Article Text

Download PDFPDF

No economic benefit of early knee reconstruction over optional delayed reconstruction for ACL tears: registry enriched randomised controlled trial data
  1. Aliasghar A Kiadaliri1,2,
  2. Martin Englund1,3,
  3. L Stefan Lohmander4,5,6,
  4. Katarina Steen Carlsson7,8,
  5. Richard B Frobell4
  1. 1Clinical Epidemiology Unit, Orthopaedics, Department of Clinical Sciences Lund, Lund University, Lund, Sweden
  2. 2Health Services Management Research Center, Institute for Futures Studies in Health, Kerman University of Medical Sciences, Kerman, Iran
  3. 3Clinical Epidemiology Research and Training Unit, Boston University School of Medicine, Boston, Massachusetts, USA
  4. 4Orthopaedics, Department of Clinical Sciences Lund, Lund University, Lund, Sweden
  5. 5Research Unit for Musculoskeletal Function and Physiotherapy, University of Southern Denmark, Odense, Denmark
  6. 6Department of Orthopedics and Traumatology, University of Southern Denmark, Odense, Denmark
  7. 7Department of Clinical Sciences, Malmö, Lund University, Lund, Sweden
  8. 8The Swedish Institute of Health Economics, Lund, Sweden
  1. Correspondence to Dr Aliasghar A Kiadaliri, Clinical Epidemiology Unit, Skåne University Hospital, Klinikgatan 22, Lund SE-221 85, Sweden; aliasghar.ahmad_kiadaliri{at}med.lu.se

Abstract

Background To analyse 5-year cost-effectiveness of early versus optional delayed acute anterior cruciate ligament (ACL) reconstruction.

Methods 121 young, active adults with acute ACL injury to a previously uninjured knee were randomised to early ACL reconstruction (n=62, within 10 weeks of injury) or optional delayed ACL reconstruction (n=59; 30 with ACL reconstruction within 6–55 months); all patients received similar structured rehabilitation. Real life data on health care utilisation and sick leave were obtained from regional and national registers. Costs and quality-adjusted life years (QALYs) were discounted at 3%. Full-analysis set (based on study randomisation) and as-treated analysis (according to actual treatment over 5 years) principles were applied.

Results Mean cost of early ACL reconstruction was €4695 higher than optional delayed ACL reconstruction (p=0.19) and provided an additional 0.13 QALYs (p=0.11). Full-analysis set showed incremental net benefit of early versus optional delayed ACL reconstruction was not statistically significantly different from zero at any level. As-treated analysis showed that costs for rehabilitation alone were €13 650 less than early ACL reconstruction (p<0.001). Results were robust to sensitivity analyses.

Conclusions In young active adults with acute ACL injury, a strategy of early ACL reconstruction did not provide extra economic value over a strategy of optional delayed ACL reconstruction over a 5-year period. Results from this and previous reports of the KANON-trial imply that early identification of individuals who would benefit from either early ACL reconstruction or rehabilitation alone might reduce resource consumption and decrease risk of unnecessary overtreatment.

Trial registration ISRCTN84752559.

  • Economics
  • Effectiveness
  • Rehabilitation
  • Knee ACL
  • Randomised controlled trial

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Acute anterior cruciate ligament (ACL) rupture is a common and serious knee injury in young active people, with approximately 250 000 new injuries per year in the USA alone.1 Treatment of these injuries is debated and a randomised controlled trial (RCT), the KANON-trial, failed to identify a significant difference in patient-reported knee function, general health status, activity level, return to pre-injury activity level (2 and 5 years), and radiographic osteoarthritis (5 years) between a strategy of structured rehabilitation plus early ACL reconstruction or the same structured rehabilitation with the option of having ACL reconstruction later if needed.2 ,3

The economic impact of ACL tears and the cost-effectiveness of treatments for this group of patients have gained interest and the results of this RCT have been used in at least two recent model-based health economic analyses.4 ,5 Unfortunately, these reports were based on misinterpreted data from the KANON-trial and relied heavily on expert opinions on the future risk of developing osteoarthritis rather than the actual 5-year radiographic osteoarthritis outcome.3 We are not aware of any high-level studies supporting an advantage of ACL reconstruction over non-surgical management with regard to the risk of osteoarthritis development. On the contrary, a recent meta-analysis6 and well conducted observational and registry studies7 ,8 have failed to find a difference in the development of osteoarthritis between surgical and non-surgical treatment, similar to our 5-year follow-up in the KANON-trial.3

Nevertheless, the absence of differences in clinical and radiographic outcomes is not sufficient for a decisive recommendation; patient quality of life, loss of productivity, and health resource use are other important aspects. We present an exploratory cost-effectiveness analysis based on the results of the careful longitudinal monitoring of the first 5 years of our RCT on the treatment of ACL rupture in young active adults, combined with real life data from registers on trial participant health care consultations and interventions as well as sick leave. Availability of real life data on consumption of health and non-health resources not only increases the generalisability of our findings but avoids prone-to-bias assumptions in the calculation of costs.

Method

Study design and participants

The KANON-trial (Current Controlled Trials ISRCTN84752559) enrolled active adults, 18–35 years of age, with an acute ACL injury to a previously uninjured knee presented at the departments of orthopaedics at Skåne University Hospital Lund and Helsingborg Hospital, Sweden.2 ,3 Major exclusion criteria were professional athletes (10 on the Tegner activity score;9 0–10, lowest to highest activity level), individuals who were less than moderately active (<5 on the same scale), a total collateral ligament rupture, a full thickness cartilage lesion visualised on MRI, and extensive meniscal fixation (see Frobell et al2 ,3 ,10 for more details). One hundred and twenty-one patients were randomly assigned to undergo either structured rehabilitation plus early ACL reconstruction within 10 weeks of injury (n=62, hereon referred to as the early reconstruction group), or structured rehabilitation with the option of delayed ACL reconstruction for those with symptomatic knee instability who met specific protocol guidelines (n=59, hereon referred to as the optional delayed reconstruction group). Twenty-three (39%) of the latter group had a delayed ACL reconstruction over the first 2 years2 and an additional seven had a similar procedure over the following 3 years; in total, 30 of the 59 patients (51%) in the optional delayed reconstruction group had a delayed ACL reconstruction over the entire 5-year follow-up period.3 One patient, assigned to early reconstruction, did not attend the 5-year clinical visit and was lost to follow-up; thus in the full analysis set, 61 individuals presented in the early reconstruction group and 59 patients in the optional delayed reconstruction group (table 1).

Table 1

Baseline characteristics of the study participants

Two patients assigned to early reconstruction did not have reconstruction or had fewer than 10 rehabilitation visits and were excluded from the as-treated analysis, resulting in 59, 30, and 29 individuals in the early reconstruction, delayed reconstruction, and rehabilitation alone groups, respectively.3 Seven patients had two ACL reconstructions performed over the follow-up period (two revisions of the index knee and five in the contralateral knee): five in the early reconstruction group and two in the delayed reconstruction group.

Costs

In the base case, costs were estimated from the societal perspective and measured in 2011 Swedish SEK (1 SEK = €0.111). Health resource consumption including operations, inpatient and outpatient visits (including rehabilitation visits and laboratory tests) and medications for every participant in the RCT were retrieved from the Skåne Healthcare Register (SHR) in Sweden. The SHR comprises routinely collected prospectively ascertained data on all health care at the individual level within the Skåne region. Prices for operations, and inpatient and outpatient visits, were based on the diagnostic codes (International Statistical Classification of Diseases and Related Health Problems 10th revision (ICD-10)) and the diagnosis related groups (DRGs) in Sweden for 2011. These data were extracted from the KPP database (https://stat.skl.se/kpp/index.htm) for operations and from Södra-regionvårdsnämnden (http://www.skane.se/sv/Webbplatser/Sodra-regionvardsnamnden/PriserAvtal/) for inpatient and outpatient visits. Prices for medications were obtained through the Dental and Pharmaceutical Benefits Agency in Sweden (http://www.tlv.se). These prices were multiplied by health resource consumption to calculate health care costs. We measured and valued all costs, knee-related and unrelated, during the 5-year follow up. In the base case, a patient who underwent more than one procedure during the same operation (eg, cruciate ligament reconstruction and meniscus surgery) received costs only for the main procedure (eg, cruciate ligament reconstruction).

Productivity loss was estimated using the human capital approach and the 2011 national average wage. Days of work absence for every trial participant were obtained from the Social Insurance Register administered by the Swedish Social Insurance Agency (SSIA) (http://www.forsakringskassan.se). All sick leave periods longer than 14 days and all disability pension payments are administered and registered by the SSIA. The cost of productivity loss was based on average wages from Statistics Sweden (http://www.scb.se) and payroll taxes (SEK 1 332 per day of absence).

All costs were aggregated across all participants in each treatment arm to obtain a mean cost per participant. No costs related to the research and/or the conducting of the trial were included.

Outcomes

The 36-Item Short-Form Health Survey (SF-36) data were collected at baseline, 3, 6, 12, 24, and 60 months after randomisation.2 ,3 A preference-based measure of health was derived from the SF-36 providing a six-dimensional health state classification (SF-6D).11 ,12 SF-6D values (range worst to best as 0.29–1) were plotted against time and quality-adjusted life years (QALYs) were calculated as the area under the curve.13 SF-6D scores were missing for two subjects at 3 months, one subject at 6 months, one subject at 12 months, and one subject at 24 months. Due to the few missing values, we used last observation carried forward for imputation.

Cost-effectiveness analysis

A within-trial stochastic cost-effectiveness analysis was conducted for the 5-year trial period using full-analysis set (primary) and as-treated analysis (secondary) principles. In the full-analysis set, the subjects were distributed across two groups (early ACL reconstruction, and optional delayed ACL reconstruction) based on their initial randomised allocation in the trial (intention-to-treat). In the as-treated analysis, subjects were distributed across three groups according to treatment actually received over the study period (early ACL reconstruction, rehabilitation alone, and delayed ACL reconstruction). A discount rate of 3% was applied for costs and outcome. The incremental net monetary benefit (incremental net benefit hereafter) was defined as: Embedded Imagewhere λ is willingness to pay for a one point difference in QALYs, and subscripts a and b denote early ACL reconstruction and optional delayed ACL reconstruction, respectively. A treatment is deemed cost-effective if incremental net benefit is positive. Decisions and recommendations from Swedish authorities do not rely on a single willingness to pay per QALY threshold but consider the incremental cost-effectiveness together with the severity of the condition. Results are presented in cost-effectiveness planes; in incremental net benefit graphs displaying the incremental net benefit for willingness to pay threshold values ranging from SEK 0 to SEK 1 000 000 (€0 to €111 000); and in acceptability curves showing the probability of early ACL reconstruction being cost-effective compared to optional delayed ACL reconstruction.

Uncertainty of the incremental net benefit estimates was evaluated using the non-parametric bootstrapping technique.14 Bootstrap bias-corrected 95% CIs for costs, QALYs, and incremental net benefits were estimated using 10 000 bootstrap replicates by sampling with replacement from the original dataset.15 Bootstrapped Studentised t tests and bootstrapped 95% CIs were used for statistical tests. All analyses were performed using STATA V.13 (StataCorp LP, College Station, Texas, USA) and cost-effectiveness analysis was performed using the macro ‘bsceaprogs’ available from the University of Pennsylvania (http://www.uphs.upenn.edu/dgimhsr/stat-cicer.htm accessed 12 November 2014).

Sensitivity analysis

We conducted a series of one- and multi-way sensitivity analyses to assess the robustness of base-case results. To examine the impact of the study's perspective, we excluded the productivity losses (ie, applying healthcare payer perspective). We included 50% of the costs of secondary procedures if a patient had multiple procedures during the same operation since secondary procedures could increase time in the surgical theatre and consume more resources than a single procedure operation. We excluded the costs of operations not related to leg and knee disorders to avoid possible impact of comorbidities on our findings. We detected potential outliers in cost and outcome variables (including those having second ACL reconstruction (n=7)) using box plots and performed additional sensitivity analyses excluding these subjects.

Results

The mean age of participants was 26 years at baseline, 73% were male, and the SF-6D score at baseline was 0.61 in both groups (table 1). The full-analysis set results showed that there were statistically significantly more cruciate ligament reconstructions, other knee and leg related operations, and other operations performed on those randomised to early ACL reconstruction, compared to those randomised to optional delayed ACL reconstruction. There were no statistically significant differences in meniscus surgery procedures between the two treatment groups (table 2). There were, on average, 79 (SD 56) and 76 (SD 49) rehabilitation visits per patient over 5 years in early and optional delayed ACL reconstruction groups, respectively (rate ratio=1.02; p=0.39). The SF-6D scores improved in both treatment groups over 5 years (mean improvement 0.24 for early and 0.23 for optional delayed ACL reconstruction) with no statistically significant difference (mean difference=0.01, 95% CI −0.03 to 0.06, figure 1). The as-treated analysis did not change this finding (see online supplementary appendix figure S1A).

Table 2

Number of operations over the 5 years of follow-up collected from registries

Figure 1

The SF-6D scores in the treatment groups (full-analysis set) over 5-year study period. The bars indicate 95% CIs.

The full-analysis set results showed that the mean discounted 5-year cost for a patient in the early ACL reconstruction group was about 236 000 SEK (€26 200) compared with SEK 193 500 (€21 478) in the optional delayed ACL reconstruction group (mean difference=SEK 42 330, p=0.19). The mean discounted QALYs were 3.96 and 3.83 for the early and the optional delayed ACL reconstruction groups, respectively (mean difference=0.13 QALYs, p=0.11, table 3). The cost-effectiveness plane (figure 2) showed that 85% of the bootstrapped incremental cost-effectiveness ratios (ICERs) indicated increased costs and QALY gains from early ACL reconstruction (12% of these ICERs were greater than SEK 1 000 000). Five per cent of the bootstrapped ICERs were located in the northwest quadrant where early ACL reconstruction was less effective and more costly. In 10% of the bootstrapped ICERs, early ACL reconstruction would be more effective and less costly.

Table 3

Results of the base case cost-effectiveness analysis using full-analysis set principle

Figure 2

Cost-effectiveness plane of early ACL reconstruction versus delayed optional ACL reconstruction (full-analysis set). NE (north east) quadrant: early ACL reconstruction more costly and more effective; NW (north west) quadrant: early ACL reconstruction more costly and less effective; SE (south east) quadrant: early ACL reconstruction less costly and more effective; SW (south west) quadrant: early ACL reconstruction less costly and less effective.

The point estimate of the incremental net benefit of early ACL reconstruction versus optional delayed ACL reconstruction was negative for willingness to pay levels lower than SEK 320 000 but positive for higher values; the wide 95% CIs always included zero indicating uncertain cost-effectiveness (figure 3). The acceptability curve (see online supplementary appendix figure S2A) showed that, for a willingness to pay of SEK 500 000, the probability of early ACL reconstruction being more cost-effective than the optional delayed ACL reconstruction was 66% (p=0.68).

Figure 3

Incremental net monetary benefit graph for the early ACL reconstruction versus delayed optional ACL reconstruction (full-analysis set).

In the sensitivity analyses, excluding productivity losses and accounting for costs of secondary operations generally resulted in the early ACL reconstruction being more expensive than the delayed optional ACL reconstruction (see online supplementary appendix table S2A). The incremental net benefit was never statistically significantly different from zero in any case, meaning that the conclusion from the base case was robust in all sensitivity analyses.

In the as-treated analysis, the base-case results showed no statistically significant differences in costs, QALYs or net benefits between the early and delayed ACL reconstruction groups (see online supplementary appendix table S3A and figures S3A–S4A). The cost-effectiveness plane (see online supplementary appendix figure S5A panel a) showed that about 2%, 26%, and 68% of the bootstrapped ICERs were located in the northwest, northeast, and southeast quadrants, respectively. Rehabilitation alone was statistically significantly less costly than early ACL reconstruction (mean difference=SEK 122 872, 95% CI SEK 68 625 to 179 469) but no statistically significant differences were identified in QALYs (see online supplementary appendix table S3A). In the cost-effectiveness plane (see online supplementary appendix figure S5A panel b), 23% and 77% of the bootstrapped ICERs were located in northwest and northeast quadrants, respectively. The net benefit analysis showed that rehabilitation alone was cost-effective for willingness to pay levels less than SEK 320 000 with no statistically significant differences for higher levels (see online supplementary appendix figures S6A, S7A). These results were generally robust in our sensitivity analyses (see online supplementary appendix table S4A).

Discussion

Over a 5-year period and from a societal perspective, the results of this study failed to identify a statistically significant difference in economic value of a strategy of early ACL reconstruction compared with a strategy of rehabilitation with the option of delayed ACL reconstruction when needed. In addition, the as-treated analysis showed that subjects treated with rehabilitation alone consumed fewer resources than the early ACL reconstruction group. Results were robust to sensitivity analyses and analyses by as-treated groups were consistent. We did not find any statistically significant differences in health utility between the treatment strategies of early or delayed optional ACL reconstruction with all patients receiving similar structured rehabilitation. This finding is in line with recent evidence suggesting early operative treatments are not superior to non-operative treatments for ACL injury in terms of patient reported outcomes.2 ,3 ,6 ,7 ,16 Further, a growing body of evidence shows that there are no important differences between reconstruction and non-operative treatment of the ruptured ACL in terms of other outcomes, including subsequent partial meniscectomy, return to sport, and osteoarthritis development.6 ,8 ,17 ,18 The overall results from the KANON RCT, including patient-centred outcomes after 2 and 5 years2 ,3 and the results provided here, highlight the need to identify those who benefit from rehabilitation alone at an early stage. Such early identification would spare patients from surgery that may provide no or marginal additional benefit and consequently prevent overtreatment.

Why do our data contradict previous economic analyses of ACL reconstruction?

Our results, based on a randomised trial and real-life data, contradict all available cost-effectiveness reports of ACL injury and reconstruction in which ACL reconstruction uniformly was suggested to be more cost-effective than non-surgical and/or delayed surgical treatment strategies.4 ,5 ,19–21 There are several potential explanations for the disparity in our findings compared to previous studies including differences in perspectives, prices, applied health utilities, and clinical practices (eg, structure of rehabilitation sessions) across studies. For example, the SF-6D scores reported from Mather et al5 study resulted in a QALY difference of 0.28 over 6 years, while a difference of 0.13 was observed over 5 years in our study. If a 0.28 QALY difference was used in this study, the probability of early ACL reconstruction being cost-effective for a willingness to pay SEK 500 000 would increase to 96% (from 66% using real world data). However, a more likely reason for the observed disparity is that prior modelling analyses relied on expert opinions as is common before actual estimates are available. For example, based on expert opinion, Mather et al5 assumed a ratio of 1.125 physical therapy visits for delayed versus early ACR reconstruction, while the corresponding real life ratio in our randomised trial was 0.98 over 5 years. Moreover, the results of prior studies were highly sensitive to variables such as mechanical instability, delayed meniscal surgery, and the estimated risk of knee osteoarthritis development; all assumed to have a better prognosis after ACL reconstruction despite evidence supporting a lack of such differences.2 ,3 ,6 ,8 ,17 ,18 In addition, the two reports4 ,5 based on the KANON-trial cohort included critical errors in extracting data from our publications (eg, not including meniscus surgeries performed at the time of ACL reconstruction, rehabilitation visits and additional surgical procedures; incorrectly assuming a lower rate of osteoarthritis in the early reconstruction group). Our present findings are based on the actual RCT results from the KANON RCT in combination with real-life economic data and suggest that the conclusions of Mather et al4 ,5 may be biased in favour of ACL reconstruction.

Limitations

The results of our study should be interpreted in light of some limitations. First, 5 years may be too short to capture the full impact of important outcomes such as osteoarthritis development. However, recent evidence does not suggest significant differences in health outcomes between those treated with surgical reconstruction or not treated over longer follow-up periods.6 ,8 Second, only sick leaves extending over 14 days are covered by the SSIA registers, implying that productivity losses in our study might be underestimated. Third, the results presented here do not apply to professional athletes or to less than moderately active individuals. Finally, the sample size of the KANON-trial was determined by differences in clinical outcomes of the two treatment strategies at two years. Sample sizes required for cost-effectiveness analysis are generally larger than those for clinical evaluation,22 ,23 and the wide CIs in our study imply that our findings should be interpreted with some caution.

Conclusion

Based on the present and previous reports from this RCT involving young active adults with an acute ACL injury,2 ,3 rehabilitation plus optional delayed ACL reconstruction provides comparable clinical and economical outcomes after 5 years, as does early and universal ACL reconstruction with rehabilitation. Our results highlight the risk of overtreatment after acute ACL injury and the importance of early identification of individuals who would benefit from either early ACL reconstruction or rehabilitation alone. The risk and costs associated with osteoarthritis development beyond 5 years need to be determined in long-term follow-up studies.

What is already known on this topic

  • In young active adults with acute ACL injury, rehabilitation plus optional delayed ACL reconstruction provides comparable clinical outcomes after 5 years, as does rehabilitation plus early ACL reconstruction.

  • ACL reconstruction was suggested to be a cost-effective option compared with non-operative interventions in reports using decision analytic modelling and assumptions not supported by evidence.

What this study adds

  • Use of high quality outcomes data from a randomised controlled trial in combination with real life data on health care utilisation and sick leave provides information important to treatment recommendations for ACL injured individuals, highlighting the risk of overtreatment after ACL injury.

  • From a societal perspective, early ACL reconstruction is unlikely to be a cost-effective option compared with optional delayed ACL reconstruction if needed over a 5-year period.

Acknowledgments

We thank all the subjects who participated in this trial; the collaborators of the KANON-trial, including Harald Roos and Ewa Roos; the physiotherapists and clinic staff members who managed the rehabilitation; Ola Olsson in Helsingborg, Lars Wahlström in Ängelholm, and Fredrik Nyquist in Malmö, for performing surgical interventions; and Kerstin Åkesson for study management and categorisation of adverse events, Björn Slaug and Ludvig Dahl for database management, and Torsten Boegård for radiologic advice and assessment.

References

Footnotes

  • KSC, RBF equal contribution as senior author.

  • Contributors AAK participated in the design of the study, analysis, and interpretation of results and drafting the manuscript. ME participated in the design of the study, the collection of register-based data, the interpretation of results and drafting the manuscript. KSC participated in the design of the study, the interpretation of results and drafting the manuscript. RBF and LSL were responsible for the RCT design, protocol and completion. RBF participated in the design of the study, recruited and followed up patients and collected data. RBF and LSL participated in the interpretation of results and drafting the manuscript. All authors reviewed the manuscript and approved the final version.

  • Funding Supported by grants from the Swedish Research Council, the Medical Faculty of Lund University and the Skåne Regional Council (to RBF, ME and LSL), the Thelma Zoegas Fund and the Stig and Ragna Gorthon Research Foundation (to RBF), the Swedish National Center for Research in Sports (to LSL and RBF), and Crafoord Foundation (to ME).

  • Competing interests None declared.

  • Ethics approval This study was approved by the IRB at Lund University, Sweden (LU 535-2001).

  • Provenance and peer review Not commissioned; externally peer reviewed.