SPINE Volume 39, Number 1, pp 23-32 ©2013, Lippincott Williams & Wilkins
RANDOMIZED TRIAL
Cost-Effectiveness of Total Disc Replacement Versus Multidisciplinary Rehabilitation in Patients With Chronic Low Back Pain A Norwegian Multicenter RCT Lars Gunnar Johnsen, MD,*t4: Christian Helium, MD, PhD,§ Kjersti Storheim, PhD,§1 0ystein P. Nygaard, MD, PhD,** Jens Ivar Brox, MD, PhD,§ Ivar Rossvoll, MD, PhD,*!* Magne R0, MD,** Hege Andresen, MN,** SLian Lydersen, PhD,|| Oliver Grundnes, MD, PhD,** Marit Pedersen, Cand. Polit,tt Gunnar Leivseth, MD, PhD,*** Gylfi Olafsson, MSc,§§11 Fredrik Borgström, PhD,§§1It Peter Fritzell, MD, PhD,|| and The Norwegian Spine Study Group
Study Design. Randomized clinical trial with 2-year follow-up. Objective. To evaluate the cost-effectiveness of total disc replacement (TDR) versus multidisciplinary rehabilitation (MDR) in patients with chronic low back pain (CLBP). Summary of Background Data. The existing studies on CLBP report cost-effectiveness of fusion surgery versus disc replacement and fusion versus rehabilitation. This study evaluated the costeffectiveness of TDR versus MDR.
From the *Neuroclinic, Center of Spinal Disorder and; tDepartment of Orthopaedic Surgery, Clinic of Orthopaedics and Rheumatology, St. Olavs Hospital, Trondheim University Hospital, Norway; tDepartment of Neuroscience, Faculty of Medicine, Norwegian University of Science and Technology, Trondheim, Norway; §Orthopaedic Department, Clinic for Surgery and Neurology, Oslo University Hospital and University of Oslo, Oslo, Norway; HFORMI, Clinic for Surgery and Neurology, Oslo University Hospital, Oslo, Norway; ¡Regional Centre for Child and Youth Mental Health and Child Welfare-Central Norway, Norwegian University of Science and Technology, Trondheim, Norway; **Aleris Hospital, Oslo, Norway; ++SINTFF Technology and Society, Trondheim, Norway; ítDepartment of Clinical Medicine, Neuromuscular Diseases and Research Croup, University of Tromso, Tromso, Norway; §§Quantify Research, Stockholm, Sweden; 11 Department of Learning, Informatics, Management and Fthics, Karolinska Institutet, Stockholm, Sweden; and |||Ryhov Hospital, Jönköping, Sweden. Acknowledgment date: May 21, 2013. Revision date: August 16, 2013. Acceptance date: September 23, 2013. The device(s)/drug(s) is/are FDA-approved or approved by corresponding national agency for this indication. Jönköping län grant funds and the South Fastern Norway Regional Health Authority and EXTRA funds from the Norwegian Foundation for Health and Rehabilitation, through the Norwegian Back Rain Association funds were received in support of this work. Relevant financial activities outside the submitted work: consultancy, payment for lecture, grants. Address correspondence and reprint requests to Lars Cunnar Johnsen, MD, Department of Orthopaedic Surgery, St. Olavs Hospital, Trondheim University Hospital, Olav Kyrres gt 4, 7005 Trondheim, Norway; F-mail: lars.gunnar. [email protected]
Methods. Between April 2004 and May 2007, 173 patients with CLBP (>1 yr) were randomized to TDR (n = 86) or MDR (n = 87). Treatment effects (Euro Qol 5D [EQ-5D] and Short Form 6D [SF-6D]) and relevant direct and indirect costs at 6 weeks and at 3, 6, 12, and 24 months after treatment were assessed. Cain in quality-adjusted life years (QALYs) after 2 years was estimated. Cost-effectiveness was expressed as an incremental cost-effectiveness ratio. Results. The mean QALYs gained (standard deviation) using EQ5D was 1.29 (0.53) in the TDR group and 0.95 (0.52) in the MDR group, a significant difference of 0.34 (95% confidence interval 0.18-0.50). The mean total cost per patient in the TDR group was €87,622 (58,351) compared with €74,116 (58,237) in the MDR group, which was not significantly different (95% confidence interval: - 4 0 4 1 to 31,755). The incremental cost-effectiveness ratio for the TDR procedure varied from €39,748 using EQ-5D (TDR cost-effective) to €128,328 using SE-6D (TDR not cost-effective). The dropout rate was 20% (15% TDR group, 24% MDR group). Five patients moved from the MDR to the TDR group, whereas 9 patients randomized to TDR declined surgery. Using per-protocol analysis instead of intention-to-treat analysis indicated that TDR was not cost-effective, irrespective of the use of EQ-5D or SF-6D. Conclusion. In this study, TDR was cost-effective compared with MDR after 2 years when using EQ-5D for assessing QALYs gained and a willingness to pay of €74,600 (kr500,000/QALY). TDR was not cost-effective when SF-6D was used; therefore, our results should be interpreted with caution. Longer follow-up is needed to accurately assess the cost-effectiveness of TDR. Key words: chronic low back pain, degenerative disc disease, total disc replacement, multidisciplinary rehabilitation, health economic evaluation, cost-utility analysis, cost-effectiveness, quality-adjusted life years (QALY), randomized controlled trial, EQ-5D, SF-6D, utility index.
Level of Evidence: 2 Spine 2014;39:23-32
DOI: 10.1097/BRS.0000000000000065 Spine
www.spinejournal.com
23
Sme
RANDOMIZED TRIAL
C
hronic low back pain (CLBP) associated with degenerative disc disease is a major health problem in most countries and a great challenge for both therapists and society.' Physicians must choose between several therapeutic options, and appropriate patient selection can be very difficult, especially when surgery is considered.'"* There is an on-going debate regarding the cost-effectiveness of surgery in CLBP,** where fusion is the standard method in highly selected patients. In 2004, Eritzell et al'-^ reported that fusion was more expensive than nonspecific rehabilitation (treatment as usual) after 2 years but that cost-effectiveness was dependent on the threshold of willingness to pay set by the society. In 2005, Rivero-Arias et aP ' reported that fusion was not cost-effective when compared with intensive back rehabilitation. However, there are few studies comparing the cost-effectiveness of surgery versus nonsurgical treatment in LBP.'--" In recent years, total disc replacement (TDR) has been introduced as an alternative to fusion in selected patients,"* and a cost-effectiveness randomized clinical trial (RCT) published in 2010 found that TDR was not cost-effective compared with fusion in CLBR'"* Whether TDR is cost-effective compared with specific nonsurgical treatment has not been reported. To evaluate this, the multicenter Norwegian Disc Prosthesis Study was initiated in 2004, and the clinical results were published in 2011.'''
MATERIALS AND METHODS This was a randomized controlled multicenter trial. Eor further details, see Helium et al. ' ^ Between April 2004 and May 2007, a total of 173 patients were randomized to either TDR or multidisciplinary rehabilitation (MDR). The most important inclusion criteria were CLBP (>1 yr) and degenerative changes in lumbosacral intervertebral discs. Data were recorded on follow-up consultations at baseline, at 6 weeks, and at 3, 6,12, and 24 months post-treatment. Patients with TDR received disc prosthesis, (ProDisc II, Synthes Spine, West Chester, PA) in the lumbar spine in 1 or 2 lower levels (L4-L5 or L5-S1). Access was from the front, and fluoroscopic guidance was used."" Patients in the MDR group attended treatment groups based on a model described by Brox et al.^'' This was an outpatient program with an emphasis on exercises and cognitive intervention. The treatment was interdisciplinary and directed by a team of physiotherapists and specialists in physical medicine and rehabilitation and lasted for approximately 60 hours during 3 to 5 weeks. Costs and Resource Use All relevant costs were identified, measured, and valued. Resource use was assessed, and the analyses were performed from a societal perspective, including index treatment, other hospital care, primary care, patients' private costs, and costs due to loss of production both for the patient and their relatives. The Norwegian krone (kr), with 2006 as a base year was used, and costs were adjusted for inflation into 24
www.spinejournal.com
Cost-Effectiveness of Total Disc Replacement • Johnsen et al
2012 prices and converted into euros using the rate 1 €,gj,= 6.7 kr^ijp^. Actual costs were assigned to patients regardless of their randomized group, so patients who were randomized to receive MDR but crossed over and underwent operation after having had MDR were assigned costs for both treatments. Index Treatment
Eor TDR, the resource use multiplied by unit costs, and incorporating spare capacity when appropriate, summarized the cost for each index treatment. Cost components included were: prosthesis, operation room time, wake-up services, postoperative stay in hospital, and postoperative radiograph. Eor MDR, we used a top-down approach, that is, the total cost of a spine clinic was estimated, and then how much of the clinic's costs were associated with MDR was determined.'^ A consequence of this approach is that the costs are the same for all patients. Spare capacity was included. A premium of 12% was added to common costs based on data from previous estimates of the cost weights for the Norwegian DRC system." Hospital Costs During Follow-up
The number of planned and unplanned readmissions, including outpatient visits and reoperations were registered in electronic patient administrative systems. Patients who underwent surgery received one mandatory consultation with a radiograph 6 weeks after surgery. Patients in the MDR group were offered 4 follow-up consultations at 6 weeks and at 3, 6, and 12 months, and costs were assigned if accepted. Primary Care Costs During Follow-up
Unplanned visits to general practitioners, physiotherapists, or other practitioners in the public health service were recorded in a cost diary kept by the patient as described in a previous cost-effectiveness study.'" Patients^ Private Resource Use
The use of medication (both prescribed and over the counter), contact with practitioners outside the public health service, and other costs were reported by the patient in a cost diary. Costs for relatives were included. Loss of Production
The human capital approach was used to estimate the costs related to days each patient spent out of work due to low back pain. Costs related to production losses were calculated as the number of days out of work multiplied by the average wage adjusted for part-time sick leave. Income before taxes was used for patients and after taxes for relatives when calculating costs related to work Ioss.^"
Treatment Effects and Health Utilities To measure treatment effects, the Euro Qol 5D (EQ-5D) utility index-' was used in the main analysis and the Short Eorm 6D (SE-6D) utility index^^ was used for comparison.^^ Both costs and effects were measured at baseline, at 6 weeks (not SE-6D), and at 3, 6, 12, and 24 months post-treatment. Combining utility indexes and time, the quality-adjusted life years January 2014
RANDOMIZED TRIAL
gained (QALYs) were estimated as area under the curve using the trapezoidal method.-'*'^'
Cost-Effectiveness of Total Disc Replacement • Johnsen et al
^TABtti.
^ TDR (n = 86)
MDR (n = 86)
41.1 (7.1)
40.7 (7.2)
47
58
Oswestry Disability Index (SD)
42.4 (9.4)
42.1(8.2)
Low back pain, visual analogue scale (SD)
69.6(19.9)
71.7(14.8)
EQ-5D (SD)
0.291 (0.297)
0.266 (0.296)
SF-6D (SD)
0.555 (0.086)
0.548(0.081)
85 (82)
79 (70)
Statistical Analysis Cost-effectiveness was analyzed as the difference in costs between the 2 treatment groups divided by their difference in QALYs gained. The results are presented as an incremental cost-effectiveness ratio (ICER), meaning the cost for each unit of effect gained using TDR instead of MDR.^* To derive a confidence interval (CI) for the ICER, we used nonparametric bootstrap method with 10,000 replications. The replications were plotted in a cost-effectiveness plane^^ to illustrate the uncertainty in the ICER. The concept of net monetary benefit was used to construct a cost-effectiveness acceptability curve.^^ The CEAC is linked to different willingness-to-pay values, where a value of WTP represents a theoretical measure of how much society is willing to pay for one QALY, in this study kr500,000 or €74,600.^'-^'' The curve shows the probability that TDR is cost-effective at 2 years for different values of WTP.28 The intention-to-treat (ITT) method was used, and perprotocol analysis was presented in the sensitivity analysis. Missing data were handled by multiple imputations with m = 5 imputed data sets.^'*'-^* Student t tests and corresponding 95% CIs were used to analyze differences in cost and utility. Two-sided P < 0.05 were regarded as significant. The short follow-up period was deemed not to require discounting of treatment effects or costs. Sensitivity was tested by applying different utility measures (EQ-5D and SE-ÖD),^^-^« using PP analysis instead of ITT, not applying the method of multiple imputations, applying different estimates for the loss of production, and excluding the cost of care provided by relatives from calculations. Study Approval The study was approved by the Committee for Medical Research Ethics in Eastern Norway. Data were stored with the Norwegian Data Inspectorate. The study was conducted in accordance with the Helsinki Declaration and the ICH-GCP guidelines and registered at clinicaltrial.gov (NCT00394732).
Age (SD) Female (%)
Duration of back pain, mo (SD)
Values are calculated with non-missing data. TDR indicates total disc replacement; MDR, multidisciplinary rehabilitation; SD, standard deviation.
Treatment Effects A significant difference in EQ-5D utilities in favor of TDR was found at all follow-up time points except at 6 weeks. After 2 years, the mean total improvement in QALYs (standard deviation) was 1.29 (0.53) in the TDR group and 0.95 (0.52) in the MDR group, a significant difference of 0.34 QALY (95% CI, 0.18-0.50) (Eigure 1). Costs and Resource Use Index Treatment Costs The mean cost of TDR was estimated as € 10,846, compared with €5977 per patient with MDR. Hospital Cost During Follow-up Unit costs are presented in Table 2, and the use of resources and costs during the 2-year follow-up are listed in Table 3. We EQ-5D at baseline and follow-ups
RESULTS Ratient Population Eull cost data were provided for 144 of 173 patients (84%), including 74 in the TDR group and 68 in the MDR group. Baseline characteristics were similar between groups (Table 1).'^ In all, 13% of resource use data during follow-up and 8.3% of utility scores were missing between baseline and 24 months. Eive patients crossed over from MDR and underwent TDR. Nine patients randomized to surgery decided not to undergo surgery. Five patients in the TDR group were reoperated: 2 patients were operated with fusion with the prosthesis in situ, 2 patients underwent partial removal of spinous processes, and one patient required revision of the prosthesis due to dislodgement of the polyethylene inlay. Spine
ó LU
10 15 Time to follow-up (months) TDR
20
25
MDR
Figure 1. Utilities (mean ± 2SE) at time to follow-up derived from EQ5D scoring. Significant difference in favor of TDR was found at all times of follow-up except at 6 weeks (independent 2-sided f test). SE indicates standard error; TDR, total disc replacement; MDR, multidisciplinary rehabilitation. www.spinejournaLcom
25
RANDOMIZED TRIAL
Cost-Effectiveness of Total Disc Replacement • Ji)hnsen et al
TABLE 2. Type of Resource
Unit Cost (€)
Source
Day of postoperative hospital stay
1487
Huseby eta/^'
Prosthesis
3460
Manufacturer
Outpatient visit to physician
270
Pedersen'"
Outpatient visit to physiotherapist
270
Pedersen'"
Outpatient visit to other caregivers
270
Pedersen'"
Consultation, general practitioner
124
Nossen"
Consultation, physiotherapist
124
Nossen'-
Consultation, complementary practitioners
124
Nossen'-
Loss of production, patients: cost per day
252
Statistics Norvi/ay"
Days spent by relatives: cost per clay
180
Statistics Norway"
Prices in €2012.
did not identify any significant differences in costs related to hospital follow-up between the groups. Primary Care Costs During Follow-up
We collected information on visits to general practitioners, physical therapists, and other health care professionals, as well as medication use. On average, the total cost was € 1801 and €3236 for patients with MDR and TDR respectively, and although considerable, this difference was not significant (P = 0.07).
€74,116 (58,237) in the TDR and MDR groups, respectively The difference of €13,505 was not significant (P = 0.14) (Table 4). Cost-Effectiveness Using EQ-5D, the mean ICER in the TDR group was € 39,748/ QALY (95% CI €15,990 to €65,645). We calculated 10,000 bootstrap estimates of the ICER, of which 2000 were plotted in the cost-effectiveness plane, illustrating the uncertainty of the ICER estimates (Eigure 2). If decision makers and relevant stakeholders are willing to pay kr500,000 (€74,600) for one QALX^" the chance of TDR being cost-effective from a societal perspective was approximately 90% using EQ-5D. This is illustrated in the cost-effectiveness acceptability curve (Eigure 3). When using a willingness-to-pay limit of 3 times the gross domestic product per capita ($233,000 in 2011 in Norway) as recommended by the World Health Organization, TDR was cost-effective irrespective of utility measure used (Eigure 3).
Sensitivity Analyses Sensitivity analyses are presented in Table 5. EQ-5D Versus SF-6D
All sensitivity analyses were performed using both EQ-5D and SE-6D. Using SE-6D, the improvement in the TDR group was 1.33 (0.21) QALY, compared with 1.22 (0.18) QALY in the MDR group, a significant gam of 0.11 QALY (95% CI, 0.05-0.17), although less than the gain of 0.34 QALY (95% CI, 0.18-0.50) if EQ-5D was used. Using SE-6D, the ICER in the TDR group was €128,328/QALY (95% CI €51,329 to €219,907), and the chance of TDR being cost-effective from a societal perspective was approximately 40%, that is not cost-effective (Eigure 3). Using per-Protocol Analysis
Patients' Private Resource Use and Costs (Patient Diary)
The MDR group spent on average €2999, whereas the TDR group spent €3630, which was not significantly different (P = 0.70).
Using per-protocol analysis instead of ITT analysis indicated that TDR was not cost-effective, irrespective of the use of EQ-5D or SE-6D. Multiple Imputation
Loss of Production
The mean total days out of work (standard deviation), were 225 (195) in the TDR group and 219 (210) days in the MDR group. The mean cost of production loss was €56,759 (49,301) in the TDR group and€55,225 (53,056) in the MDR group; this difference was not significant (P = 0.85). Relatives of patients with TDR spent on average 65 (159) days helping a patient, whereas relatives of patients with MDR spent 35 (93) days. The alternative cost was estimated to be €11,636 (28,624) for TDR relatives and €6303 (16,737) MDR relatives, and this difference was not significant (P = 0.23). Total Cost The total costs included index treatment and costs during the 2-year follow-up. The total cost was €87,622 (58,351) and 26
www.spinejournal.com
Multiple imputation is the method of inserting a distribution of values into missing fields.''^" This is done to avoid excluding patients with a few missing values and to avoid bias. Eive data sets were created where missing values were replaced with different imputed values, reflecting the uncertainty around the missing value. Not using these methods resulted in a considerable loss of observations and higher ICER, rendering TDR not cost-effective. Loss of Production
As the costs for 1 day off work (production losses) and time spent by relatives to help the patient constituted a high cost component, we arbitrarily raised and lowered the estimated cost per day by 30%. This only had a moderate effect on costeffectiveness and did not alter the conclusions. January 2014
RANDOMIZED TRIAI
TABLE 3.
Cost-Effectiveness of Total Disc Replacement • Johnsen et al
.gnrsp»,^, No. per Patient TDR
Cost per Patient (€)
MDR
TDR
No.
SD
No.
SD
Follow-up surgery (reoperations) (no.)
0.05
0.24
0.09
0.31
Unplanned readmissions to hospital (d)
0.81
3.39
1.16
Visit to physician (no.)
1.30
1.41
Physiotherapy outpatient clinics (no.)
0.69
Cost
MDR SD
Cost
SD
633
2746
1037
3528
5.03
488
2047
700
3031
0.25
0.79
349
379
66
211
3.27
0.02
1.24
187
882
4
334
224.97
195.41
218.90
210.30
56,759
49,301
55,225
53,056
64.53
158.73
34.95
92.82
11,636
28,624
6303
16,737
4.13
7.89
5.14
14.43
513
981
640
1797
20.36
34.39
8.04
20.67
2535
4282
1002
2573
Visits to complementary practitioners (no.)
1.12
6.03
0.88
3.96
140
751
no
493
Medication (daily doses)
3.68
4.20
3.77
5.20
48
55
49
68
Care paid by patient
NA
NA
NA
NA
2105
3524
1716
4185
Private general practitioner
0.30
4.43
1.50
13.56
21
488
188
1689
Private physiotherapist
7.89
17.64
3.02
10.78
1061
2201
376
1342
Private complementary practitioners
3.67
12.67
5.78
14.70
443
1611
719
1830
76,776
57,735
68,139
57,737
Hospital follow-up
Loss of production Days out of work Days spent by relatives Other back pain-related cost General practitioner consultations (no.) Physical therapist consultations (no.)
Patients' private cost
Total follow-up costs Cost difference (CI)
8637 (-8742 to 26,026)
TDR indicates total disc replacement; MDR, multidisciplinary rehabilitation; SD, standard deviation; Ci, confidence interval.
Relatives Days spent by relatives were excluded to see how this changed the results. As patients undergoing surgery required more care by their relatives, excluding these costs decreased the ICER and thus increased the likelihood of TDR being cost-effective.
DISCUSSION Main Findings Both TDR and MDR improved quality of life considerably during the first 2 years after the index treatment. Using
^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ B > > t i a l Irlitment lûpd Follpw-up Costs and Effect J j ^ ^ ^ ^ ^ ^ ^ ^ ^ m ^ ^ ^ ^ ^ TDR
MDR
Mean
SD
Mean
SD
Mean Difference
CI
P
QALY EQ-5D
1.29
0.53
0.95
0.52
0.34
(0.18-0.5)
RANDOMIZED TRIAL
Cost-Effectiveness of Total Disc Replacement Versus Multidisciplinary Rehabilitation in Patients With Chronic Low Back Pain A Norwegian Multicenter RCT Lars Gunnar Johnsen, MD,*t4: Christian Helium, MD, PhD,§ Kjersti Storheim, PhD,§1 0ystein P. Nygaard, MD, PhD,** Jens Ivar Brox, MD, PhD,§ Ivar Rossvoll, MD, PhD,*!* Magne R0, MD,** Hege Andresen, MN,** SLian Lydersen, PhD,|| Oliver Grundnes, MD, PhD,** Marit Pedersen, Cand. Polit,tt Gunnar Leivseth, MD, PhD,*** Gylfi Olafsson, MSc,§§11 Fredrik Borgström, PhD,§§1It Peter Fritzell, MD, PhD,|| and The Norwegian Spine Study Group
Study Design. Randomized clinical trial with 2-year follow-up. Objective. To evaluate the cost-effectiveness of total disc replacement (TDR) versus multidisciplinary rehabilitation (MDR) in patients with chronic low back pain (CLBP). Summary of Background Data. The existing studies on CLBP report cost-effectiveness of fusion surgery versus disc replacement and fusion versus rehabilitation. This study evaluated the costeffectiveness of TDR versus MDR.
From the *Neuroclinic, Center of Spinal Disorder and; tDepartment of Orthopaedic Surgery, Clinic of Orthopaedics and Rheumatology, St. Olavs Hospital, Trondheim University Hospital, Norway; tDepartment of Neuroscience, Faculty of Medicine, Norwegian University of Science and Technology, Trondheim, Norway; §Orthopaedic Department, Clinic for Surgery and Neurology, Oslo University Hospital and University of Oslo, Oslo, Norway; HFORMI, Clinic for Surgery and Neurology, Oslo University Hospital, Oslo, Norway; ¡Regional Centre for Child and Youth Mental Health and Child Welfare-Central Norway, Norwegian University of Science and Technology, Trondheim, Norway; **Aleris Hospital, Oslo, Norway; ++SINTFF Technology and Society, Trondheim, Norway; ítDepartment of Clinical Medicine, Neuromuscular Diseases and Research Croup, University of Tromso, Tromso, Norway; §§Quantify Research, Stockholm, Sweden; 11 Department of Learning, Informatics, Management and Fthics, Karolinska Institutet, Stockholm, Sweden; and |||Ryhov Hospital, Jönköping, Sweden. Acknowledgment date: May 21, 2013. Revision date: August 16, 2013. Acceptance date: September 23, 2013. The device(s)/drug(s) is/are FDA-approved or approved by corresponding national agency for this indication. Jönköping län grant funds and the South Fastern Norway Regional Health Authority and EXTRA funds from the Norwegian Foundation for Health and Rehabilitation, through the Norwegian Back Rain Association funds were received in support of this work. Relevant financial activities outside the submitted work: consultancy, payment for lecture, grants. Address correspondence and reprint requests to Lars Cunnar Johnsen, MD, Department of Orthopaedic Surgery, St. Olavs Hospital, Trondheim University Hospital, Olav Kyrres gt 4, 7005 Trondheim, Norway; F-mail: lars.gunnar. [email protected]
Methods. Between April 2004 and May 2007, 173 patients with CLBP (>1 yr) were randomized to TDR (n = 86) or MDR (n = 87). Treatment effects (Euro Qol 5D [EQ-5D] and Short Form 6D [SF-6D]) and relevant direct and indirect costs at 6 weeks and at 3, 6, 12, and 24 months after treatment were assessed. Cain in quality-adjusted life years (QALYs) after 2 years was estimated. Cost-effectiveness was expressed as an incremental cost-effectiveness ratio. Results. The mean QALYs gained (standard deviation) using EQ5D was 1.29 (0.53) in the TDR group and 0.95 (0.52) in the MDR group, a significant difference of 0.34 (95% confidence interval 0.18-0.50). The mean total cost per patient in the TDR group was €87,622 (58,351) compared with €74,116 (58,237) in the MDR group, which was not significantly different (95% confidence interval: - 4 0 4 1 to 31,755). The incremental cost-effectiveness ratio for the TDR procedure varied from €39,748 using EQ-5D (TDR cost-effective) to €128,328 using SE-6D (TDR not cost-effective). The dropout rate was 20% (15% TDR group, 24% MDR group). Five patients moved from the MDR to the TDR group, whereas 9 patients randomized to TDR declined surgery. Using per-protocol analysis instead of intention-to-treat analysis indicated that TDR was not cost-effective, irrespective of the use of EQ-5D or SF-6D. Conclusion. In this study, TDR was cost-effective compared with MDR after 2 years when using EQ-5D for assessing QALYs gained and a willingness to pay of €74,600 (kr500,000/QALY). TDR was not cost-effective when SF-6D was used; therefore, our results should be interpreted with caution. Longer follow-up is needed to accurately assess the cost-effectiveness of TDR. Key words: chronic low back pain, degenerative disc disease, total disc replacement, multidisciplinary rehabilitation, health economic evaluation, cost-utility analysis, cost-effectiveness, quality-adjusted life years (QALY), randomized controlled trial, EQ-5D, SF-6D, utility index.
Level of Evidence: 2 Spine 2014;39:23-32
DOI: 10.1097/BRS.0000000000000065 Spine
www.spinejournal.com
23
Sme
RANDOMIZED TRIAL
C
hronic low back pain (CLBP) associated with degenerative disc disease is a major health problem in most countries and a great challenge for both therapists and society.' Physicians must choose between several therapeutic options, and appropriate patient selection can be very difficult, especially when surgery is considered.'"* There is an on-going debate regarding the cost-effectiveness of surgery in CLBP,** where fusion is the standard method in highly selected patients. In 2004, Eritzell et al'-^ reported that fusion was more expensive than nonspecific rehabilitation (treatment as usual) after 2 years but that cost-effectiveness was dependent on the threshold of willingness to pay set by the society. In 2005, Rivero-Arias et aP ' reported that fusion was not cost-effective when compared with intensive back rehabilitation. However, there are few studies comparing the cost-effectiveness of surgery versus nonsurgical treatment in LBP.'--" In recent years, total disc replacement (TDR) has been introduced as an alternative to fusion in selected patients,"* and a cost-effectiveness randomized clinical trial (RCT) published in 2010 found that TDR was not cost-effective compared with fusion in CLBR'"* Whether TDR is cost-effective compared with specific nonsurgical treatment has not been reported. To evaluate this, the multicenter Norwegian Disc Prosthesis Study was initiated in 2004, and the clinical results were published in 2011.'''
MATERIALS AND METHODS This was a randomized controlled multicenter trial. Eor further details, see Helium et al. ' ^ Between April 2004 and May 2007, a total of 173 patients were randomized to either TDR or multidisciplinary rehabilitation (MDR). The most important inclusion criteria were CLBP (>1 yr) and degenerative changes in lumbosacral intervertebral discs. Data were recorded on follow-up consultations at baseline, at 6 weeks, and at 3, 6,12, and 24 months post-treatment. Patients with TDR received disc prosthesis, (ProDisc II, Synthes Spine, West Chester, PA) in the lumbar spine in 1 or 2 lower levels (L4-L5 or L5-S1). Access was from the front, and fluoroscopic guidance was used."" Patients in the MDR group attended treatment groups based on a model described by Brox et al.^'' This was an outpatient program with an emphasis on exercises and cognitive intervention. The treatment was interdisciplinary and directed by a team of physiotherapists and specialists in physical medicine and rehabilitation and lasted for approximately 60 hours during 3 to 5 weeks. Costs and Resource Use All relevant costs were identified, measured, and valued. Resource use was assessed, and the analyses were performed from a societal perspective, including index treatment, other hospital care, primary care, patients' private costs, and costs due to loss of production both for the patient and their relatives. The Norwegian krone (kr), with 2006 as a base year was used, and costs were adjusted for inflation into 24
www.spinejournal.com
Cost-Effectiveness of Total Disc Replacement • Johnsen et al
2012 prices and converted into euros using the rate 1 €,gj,= 6.7 kr^ijp^. Actual costs were assigned to patients regardless of their randomized group, so patients who were randomized to receive MDR but crossed over and underwent operation after having had MDR were assigned costs for both treatments. Index Treatment
Eor TDR, the resource use multiplied by unit costs, and incorporating spare capacity when appropriate, summarized the cost for each index treatment. Cost components included were: prosthesis, operation room time, wake-up services, postoperative stay in hospital, and postoperative radiograph. Eor MDR, we used a top-down approach, that is, the total cost of a spine clinic was estimated, and then how much of the clinic's costs were associated with MDR was determined.'^ A consequence of this approach is that the costs are the same for all patients. Spare capacity was included. A premium of 12% was added to common costs based on data from previous estimates of the cost weights for the Norwegian DRC system." Hospital Costs During Follow-up
The number of planned and unplanned readmissions, including outpatient visits and reoperations were registered in electronic patient administrative systems. Patients who underwent surgery received one mandatory consultation with a radiograph 6 weeks after surgery. Patients in the MDR group were offered 4 follow-up consultations at 6 weeks and at 3, 6, and 12 months, and costs were assigned if accepted. Primary Care Costs During Follow-up
Unplanned visits to general practitioners, physiotherapists, or other practitioners in the public health service were recorded in a cost diary kept by the patient as described in a previous cost-effectiveness study.'" Patients^ Private Resource Use
The use of medication (both prescribed and over the counter), contact with practitioners outside the public health service, and other costs were reported by the patient in a cost diary. Costs for relatives were included. Loss of Production
The human capital approach was used to estimate the costs related to days each patient spent out of work due to low back pain. Costs related to production losses were calculated as the number of days out of work multiplied by the average wage adjusted for part-time sick leave. Income before taxes was used for patients and after taxes for relatives when calculating costs related to work Ioss.^"
Treatment Effects and Health Utilities To measure treatment effects, the Euro Qol 5D (EQ-5D) utility index-' was used in the main analysis and the Short Eorm 6D (SE-6D) utility index^^ was used for comparison.^^ Both costs and effects were measured at baseline, at 6 weeks (not SE-6D), and at 3, 6, 12, and 24 months post-treatment. Combining utility indexes and time, the quality-adjusted life years January 2014
RANDOMIZED TRIAL
gained (QALYs) were estimated as area under the curve using the trapezoidal method.-'*'^'
Cost-Effectiveness of Total Disc Replacement • Johnsen et al
^TABtti.
^ TDR (n = 86)
MDR (n = 86)
41.1 (7.1)
40.7 (7.2)
47
58
Oswestry Disability Index (SD)
42.4 (9.4)
42.1(8.2)
Low back pain, visual analogue scale (SD)
69.6(19.9)
71.7(14.8)
EQ-5D (SD)
0.291 (0.297)
0.266 (0.296)
SF-6D (SD)
0.555 (0.086)
0.548(0.081)
85 (82)
79 (70)
Statistical Analysis Cost-effectiveness was analyzed as the difference in costs between the 2 treatment groups divided by their difference in QALYs gained. The results are presented as an incremental cost-effectiveness ratio (ICER), meaning the cost for each unit of effect gained using TDR instead of MDR.^* To derive a confidence interval (CI) for the ICER, we used nonparametric bootstrap method with 10,000 replications. The replications were plotted in a cost-effectiveness plane^^ to illustrate the uncertainty in the ICER. The concept of net monetary benefit was used to construct a cost-effectiveness acceptability curve.^^ The CEAC is linked to different willingness-to-pay values, where a value of WTP represents a theoretical measure of how much society is willing to pay for one QALY, in this study kr500,000 or €74,600.^'-^'' The curve shows the probability that TDR is cost-effective at 2 years for different values of WTP.28 The intention-to-treat (ITT) method was used, and perprotocol analysis was presented in the sensitivity analysis. Missing data were handled by multiple imputations with m = 5 imputed data sets.^'*'-^* Student t tests and corresponding 95% CIs were used to analyze differences in cost and utility. Two-sided P < 0.05 were regarded as significant. The short follow-up period was deemed not to require discounting of treatment effects or costs. Sensitivity was tested by applying different utility measures (EQ-5D and SE-ÖD),^^-^« using PP analysis instead of ITT, not applying the method of multiple imputations, applying different estimates for the loss of production, and excluding the cost of care provided by relatives from calculations. Study Approval The study was approved by the Committee for Medical Research Ethics in Eastern Norway. Data were stored with the Norwegian Data Inspectorate. The study was conducted in accordance with the Helsinki Declaration and the ICH-GCP guidelines and registered at clinicaltrial.gov (NCT00394732).
Age (SD) Female (%)
Duration of back pain, mo (SD)
Values are calculated with non-missing data. TDR indicates total disc replacement; MDR, multidisciplinary rehabilitation; SD, standard deviation.
Treatment Effects A significant difference in EQ-5D utilities in favor of TDR was found at all follow-up time points except at 6 weeks. After 2 years, the mean total improvement in QALYs (standard deviation) was 1.29 (0.53) in the TDR group and 0.95 (0.52) in the MDR group, a significant difference of 0.34 QALY (95% CI, 0.18-0.50) (Eigure 1). Costs and Resource Use Index Treatment Costs The mean cost of TDR was estimated as € 10,846, compared with €5977 per patient with MDR. Hospital Cost During Follow-up Unit costs are presented in Table 2, and the use of resources and costs during the 2-year follow-up are listed in Table 3. We EQ-5D at baseline and follow-ups
RESULTS Ratient Population Eull cost data were provided for 144 of 173 patients (84%), including 74 in the TDR group and 68 in the MDR group. Baseline characteristics were similar between groups (Table 1).'^ In all, 13% of resource use data during follow-up and 8.3% of utility scores were missing between baseline and 24 months. Eive patients crossed over from MDR and underwent TDR. Nine patients randomized to surgery decided not to undergo surgery. Five patients in the TDR group were reoperated: 2 patients were operated with fusion with the prosthesis in situ, 2 patients underwent partial removal of spinous processes, and one patient required revision of the prosthesis due to dislodgement of the polyethylene inlay. Spine
ó LU
10 15 Time to follow-up (months) TDR
20
25
MDR
Figure 1. Utilities (mean ± 2SE) at time to follow-up derived from EQ5D scoring. Significant difference in favor of TDR was found at all times of follow-up except at 6 weeks (independent 2-sided f test). SE indicates standard error; TDR, total disc replacement; MDR, multidisciplinary rehabilitation. www.spinejournaLcom
25
RANDOMIZED TRIAL
Cost-Effectiveness of Total Disc Replacement • Ji)hnsen et al
TABLE 2. Type of Resource
Unit Cost (€)
Source
Day of postoperative hospital stay
1487
Huseby eta/^'
Prosthesis
3460
Manufacturer
Outpatient visit to physician
270
Pedersen'"
Outpatient visit to physiotherapist
270
Pedersen'"
Outpatient visit to other caregivers
270
Pedersen'"
Consultation, general practitioner
124
Nossen"
Consultation, physiotherapist
124
Nossen'-
Consultation, complementary practitioners
124
Nossen'-
Loss of production, patients: cost per day
252
Statistics Norvi/ay"
Days spent by relatives: cost per clay
180
Statistics Norway"
Prices in €2012.
did not identify any significant differences in costs related to hospital follow-up between the groups. Primary Care Costs During Follow-up
We collected information on visits to general practitioners, physical therapists, and other health care professionals, as well as medication use. On average, the total cost was € 1801 and €3236 for patients with MDR and TDR respectively, and although considerable, this difference was not significant (P = 0.07).
€74,116 (58,237) in the TDR and MDR groups, respectively The difference of €13,505 was not significant (P = 0.14) (Table 4). Cost-Effectiveness Using EQ-5D, the mean ICER in the TDR group was € 39,748/ QALY (95% CI €15,990 to €65,645). We calculated 10,000 bootstrap estimates of the ICER, of which 2000 were plotted in the cost-effectiveness plane, illustrating the uncertainty of the ICER estimates (Eigure 2). If decision makers and relevant stakeholders are willing to pay kr500,000 (€74,600) for one QALX^" the chance of TDR being cost-effective from a societal perspective was approximately 90% using EQ-5D. This is illustrated in the cost-effectiveness acceptability curve (Eigure 3). When using a willingness-to-pay limit of 3 times the gross domestic product per capita ($233,000 in 2011 in Norway) as recommended by the World Health Organization, TDR was cost-effective irrespective of utility measure used (Eigure 3).
Sensitivity Analyses Sensitivity analyses are presented in Table 5. EQ-5D Versus SF-6D
All sensitivity analyses were performed using both EQ-5D and SE-6D. Using SE-6D, the improvement in the TDR group was 1.33 (0.21) QALY, compared with 1.22 (0.18) QALY in the MDR group, a significant gam of 0.11 QALY (95% CI, 0.05-0.17), although less than the gain of 0.34 QALY (95% CI, 0.18-0.50) if EQ-5D was used. Using SE-6D, the ICER in the TDR group was €128,328/QALY (95% CI €51,329 to €219,907), and the chance of TDR being cost-effective from a societal perspective was approximately 40%, that is not cost-effective (Eigure 3). Using per-Protocol Analysis
Patients' Private Resource Use and Costs (Patient Diary)
The MDR group spent on average €2999, whereas the TDR group spent €3630, which was not significantly different (P = 0.70).
Using per-protocol analysis instead of ITT analysis indicated that TDR was not cost-effective, irrespective of the use of EQ-5D or SE-6D. Multiple Imputation
Loss of Production
The mean total days out of work (standard deviation), were 225 (195) in the TDR group and 219 (210) days in the MDR group. The mean cost of production loss was €56,759 (49,301) in the TDR group and€55,225 (53,056) in the MDR group; this difference was not significant (P = 0.85). Relatives of patients with TDR spent on average 65 (159) days helping a patient, whereas relatives of patients with MDR spent 35 (93) days. The alternative cost was estimated to be €11,636 (28,624) for TDR relatives and €6303 (16,737) MDR relatives, and this difference was not significant (P = 0.23). Total Cost The total costs included index treatment and costs during the 2-year follow-up. The total cost was €87,622 (58,351) and 26
www.spinejournal.com
Multiple imputation is the method of inserting a distribution of values into missing fields.''^" This is done to avoid excluding patients with a few missing values and to avoid bias. Eive data sets were created where missing values were replaced with different imputed values, reflecting the uncertainty around the missing value. Not using these methods resulted in a considerable loss of observations and higher ICER, rendering TDR not cost-effective. Loss of Production
As the costs for 1 day off work (production losses) and time spent by relatives to help the patient constituted a high cost component, we arbitrarily raised and lowered the estimated cost per day by 30%. This only had a moderate effect on costeffectiveness and did not alter the conclusions. January 2014
RANDOMIZED TRIAI
TABLE 3.
Cost-Effectiveness of Total Disc Replacement • Johnsen et al
.gnrsp»,^, No. per Patient TDR
Cost per Patient (€)
MDR
TDR
No.
SD
No.
SD
Follow-up surgery (reoperations) (no.)
0.05
0.24
0.09
0.31
Unplanned readmissions to hospital (d)
0.81
3.39
1.16
Visit to physician (no.)
1.30
1.41
Physiotherapy outpatient clinics (no.)
0.69
Cost
MDR SD
Cost
SD
633
2746
1037
3528
5.03
488
2047
700
3031
0.25
0.79
349
379
66
211
3.27
0.02
1.24
187
882
4
334
224.97
195.41
218.90
210.30
56,759
49,301
55,225
53,056
64.53
158.73
34.95
92.82
11,636
28,624
6303
16,737
4.13
7.89
5.14
14.43
513
981
640
1797
20.36
34.39
8.04
20.67
2535
4282
1002
2573
Visits to complementary practitioners (no.)
1.12
6.03
0.88
3.96
140
751
no
493
Medication (daily doses)
3.68
4.20
3.77
5.20
48
55
49
68
Care paid by patient
NA
NA
NA
NA
2105
3524
1716
4185
Private general practitioner
0.30
4.43
1.50
13.56
21
488
188
1689
Private physiotherapist
7.89
17.64
3.02
10.78
1061
2201
376
1342
Private complementary practitioners
3.67
12.67
5.78
14.70
443
1611
719
1830
76,776
57,735
68,139
57,737
Hospital follow-up
Loss of production Days out of work Days spent by relatives Other back pain-related cost General practitioner consultations (no.) Physical therapist consultations (no.)
Patients' private cost
Total follow-up costs Cost difference (CI)
8637 (-8742 to 26,026)
TDR indicates total disc replacement; MDR, multidisciplinary rehabilitation; SD, standard deviation; Ci, confidence interval.
Relatives Days spent by relatives were excluded to see how this changed the results. As patients undergoing surgery required more care by their relatives, excluding these costs decreased the ICER and thus increased the likelihood of TDR being cost-effective.
DISCUSSION Main Findings Both TDR and MDR improved quality of life considerably during the first 2 years after the index treatment. Using
^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ B > > t i a l Irlitment lûpd Follpw-up Costs and Effect J j ^ ^ ^ ^ ^ ^ ^ ^ ^ m ^ ^ ^ ^ ^ TDR
MDR
Mean
SD
Mean
SD
Mean Difference
CI
P
QALY EQ-5D
1.29
0.53
0.95
0.52
0.34
(0.18-0.5)
