no significant differences between the DB and SB groups in terms of knee laxity measurements, range of motion, Lysholm knee scoring scale, Tegner activity scale, Knee Injury and Osteoarthritis and Outcome Score (KOOS) or functional tests. A significant improvement was seen in both groups compared with the pre-operative values in terms of most clinical assessments.

Surgical method
Standard antero-lateral and antero-medial portals were established. Associated intra-articular injuries, such as meniscal injuries and chondral lesions, were addressed at the time of the index operation.

Follow-Up
All paƟents at the clinics of the parƟcipaƟng surgeons were assessed for eligibility Randomised (n=105)

3
The surgical procedure for both the single-and doublebundle techniques is described in detail in the article by Ahldén et al. [1].

Post-operative procedure
All patients had their surgery performed in an outpatient procedure, and none had to stay overnight because of complications.
The general recommendation was to refrain from work for 6 weeks, but people with a physically light job were able to return to work after 2 weeks.
In the present study, post-operative appointments were made at the orthopaedic ward after 4-6 weeks, seeing only the surgeon, and at 6 and 24 months after surgery, seeing both the surgeon and the physiotherapist. The clinical examination included range of motion (ROM), knee laxity measurements, functional tests commonly used to evaluate ACLreconstructed patients, as well as radiographic examination. The routine is one post-operative visit to the surgeon after 4-6 weeks and one to the physiotherapist after 6 months, but, as these patients were included in a randomised clinical trial study (RCT) comparing the two surgical methods, visits to the surgeon at six and 24 months were included, as well as the X-ray examination.

Post-operative rehabilitation programme
A referral was given to the patients when leaving the hospital with the recommendation to contact their local physiotherapist and start rehabilitation seven to 10 days post-operatively [1]. All the patients were rehabilitated according to the same guidelines.

Data extraction
All patient reports, as well as surgical reports, were reviewed by two persons independently. Disagreement or uncertainty between two reviewers was resolved by consensus or after discussion with a third party, one of the surgeons.

Outcome measures
Health-related quality of life and quality-adjusted life years EQ-5D The EQ-5D is a generic instrument for measuring overall health-related quality of life based on five dimensions (mobility, self-care, usual activities, pain/discomfort and depression/anxiety) and including three levels (none, moderate and severe problems) of answers and a rating scale. In the present study, the UK tariff values of scale between −0.59 and 1 have been used. These social tariff values (elicited health state valuations from a representative sample of the UK health population) have previously been published [12]. A summary index between 1 (full health), 0 (death) and values below 0 corresponding to worse than death has been calculated [8,34]. A difference of 0.07 or more is regarded as clinically relevant [39]. It was intended to be used for economic analysis (cost-utility analysis) and to make it possible to calculate the cost per QALY [8]. Brazier et al. [6] evaluated the EQ-5D in a group of patients with osteoarthritis of the knee and concluded that it could be used for economic evaluations of surgery.

QALY
The quality of life weighting was then used to calculate QALYs. This measurement combines years of life with quality of life, so that the QALY, as a result of a treatment, can consist of increasing life expectancy and/or increased quality of life. QALY calculations (here using the EQ-5D) were made at individual level, reflecting the change from baseline to 2 years, assuming a linear change in QoL between the two measurements. QALY was calculated by using the areaunder-the-curve approach.

Costs
The cost in the current study consisted of direct and indirect costs (production loss). The direct costs were calculated from resources used in specialised outpatient care. These costs included all fixed costs, i.e. administration and other overhead costs, staff salaries and accommodation at recovery, as well as individual costs, such as examination, surgery (operating room (OpR), anaesthesia and material), post-operative visits, rehabilitation, laboratory tests or imaging. All costs were collected from the hospitals' administrative records registered at individual level (e.g. hospitalisation), including fixed cost for the hospital stay and specific resources used during the treatment. The indirect costs are costs related to reduce work ability for health reasons. The human capital method was used in this study, estimating the potential production loss, based on the assumption that earnings reflect productivity [13]. Using this method, the daily production loss on an individual basis was calculated and consisted of each individual´s number of days of sick leave, obtained from the National Social Insurance Board's (NSIB) central register (http://www.forsakringskassan.se/privatpers). All sick leave periods longer than 14 days and all disability pension payments are administered and registered by the NSIB. Statistics Sweden (http://www. scb.se/Pages/List____250619.aspx) supplied the information relating to the national mean monthly salary for males and females presented for defined professions. As 15% and 24%, respectively, were students, the imputation of missing data for indirect costs was performed. Furthermore, the average salary (for defined professions among the study participants) from Statistics Sweden (http://www.scb.se) and payroll taxes (SEK 1334 per day of absence) were used to calculate the cost of productivity losses.
Other costs associated with cruciate ligament damage, such as travelling to and from treatment, are not valued.
All the costs were obtained and analysed retrospectively. The costs are presented in euros, using the 2010 average exchange rate (Euro 1 = SEK 9.55).

Cost-utility analysis
Cost-utility analysis (CUA) refers to a particular form of cost-effectiveness analysis where the outcome is measured in terms of QALY gained. This can be compared with the mean total cost (direct + indirect costs) of the programme to determine the cost per QALY gained.
In this study, both QALY and costs are discounted by 3%.
The ICER can be interpreted as the extra cost of obtaining one extra QALY and allows comparisons between interventions in all areas of health care, not necessarily comparable programmes [10]. The results of the analysis are able to demonstrate whether the DB technique yields higher or reduced costs, reduced or improved utility in comparison with the SB technique and are presented in a so-called costeffectiveness plane [13].

ICER uncertainty
The cost-effectiveness plane was used to visualise the uncertainty surrounding the mean ICER point estimate. The plane displays the incremental costs on the y-axis and the incremental effects on the x-axis and is divided into four quadrants: north-east (NE), south-east (SE), south-west (SW) and north-west (NW) [15]. If the ICERs are plotted in the NE quadrant, the DB technique yields higher costs and an improved effect and a willingness-topay (WTP) threshold would therefore be necessary, if the DB technique were implemented. Conversely, if ICERs are plotted in the SE quadrant, the DB technique both saves costs and has an improved effect; the DB technique is therefore the dominant strategy. If the ICERs are located in the SW quadrant, the DB technique reduces costs, but the effect decreases. Finally, if the ICERs are placed in the NW quadrant, the DB technique is both more expensive and reduces the effect, which explains why the existing treatment, the SB technique, clearly dominates. The cost-effectiveness acceptability curve (CEAC) is constructed so that it shows the proportion of simulations that are deemed cost-effective, given different threshold values for cost-effectiveness. It is the proportion of simulations to the right of a budget line with the slope corresponding to the threshold value in the figures. The CEAC illustrates the probability that the DB technique is cost-effective compared with the SB technique at different willingness-to-pay thresholds [13]. In this analysis, the maximum willingness to pay was set at a level of Euro 50,000 (SEK 500,000), based on the Swedish National Board of Health and Welfare (https://www.socialstyrelsen.se/SiteCollectionDocuments/metodbeskrivningnationella-riktlinjer.pdf).

Sensitivity analysis
To assess the strength of the result, an average salary from Statistics Sweden (http://www.scb.se) and payroll taxes were used to estimate the cost of productivity loss (SEK 1400 (=Euro 147) per day of absence).
As a sensitivity analysis, both costs and effects are discounted at 5 and 0% (no discount) and costs are discounted at 3%, but no discount is made for effects. A health-care perspective using the direct costs is also presented.

Statistical analysis
Sample size calculation was based on the pivot shift test with one degree of difference, a power of 0.85 and a level of significance of 0.05, which gives 38 patients in each group.
Mean (SD) and median (range) values are presented when applicable. For comparisons of dichotomous variables between the groups, the Chi-square test was used. In terms of comparison between the groups of continuous variables, a t test was used, while for discrete variables the Mann-Whitney U test was used. A p-value of ≤ 0.05 was considered statistically significant. Effect size was calculated as the difference between pre-operative values and post-operative values.
By using bootstrapping, we wanted to assess the statistical uncertainty associated with the ICER. We simulate new ICERs as if they were single mean estimates from the trial by building up an empirical sampling distribution (based on the original data) in order to make inferences about the ICER. The number of units is the same as the number of patients in the analysis and the CI is based on 1000 such samplings. The bootstrapping analysis was performed using the SAS 9.2 computer program, Cary, NC.

Ethics
Eligible patients were informed about the study and gave their signed informed consent. The Regional Ethical Review Boards at the University of Gothenburg approved the study, (2011-07-25) Dnr 162-11, and permission was also obtained from the heads of the individual clinics.

Results
Pre-operatively, the groups were comparable in terms of demographic data (Table 1).
Descriptive data relating to the total intervention procedure are presented in Tables 2 and 3. The sensitivity analysis is presented in Table 4.
The groups were comparable pre-operatively and at follow-up in terms of HRQoL ( Table 5).
The ICER points (based on the average salary for defined professions among the study participants) were distributed in each of the four quadrants, where 20.2% were located in the NW quadrant, 34.9% in the SW quadrant, 38.0% in the SE quadrant and 6.9% in the NE quadrant. In other words, the ICERS are spread out over all the quadrants (Fig. 2a).
The CEAC (Fig. 3) is based on the average salary for defined professions and means that there is a 50% probability of the DB technique being cost-effective at a willingness-topay threshold of Euro 50,000 with great uncertainty in the ICER estimates.

Sensitivity analysis
The ICER point (based on average salary for defined professions) was distributed in each of the four quadrants, where 20.2% was located in the NW quadrant, 34.9% in the SW quadrant, 38.0% in the SE quadrant and 6.9% in the NE quadrant (Fig. 2a).
The CEAC (Fig. 3) based on average salary for defined professions there is a 50 probability of DBG being costeffective at a threshold of willingness to pay of Euro 53,000, however with a great uncertainty in the ICER estimates.
The total mean cost for DBG and SBG, when using the average salary from Statistics Sweden, was Euro 13,718 and Euro 14,913, respectively (Table 4), which is slightly higher costs in comparison with the national mean monthly salary for males and females presented for defined professions ( Table 3). The ICERs based on average salary (Fig. 2b) showed that 40.6% was located in the SW, 40.1% in the SE, 4.3% in the NW and 15% in the NE quadrant. Here the CEAC showed that around 55 probability of DB being costeffective at a willingness-to-pay threshold of Euro 53,000.
The discount rate 0, 3 and 5% for costs and QALYs is presented in Table 6.

Discussion
The principal findings in the present study are that the double-bundle technique is more expensive, as it takes longer in the operating room and requires more materials (direct costs p = 0.005), which means that the DB is more expensive from a health-care perspective. However, from a societal perspective there was no statistically significant difference between the groups receiving SBG and DBG, due to lower costs for production loss among patients in the DB group. This was confirmed when analysing both average salary for defined professions among the study participants and for national mean monthly salary (Fig. 2a, b). There was no difference between the groups in terms of QALYs.

Single-bundle versus double-bundle technique
In a recent systematic review by Mascarenhas et al. [29] analysing nine meta-analyses, the conclusion was that the doublebundle technique provided better post-operative knee stability, as measured with the KT 1000 arthrometer and pivot shift. This is also in line with the present study [1]. The 5-year follow-up of the present study has recently been published, and no statistical differences could be found between the groups in terms of clinical outcome or the presence of osteoarthritis [24]. However, the development of OA may take longer than 5 years to be radiographically visible. From a societal perspective, the future costs may include the consequences of the development of OA, such as sick leave and the need for a total knee arthroplasty. Today, according to the Swedish ACL Register, the use of the DB technique accounted for fewer than 1% of all the operations performed in 2015 [17]. The main reason from the health-care perspective for not implementing the DB technique is higher direct costs, with a clinical outcome comparable to that produced by the SB technique. In a meta-analysis, Li et al. [28] pooled data from seven studies and found no evidence of a significant difference in complication rate between the single-and double-bundle techniques. This is in line with the present study. No complications were registered in 90% of the patients, both groups, respectively. The low complication rate is also seen in the post-operative visits and consulting phone calls to the surgeon or nurse on the orthopaedic ward within standard care. No re-ruptures or revision surgery was registered during the follow-up period in the present study [1].

Direct costs
Orthopaedics, especially sporting injuries, are often given as an explanation of the increased costs within the health-care system and orthopaedic surgeons have to justify their choice of procedure [5,20]. These costs should be compared with the reduction in hospital stay or outpatient surgery and earlier, accelerated rehabilitation. This results in an improved functional outcome for the patient and hopefully a smaller burden on society [11].
The cost of ACL reconstructions, as well as other orthopaedic sports medicine injuries, is part of the entire     [26] found no statistically significant difference in economic value in terms of early ACL reconstruction compared with rehabilitation and ACL reconstruction when needed. They may improve the outcome, but at greater expense and with higher administrative costs, plus lapses in quality [2]. An ACL injury is a serious knee injury, and it is more common in young athletes performing highly demanding sports.  [33,35]. According to the ACL Register, in 2014, only 28 ACL reconstructions of 3430 were performed using the double-bundle technique [16].

Production loss (indirect costs)
The total days of sick leave differed between the two groups, 80 days in the SBG and 65 days in the DBG, but this difference was not statistically significant. The median difference of 15 min in time in the operating room may not result in post-operative discomfort and longer sick leave (production loss). The difference in sick leave could reflect differences in occupation between the two groups, prolonging the return to work, irrespective of the type of reconstruction technique used. There were, for example, more self-employed people in the DBG (30 vs 14%), which could suggest a greater eagerness to resume work. The larger number of days off work in the SBG might, however, randomly also reflect the fact that working ability and sick leave are highly subjective measurements with a large impact on costs in a study like this [21]. There are missing data on sick leave in 22 patients (13 in the SBG and nine in the DBG), and this might be due to unemployment, being student or being left out of the insurance system for other reasons. However, imputed data were used in the analysis. In this study, only post-operative sick leave periods longer than 14 days and all payments are included. The difference in health-care organisation and insurance system between countries must also be taken into account.
In recent years, focusing on rehabilitation programmes and the criteria for knee function has been a topic of great interest, as the return to competitive sport is relatively low [23]. However, the primary objective, after all, is to be able to return to work. Even though the individual has a legal right to compensation due to lost income through the Swedish social insurance system, this does not cover the whole loss of income. Heavy industrial workers and craftsmen have to regain full muscle strength, as well as health-related quality of life.
The health-related quality of life result, as measured with the EQ-5D, was in line with data from the Swedish National ACL database, comparing the single-bundle and doublebundle technique [4], but lower compared with the general Swedish population [36].
As the DB trial was primarily designed to detect a difference of one degree in the pivot shift test, the number of enrolled subjects may have been too low to detect a difference in QALY during the trial period. The baseline analysis conducted prior to the CUA did not detect any statistically significant differences in baseline utility. Further, the collection of QALYs in the DB trial could have been improved, as the EQ-5D score was only measured at baseline and at the 2-year follow-up. In the light of the fact that patients after ACL reconstruction are not re-admitted to the hospital due to re-injury to the ACL, it was not necessary to include further EQ-5D score measurements in the trial and the follow-up period, as this would not have increased the accuracy of the results.

QALY
In the present study, the QALY for the single-bundle group and double-bundle group was 1.44 and 1.43, respectively. This means that there was no gain in QALYs for the DB group (−0.01). Performing a cost-effectiveness comparison of different types of graft, Genuario et al. [19] reported a QALY of between 0.904 and 0.912, while Farshad et al. [14] compared conservative treatment with ACL reconstruction and reported a QALY of 0.66.
The gain in QALY is also higher (0.212 and 0.201, respectively) compared with the cost-utility analysis The cost-effectiveness plane, where the 1000 bootstrap replicates of the ICER were spread out of over the quadrants, demonstrated that the DB cost almost as much as the SB and had the same effect as the SB (Fig. 2). The sensitivity analysis revealed marginal/no differences between the discount rates of 0 and 5% for costs and QALYs. From a health-care perspective, the DB technique cannot be recommended. However, from the patients' perspective, the physician should consider individual treatment based on who the patient is and what his/ her expectations and requirements are [25]. Other aspects to take into consideration are the preferences, experiences and skill of the orthopaedic surgeons when choosing the technique.

Strengths and limitations
One strength of the present study is the randomised design and the follow-up rate of 98%. The clinical results are also in line with those in previous studies. It is common for the sample size of a trial to be based on the primary clinical outcome, and in this study, the sample size was calculated on the difference in pivot shift and not on differences in health-related quality of life. This is a major limitation, as, in a calculation based on the difference in sick leave or total costs, the power will be low. However, comparing costs, as in the present study, takes place alongside an analysis based on a clinical randomised trial. Moreover, the change in the EQ-5D over time is based on two measurement points (preoperatively and at the 2-year follow-up), assuming a linear function using an area-under-the-curve approach. The first 2 weeks of sick leave are paid by the employer, while sick leave periods longer than 14 days are administered and registered by the NSIB. As the recommended period of sick leave after an ACL reconstruction is at least 6 weeks, the total costs including the employers' contribution may have been higher, but the difference between groups would be the same. Another limitation is that the direct costs at a university hospital may have been higher than those at a county hospital. There was, however, no intention to compare the two hospitals. The fact that the cost analyses are dependent on the health-care cost and policy of the specific country, thereby limiting the external validity, must also be taken into consideration.

Clinical perspective
The results of the present study give the physician the possibility to choose an individualised treatment based on age, gender, symptoms and the size of the native ACL insertion site, but also have a detailed discussion with the patient relating to expectations and requirements [25], such as rehabilitation intensity and demands relating to activity level, in Fig. 3 Cost-effectiveness acceptability curve (CEAC), based on average salary of included professions. The red dotted line shows that the maximum willingness to pay was set at a level of €50,000 (SEK 500,000) based on the Swedish National Board of Health Care  [3]. In the long term, this will hopefully be more cost-effective for both the patient and society.

Conclusion
The double-bundle technique is more expensive than the single-bundle technique, as it takes longer in the operating room and requires more materials. On the other hand, no difference was found in total costs, due to lower costs for production losses among patients in the DB group. Furthermore, no difference in QALYs was seen between the groups.