Surgical stabilisation of the spine compared with a programme of inten
http://www.100md.com
《英国医生杂志》
1 Health Economics Research Centre, Department of Public Health, University of Oxford, Oxford OX3 7LF Oliver, 2 Nuffield Orthopaedic Centre, Oxford OX3 7LD, 3 Division of Health in the Community, University of Warwick, Warwick CV4 7AL
Correspondence to: H Campbell helen.campbell@dphpc.ox.ac.uk
Objective To determine whether, from a health provider and patient perspective, surgical stabilisation of the spine is cost effective when compared with an intensive programme of rehabilitation in patients with chronic low back pain.
Design Economic evaluation alongside a pragmatic randomised controlled trial.
Setting Secondary care.
Participants 349 patients randomised to surgery (n = 176) or to an intensive rehabilitation programme (n = 173) from 15 centres across the United Kingdom between June 1996 and February 2002.
Main outcome measures Costs related to back pain and incurred by the NHS and patients up to 24 months after randomisation. Return to paid employment and total hours worked. Patient utility as estimated by using the EuroQol EQ-5D questionnaire at several time points and used to calculate quality adjusted life years (QALYs). Cost effectiveness was expressed as an incremental cost per QALY.
Results At two years, 38 patients randomised to rehabilitation had received rehabilitation and surgery whereas just seven surgery patients had received both treatments. The mean total cost per patient was estimated to be ￡7830 (SD ￡5202) in the surgery group and ￡4526 (SD ￡4155) in the intensive rehabilitation arm, a significant difference of ￡3304 (95% confidence interval ￡2317 to ￡4291). Mean QALYs over the trial period were 1.004 (SD 0.405) in the surgery group and 0.936 (SD 0.431) in the intensive rehabilitation group, giving a non-significant difference of 0.068 (–0.020 to 0.156). The incremental cost effectiveness ratio was estimated to be ￡48 588 per QALY gained (–￡279 883 to ￡372 406).
Conclusion Two year follow-up data show that surgical stabilisation of the spine may not be a cost effective use of scarce healthcare resources. However, sensitivity analyses show that this could change—for example, if the proportion of rehabilitation patients requiring subsequent surgery continues to increase.
Chronic low back pain, defined as pain lasting for more than three months, is common and places a major economic burden on individuals, the healthcare system, and society as a whole. Direct costs associated with the disability were estimated to be around ￡1.6bn in the United Kingdom in 1998,1 and the condition is estimated to be responsible for close to 120 million UK work days lost per year.2
The optimal treatment strategy for patients with chronic low back pain in whom conservative therapy has failed remains uncertain. For three trials, the results of randomised comparisons between surgical and conservative management techniques have been published.3–5 Evidence from these trials shows that surgery may have some clinical benefit, but it is not clear whether intensive rehabilitation in conjunction with cognitive educational programmes can generate similar benefits for patients. Results from the first UK based trial, the spine stabilisation trial, show a significant difference in the Oswestry disability index at two years in patients randomised to spinal fusion surgery compared with intensive rehabilitation, which is arguably of clinical importance.6 This statistical difference between treatment groups in only one of the two primary outcome measures was marginal and only just reached the predefined minimal clinical difference. The potential risk and additional cost of surgery also need to be considered. No clear evidence emerged that primary fusion was any more beneficial than intensive rehabilitation. We report an economic evaluation conducted prospectively alongside the UK spine stabilisation trial. We employ a cost utility framework to determine whether any net health gain from using surgery would be sufficient to justify a likely increase in the costs of treatment. The chosen form of analysis will facilitate comparisons between the cost effectiveness of surgery and that of other healthcare interventions competing for healthcare resources.
Methods
Full details of the randomised controlled trial are published in parallel with this paper.6 Briefly, the trial was powered to detect a four point difference on the Oswestry disability index (a questionnaire designed to assess limitations of various activities of daily living7 8) between surgery and intensive rehabilitation at 24 months. We recruited 349 patients who met trial eligibility criteria from 15 centres around the UK between June 1996 and February 2002. Of these patients, 176 were randomised to spinal fusion surgery and 173 to intensive rehabilitation.
For surgery patients, the local operating surgeon decided the type of spinal stabilisation used. Rehabilitation patients attended a paced exercise and education programme based on principles of cognitive behaviour therapy totalling about 75 hoursl. We followed patients and collected back pain related NHS data and data on use of resources by patients to 24 months after randomisation. Patients who considered that their allocated treatment for chronic low back pain had failed could have further treatment including surgery. At baseline, six, 12, and 24 months, patients completed the EuroQol EQ-5D questionnaire, a generic health outcome instrument used to estimate utility scores9 and quality adjusted life years (QALYs).
Resource use
Patient specific data on the use of NHS resources included initial treatments, other back pain related hospital inpatient and outpatient visits, primary care contacts, and prescribed items of medication. We also collected data on over the counter medications purchased and visits made to private practitioners. The number of centres participating in the trial and constraints on resources precluded the collection of centre specific unit costs. Unless otherwise indicated, we used national average unit costs. All costs calculated are expressed in 2002-3 pounds sterling, inflated to this base year where appropriate.10
Spinal fusion surgery
A "micro" approach to the costing of surgery used patient specific data itemised by use of resources. We costed duration spent by each patient in the operating theatre to allow for the time of staff involved and use of the theatre.10 11 We used unit costs obtained from the lead investigating centre to value types and numbers of surgical implants and intraoperative spinal x rays.
We calculated costs for anaesthetic agents and blood products administered during each patient's surgery.12 We assumed that the costs of any surgical complications were reflected in the time spent by the patient in theatre. Finally, we costed each patient's surgery related inpatient stay in hospital.13
Intensive rehabilitation
For each patient, we collected information on the number of half day rehabilitation sessions attended and applied staff costs per session.10 Patients had one hydrotherapy session per day, valued by using a unit cost from the lead investigating centre. We costed exercise equipment and use of the hospital gym and a meeting room, by adding 15% (the overhead rate employed by the lead investigating centre) to staff, hydrotherapy, and equipment costs. Finally, we costed overnight accommodation at either a private bed and breakfast (paid for by the NHS) or on a hospital ward.14
Other back pain related NHS contacts
Patients reported attendances at hospital outpatient clinics for spinal surgery, physiotherapy, and other back pain related care at six, 12, and 24 months, which we then costed.10 13 15 16 We used the mean cost of the initial fusion procedures (calculated as described above) to cost hospital admissions for unplanned spinal fusion surgeries. Admissions for investigations included the cost of the evaluative procedure (provided by the lead investigating centre) plus overnight hotel costs on a general medical ward.14 We costed visits to and home visits from general practitioners and practice nurses.10 We used the average cost of a rehabilitation programme (calculated as described above) to cost any additional intensive rehabilitation.
Patients' costs
Patients reported contacts with private complementary practitioners, for which we obtained costs from relevant national organisations. Patients also documented items of medication prescribed, and the cost of over the counter medication purchased for back pain (see bmj.com for more details of costing methods).
Paid employment
Patients reported their employment status, occupation, and hours worked at baseline, six, 12, and 24 months. We calculated and costed total hours worked by each patient.17
Health related quality of life and quality adjusted life years
We used the EuroQol EQ-5D social tariff, estimated from a representative sample of the UK population, to convert patients' responses to the EuroQol EQ-5D questionnaire at baseline, six, 12, and 24 months into single utility levels.18 We then constructed patient specific utility profiles, assuming a straight line relation between each of the patient's utility levels. We calculated the number of QALYs experienced by each patient from baseline to 24 months as the area beneath this profile.
Discounting
We discounted costs and effects at an annual rate of 3.5%.19
Statistical analysis
A small amount of trial data (12% of follow-up resource use items, 10% of utility scores, and 14% of work status data) were missing between baseline and 24 months. We used multiple imputation,20 which replaces each missing value with a set of m plausible values, to generate three replacement values (m = 3) for each of the missing cells in these datasets, using multiple linear regression models containing the covariates intervention group, age, and sex. Arithmetic means presented for resource use, costs, and QALYs in each trial arm are an average of the means from the three datasets created. Associated standard deviations include a variance correction factor to account for variability as a result of the imputation process.
Arithmetic means and 95% confidence intervals are presented when making cost and QALY comparisons between the two arms of the trial. Skewness in cost data was modest, and we therefore report conventional parametric confidence intervals.
We carried out incremental analysis, with the mean cost difference between surgery and rehabilitation divided by the mean QALY difference to give the incremental cost effectiveness ratio (ICER). The non-parametric percentile method21 for calculating the confidence interval around this ratio used 1000 bootstrap estimates of the mean cost and QALY differences. We used the cost effectiveness acceptability curve to show the probability that surgery is cost effective at two years for different values of the NHS's willingness to pay for an additional QALY.22
Results
Baseline patient characteristics are summarised in table 1 and reported in detail in the companion paper.6
Table 1 Patients' demographics at baseline. Values are numbers (percentages) of patients unless otherwise indicated
Resource use and costs: initial interventions
Surgery—Spinal stabilisation was carried out for 139/176 (79%) patients randomised to surgery. Procedures were divided into three different groups: posterolateral fusion (n = 57), 360° fusion (n = 57), and Graf stabilisation (n = 25). Table 2 presents data on use of surgical resources and cost, averaged across all 139 patients who had surgery. The mean total cost of a spinal operation was estimated at ￡7610 (SD ￡2643). Zero surgery costs were assigned to the 37 patients who did not have spinal fusion and an average treatment cost of ￡6011 (SD ￡3896) calculated across all surgery patients.
Table 2 Breakdown of resource use and costs associated with initial treatments (in 2002-3 pounds sterling)
Intensive rehabilitation—151/173 (87%) of the patients randomised to intensive rehabilitation attended some proportion of their programme. Table 2 shows a breakdown of the mean total cost of intensive rehabilitation among the 151 patients who attended rehabilitation. The total cost was estimated to be ￡1615 (SD ￡644). Including zero rehabilitation programme costs for the 22 patients who did not attend, averaging across all 173 patients generated a cost estimate of ￡1410 (SD ￡808).
Intensive rehabilitation was substantially less costly than surgery (cost difference ￡4601, 95% confidence interval ￡4013 to ￡5189, P < 0.001).
Other back pain related NHS costs
Forty eight patients randomised to rehabilitation underwent surgical stabilisation of the spine—10 instead of rehabilitation, 38 in addition to rehabilitation. Table 3 shows that these unplanned surgery costs averaged ￡2128 per patient across the rehabilitation group. This was greater than the corresponding cost of ￡451 in the surgery group, which was primarily attributable to 11 patients who required spinal re-operations.
Table 3 Other back pain related NHS contacts and patient costs to 24 months after randomisation (2002-3 pounds sterling)
Fourteen surgery patients underwent unplanned intensive rehabilitation (seven instead of surgery, seven as well as surgery). These costs amounted to ￡162 per patient. The overall mean cost per patient of follow-up back pain related NHS contacts was ￡1302 lower in the surgery group (95% confidence interval –￡1999 to –￡605, P < 0.001).
Patient costs
Table 3 shows that patient costs related to back pain were similar in both arms.
Overall costs
Table 4 shows that at two years, spinal fusion costs ￡7830 (SD ￡5202), and intensive rehabilitation ￡4526 (SD ￡4155). The cost difference of ￡3304 favoured intensive rehabilitation (￡2317 to ￡4291, P < 0.001).
Table 4 Summary of initial treatment and 24 month follow-up costs (2002-03 pounds sterling)
Return to work
At baseline, 88/176 (50%) of the surgery group and 79/173 (46%) of the rehabilitation group were not in paid employment. By 24 months, 18 of these 88 in the surgery group (20%) and 19 of the 79 in the rehabilitation group (24%) had started some form of employment, a non-significant difference of 4% (–8% to 12%, P = 0.71). The mean number of days to obtaining paid employment was 326 (SD 167) days and 323 (SD 278) days, respectively.
The mean total number of hours worked from baseline to 24 months in the surgery group was 1678 (SD 1847) hours and in the rehabilitation group 1707 (SD 1870) hours (difference –29, 95% confidence interval –419 hours to 361 hours, P = 0.89). Corresponding gross earnings were ￡19 648 (SD ￡22 256) and ￡20 034 (SD ￡22 564), respectively—a non-significant difference of –￡386 (–￡5088 to ￡4317, P = 0.87).
Utility
Figure 1 shows utility levels at baseline, six, 12, and 24 months. We found no significant differences in utility at any of the follow-up points. A notable difference in utility existed at baseline (0.35 for surgery, 0.41 for rehabilitation). Adjusting for such a difference (using a regression based approach with trial arm and baseline score as explanatory variables) and recalculating the area under utility frontiers specific to patients produced a mean QALY difference in favour of surgery of 0.068 (–0.02 to 0.156, P = 0.13; mean 1.004 (SD 0.405) for surgery and 0.936 (SD 0.431) for rehabilitation).
Fig 1 Mean utility levels (with 95% confidence intervals) generated by applying the EuroQol EQ-5D social tariff to patients' self reported health state descriptions
Cost utility
The incremental cost per QALY of using a policy of immediate surgery was estimated to be ￡48 588 (–￡279 883 to ￡372 406). Figure 2 shows the cost effectiveness acceptability curve. Reading off from the curve shows that if decision makers are willing to pay ￡30 000 for a QALY (the value above which the National Institute for Clinical Excellence is less likely to accept a technology as cost effective23), at two years, the chance that surgery will be cost effective is less than 20%.
Fig 2 Cost effectiveness acceptability curve showing the probability that surgery is cost effective for different ceilings of willingness to pay
Sensitivity analysis
Although uncertainty surrounds several trial variables, alternative assumptions for some would not affect the baseline conclusion. For example, replacing unit costs provided by the lead investigating centre with national averages had they been available would make little difference. Similarly, alternative discount rates will have little effect over a two year time horizon.
We used sensitivity analysis to examine uncertainty surrounding the use of different surgical techniques for spinal stabilisation. Assuming any patient in the trial receiving surgery underwent posterolateral fusion, the least costly technique at ￡6170 (￡5638 to ￡6803), reduced the total cost in the surgery group to ￡6655 and in the rehabilitation group to ￡4252. The incremental cost per QALY fell to ￡35 338 (–￡188 876 to ￡410 404). Alternatively had all patients undergone 360° fusion, the most costly technique at ￡9279 (￡8632 to ￡9917), then the mean cost difference would have increased to ￡4132 (￡3065 to ￡5199) and the incremental cost per QALY to ￡60 765 (–￡420 210 to ￡617 081).
If the difference in utility observed at 24 months (0.566 for surgery and 0.532 for rehabilitation after adjustments for baseline) was maintained for a further two years, the incremental cost per QALY at four years would fall to ￡25 398 (￡13 121 to ￡75 916).
We also examined the impact of patients receiving other treatments subsequent to their allocated therapy. At two years, 45 patients (38 in the rehabilitation group and seven in the surgery group) had received both treatments under comparison. Holding all else constant and assuming patients in each arm would continue to receive both treatments in years three, four, and five at the rates observed in years one and two, the cost difference is reduced to ￡1144 (–￡312 to ￡2600) and the cost per QALY to ￡16 824 (–￡156 358 to ￡138 911). If the trend continued but at half the rate observed in years one and two, the excess cost of the surgery arm at five years would fall to ￡2165 (￡904 to ￡3425) and the cost per QALY to ￡31 838 (–￡407 056 to ￡283 783).
Discussion
Maniadakis N, Gray A. The economic burden of back pain in the UK. Pain 2000;84: 95-103.
BUPA. Health information, ABC of health, back pain. http://hcd2.bupa.co.uk/fact_sheets/mosby_factsheets/backpain.html (accessed 8 Apr 2005).
Fritzell P, H?gg P, Wessburg P, Nordwall A. 2001 Volvo award winner in clinical studies: lumbar fusion versus nonsurgical treatment for chronic low back pain. A multicenter randomized controlled trial from the Swedish lumbar spine group. Spine 2001;26: 2521-34.
Moller H, Hedlund R. Surgery versus conservative management in adult isthmic spondylolisthesis—a prospective randomized study: part 1. Spine 2000;25: 1711-5.
Brox JI, Sorensen R, Friis A, Nygaard O, Indahl A, Keller A, et al. Randomized clinical trial of lumbar instrumented fusion and cognitive intervention and exercises in patients with chronic low back pain and disc degeneration. Spine 2003;28: 1913-21.
Fairbank J, Frost H, Wilson-MacDonald J, Yu L, Barker K, Collins R, for the Spine Stabilisation Trial Group. The MRC spine stabilisation trial: a randomised controlled trial to compare surgical stabilisation of the lumbar spine versus an intensive rehabilitation programme for patients with chronic low back pain. BMJ 2005;330: (in press, May 28 issue).
Fairbank JC, Couper J, Davies JB, O'Brien JP. The Oswestry low back pain disability questionnaire. Physiotherapy 1980;66: 271-3.
Fairbank JC, Pynsent PB. The Oswestry disability index. Spine 2000;25: 2940-52.
EuroQol Group. EuroQol - a new facility for the measurement of health-related quality of life. Health Policy 1990;16: 199-208.
Netten A, Curtis L. Unit costs of health and social care. Canterbury: Personal Social Services Research Unit, University of Kent, 2003.
MASS Group. Multicentre aneurysm screening study (MASS): cost-effectiveness analysis of screening for abdominal aortic aneurysms based on four year results from a randomised controlled trial. BMJ 2002;325: 1135-41.
Dion P. The cost of anaesthetic vapours. Can J Anaesthesia 1992;39: 633.
Chartered Institute of Public Finance and Accountancy. The health service financial database and comparative tool. Croydon: Institute of Public Finance, 2000.
Mallender Hancock Associates (MHA). National average specialty treatment and hotel costs. London: Department of Health, 1998.
Information Services Division NHS Scotland. The cost book 2003. www.isdscotland.org/isd/info3.jsp?p_service=Content.show&pContentID=360&p_applic=CCC& (accessed 1 Jul 2004).
Department of Health. NHS reference costs 2003. Wetherby: DoH, 2004.
Bulman J. Patterns of pay: results of the 2003 new earnings survey. In: Labour Market Trends. Newport, Wales: Office for National Statistics, 2003: 601-12.
Dolan P, Gudex C, Kind P, Williams A. The time trade-off method: results from a general population study. Health Econ 1996;5: 141-54.
Great Britain H.M. Treasury. Green book, appraisal and evaluation in central government. London: Stationery Office, 2003.
Van Buuren S, Boshuizen H, Knook D. Multiple imputation of missing blood pressure covariates in survival analysis. Stat Med 2000;18: 681-94.
Briggs AH, Wonderling DE, Mooney CZ. Pulling cost-effectiveness analysis up by its bootstraps: a non-parametric approach to confidence interval estimation. Health Econ 1997;6: 327-40.
Van Hout BA, Al MJ, Gordon GS, Rutten FF. Costs, effects and the C/E ratios alongside a clinical trial. Health Econ 1994;3: 309-19.
Rawlins MD, Culyer AJ. National Institute for Clinical Excellence and its value judgments. BMJ 2004;329: 224-7.
Fritzell P, Hagg O, Jonsson D, Nordwall A. Cost-effectiveness of lumbar fusion and nonsurgical treatment for chronic low back pain in the Swedish lumbar spine study: A multicenter, randomized, controlled trial from the Swedish Lumbar Spine Study Group. Spine 2004;29: 421-34.
Skouen JS, Gradsal AL, Haldorsen EM, Ursin H. Relative cost-effectiveness of extensive and light multidisciplinary treatment programs versus treatment as usual for patients with chronic low back pain on long term sick leave: randomized controlled study. Spine 2002;27: 901-9.
Kuntz KM, Snider RK, Weinstein JN, Pope MH, Katz JN. Cost-effectiveness of fusion with and without instrumentation for patients with degenerative spondylolisthesis and spinal stenosis. Spine 2000;25: 1132-9.(Oliver Rivero-Arias, research officer1, )
Correspondence to: H Campbell helen.campbell@dphpc.ox.ac.uk
Objective To determine whether, from a health provider and patient perspective, surgical stabilisation of the spine is cost effective when compared with an intensive programme of rehabilitation in patients with chronic low back pain.
Design Economic evaluation alongside a pragmatic randomised controlled trial.
Setting Secondary care.
Participants 349 patients randomised to surgery (n = 176) or to an intensive rehabilitation programme (n = 173) from 15 centres across the United Kingdom between June 1996 and February 2002.
Main outcome measures Costs related to back pain and incurred by the NHS and patients up to 24 months after randomisation. Return to paid employment and total hours worked. Patient utility as estimated by using the EuroQol EQ-5D questionnaire at several time points and used to calculate quality adjusted life years (QALYs). Cost effectiveness was expressed as an incremental cost per QALY.
Results At two years, 38 patients randomised to rehabilitation had received rehabilitation and surgery whereas just seven surgery patients had received both treatments. The mean total cost per patient was estimated to be ￡7830 (SD ￡5202) in the surgery group and ￡4526 (SD ￡4155) in the intensive rehabilitation arm, a significant difference of ￡3304 (95% confidence interval ￡2317 to ￡4291). Mean QALYs over the trial period were 1.004 (SD 0.405) in the surgery group and 0.936 (SD 0.431) in the intensive rehabilitation group, giving a non-significant difference of 0.068 (–0.020 to 0.156). The incremental cost effectiveness ratio was estimated to be ￡48 588 per QALY gained (–￡279 883 to ￡372 406).
Conclusion Two year follow-up data show that surgical stabilisation of the spine may not be a cost effective use of scarce healthcare resources. However, sensitivity analyses show that this could change—for example, if the proportion of rehabilitation patients requiring subsequent surgery continues to increase.
Chronic low back pain, defined as pain lasting for more than three months, is common and places a major economic burden on individuals, the healthcare system, and society as a whole. Direct costs associated with the disability were estimated to be around ￡1.6bn in the United Kingdom in 1998,1 and the condition is estimated to be responsible for close to 120 million UK work days lost per year.2
The optimal treatment strategy for patients with chronic low back pain in whom conservative therapy has failed remains uncertain. For three trials, the results of randomised comparisons between surgical and conservative management techniques have been published.3–5 Evidence from these trials shows that surgery may have some clinical benefit, but it is not clear whether intensive rehabilitation in conjunction with cognitive educational programmes can generate similar benefits for patients. Results from the first UK based trial, the spine stabilisation trial, show a significant difference in the Oswestry disability index at two years in patients randomised to spinal fusion surgery compared with intensive rehabilitation, which is arguably of clinical importance.6 This statistical difference between treatment groups in only one of the two primary outcome measures was marginal and only just reached the predefined minimal clinical difference. The potential risk and additional cost of surgery also need to be considered. No clear evidence emerged that primary fusion was any more beneficial than intensive rehabilitation. We report an economic evaluation conducted prospectively alongside the UK spine stabilisation trial. We employ a cost utility framework to determine whether any net health gain from using surgery would be sufficient to justify a likely increase in the costs of treatment. The chosen form of analysis will facilitate comparisons between the cost effectiveness of surgery and that of other healthcare interventions competing for healthcare resources.
Methods
Full details of the randomised controlled trial are published in parallel with this paper.6 Briefly, the trial was powered to detect a four point difference on the Oswestry disability index (a questionnaire designed to assess limitations of various activities of daily living7 8) between surgery and intensive rehabilitation at 24 months. We recruited 349 patients who met trial eligibility criteria from 15 centres around the UK between June 1996 and February 2002. Of these patients, 176 were randomised to spinal fusion surgery and 173 to intensive rehabilitation.
For surgery patients, the local operating surgeon decided the type of spinal stabilisation used. Rehabilitation patients attended a paced exercise and education programme based on principles of cognitive behaviour therapy totalling about 75 hoursl. We followed patients and collected back pain related NHS data and data on use of resources by patients to 24 months after randomisation. Patients who considered that their allocated treatment for chronic low back pain had failed could have further treatment including surgery. At baseline, six, 12, and 24 months, patients completed the EuroQol EQ-5D questionnaire, a generic health outcome instrument used to estimate utility scores9 and quality adjusted life years (QALYs).
Resource use
Patient specific data on the use of NHS resources included initial treatments, other back pain related hospital inpatient and outpatient visits, primary care contacts, and prescribed items of medication. We also collected data on over the counter medications purchased and visits made to private practitioners. The number of centres participating in the trial and constraints on resources precluded the collection of centre specific unit costs. Unless otherwise indicated, we used national average unit costs. All costs calculated are expressed in 2002-3 pounds sterling, inflated to this base year where appropriate.10
Spinal fusion surgery
A "micro" approach to the costing of surgery used patient specific data itemised by use of resources. We costed duration spent by each patient in the operating theatre to allow for the time of staff involved and use of the theatre.10 11 We used unit costs obtained from the lead investigating centre to value types and numbers of surgical implants and intraoperative spinal x rays.
We calculated costs for anaesthetic agents and blood products administered during each patient's surgery.12 We assumed that the costs of any surgical complications were reflected in the time spent by the patient in theatre. Finally, we costed each patient's surgery related inpatient stay in hospital.13
Intensive rehabilitation
For each patient, we collected information on the number of half day rehabilitation sessions attended and applied staff costs per session.10 Patients had one hydrotherapy session per day, valued by using a unit cost from the lead investigating centre. We costed exercise equipment and use of the hospital gym and a meeting room, by adding 15% (the overhead rate employed by the lead investigating centre) to staff, hydrotherapy, and equipment costs. Finally, we costed overnight accommodation at either a private bed and breakfast (paid for by the NHS) or on a hospital ward.14
Other back pain related NHS contacts
Patients reported attendances at hospital outpatient clinics for spinal surgery, physiotherapy, and other back pain related care at six, 12, and 24 months, which we then costed.10 13 15 16 We used the mean cost of the initial fusion procedures (calculated as described above) to cost hospital admissions for unplanned spinal fusion surgeries. Admissions for investigations included the cost of the evaluative procedure (provided by the lead investigating centre) plus overnight hotel costs on a general medical ward.14 We costed visits to and home visits from general practitioners and practice nurses.10 We used the average cost of a rehabilitation programme (calculated as described above) to cost any additional intensive rehabilitation.
Patients' costs
Patients reported contacts with private complementary practitioners, for which we obtained costs from relevant national organisations. Patients also documented items of medication prescribed, and the cost of over the counter medication purchased for back pain (see bmj.com for more details of costing methods).
Paid employment
Patients reported their employment status, occupation, and hours worked at baseline, six, 12, and 24 months. We calculated and costed total hours worked by each patient.17
Health related quality of life and quality adjusted life years
We used the EuroQol EQ-5D social tariff, estimated from a representative sample of the UK population, to convert patients' responses to the EuroQol EQ-5D questionnaire at baseline, six, 12, and 24 months into single utility levels.18 We then constructed patient specific utility profiles, assuming a straight line relation between each of the patient's utility levels. We calculated the number of QALYs experienced by each patient from baseline to 24 months as the area beneath this profile.
Discounting
We discounted costs and effects at an annual rate of 3.5%.19
Statistical analysis
A small amount of trial data (12% of follow-up resource use items, 10% of utility scores, and 14% of work status data) were missing between baseline and 24 months. We used multiple imputation,20 which replaces each missing value with a set of m plausible values, to generate three replacement values (m = 3) for each of the missing cells in these datasets, using multiple linear regression models containing the covariates intervention group, age, and sex. Arithmetic means presented for resource use, costs, and QALYs in each trial arm are an average of the means from the three datasets created. Associated standard deviations include a variance correction factor to account for variability as a result of the imputation process.
Arithmetic means and 95% confidence intervals are presented when making cost and QALY comparisons between the two arms of the trial. Skewness in cost data was modest, and we therefore report conventional parametric confidence intervals.
We carried out incremental analysis, with the mean cost difference between surgery and rehabilitation divided by the mean QALY difference to give the incremental cost effectiveness ratio (ICER). The non-parametric percentile method21 for calculating the confidence interval around this ratio used 1000 bootstrap estimates of the mean cost and QALY differences. We used the cost effectiveness acceptability curve to show the probability that surgery is cost effective at two years for different values of the NHS's willingness to pay for an additional QALY.22
Results
Baseline patient characteristics are summarised in table 1 and reported in detail in the companion paper.6
Table 1 Patients' demographics at baseline. Values are numbers (percentages) of patients unless otherwise indicated
Resource use and costs: initial interventions
Surgery—Spinal stabilisation was carried out for 139/176 (79%) patients randomised to surgery. Procedures were divided into three different groups: posterolateral fusion (n = 57), 360° fusion (n = 57), and Graf stabilisation (n = 25). Table 2 presents data on use of surgical resources and cost, averaged across all 139 patients who had surgery. The mean total cost of a spinal operation was estimated at ￡7610 (SD ￡2643). Zero surgery costs were assigned to the 37 patients who did not have spinal fusion and an average treatment cost of ￡6011 (SD ￡3896) calculated across all surgery patients.
Table 2 Breakdown of resource use and costs associated with initial treatments (in 2002-3 pounds sterling)
Intensive rehabilitation—151/173 (87%) of the patients randomised to intensive rehabilitation attended some proportion of their programme. Table 2 shows a breakdown of the mean total cost of intensive rehabilitation among the 151 patients who attended rehabilitation. The total cost was estimated to be ￡1615 (SD ￡644). Including zero rehabilitation programme costs for the 22 patients who did not attend, averaging across all 173 patients generated a cost estimate of ￡1410 (SD ￡808).
Intensive rehabilitation was substantially less costly than surgery (cost difference ￡4601, 95% confidence interval ￡4013 to ￡5189, P < 0.001).
Other back pain related NHS costs
Forty eight patients randomised to rehabilitation underwent surgical stabilisation of the spine—10 instead of rehabilitation, 38 in addition to rehabilitation. Table 3 shows that these unplanned surgery costs averaged ￡2128 per patient across the rehabilitation group. This was greater than the corresponding cost of ￡451 in the surgery group, which was primarily attributable to 11 patients who required spinal re-operations.
Table 3 Other back pain related NHS contacts and patient costs to 24 months after randomisation (2002-3 pounds sterling)
Fourteen surgery patients underwent unplanned intensive rehabilitation (seven instead of surgery, seven as well as surgery). These costs amounted to ￡162 per patient. The overall mean cost per patient of follow-up back pain related NHS contacts was ￡1302 lower in the surgery group (95% confidence interval –￡1999 to –￡605, P < 0.001).
Patient costs
Table 3 shows that patient costs related to back pain were similar in both arms.
Overall costs
Table 4 shows that at two years, spinal fusion costs ￡7830 (SD ￡5202), and intensive rehabilitation ￡4526 (SD ￡4155). The cost difference of ￡3304 favoured intensive rehabilitation (￡2317 to ￡4291, P < 0.001).
Table 4 Summary of initial treatment and 24 month follow-up costs (2002-03 pounds sterling)
Return to work
At baseline, 88/176 (50%) of the surgery group and 79/173 (46%) of the rehabilitation group were not in paid employment. By 24 months, 18 of these 88 in the surgery group (20%) and 19 of the 79 in the rehabilitation group (24%) had started some form of employment, a non-significant difference of 4% (–8% to 12%, P = 0.71). The mean number of days to obtaining paid employment was 326 (SD 167) days and 323 (SD 278) days, respectively.
The mean total number of hours worked from baseline to 24 months in the surgery group was 1678 (SD 1847) hours and in the rehabilitation group 1707 (SD 1870) hours (difference –29, 95% confidence interval –419 hours to 361 hours, P = 0.89). Corresponding gross earnings were ￡19 648 (SD ￡22 256) and ￡20 034 (SD ￡22 564), respectively—a non-significant difference of –￡386 (–￡5088 to ￡4317, P = 0.87).
Utility
Figure 1 shows utility levels at baseline, six, 12, and 24 months. We found no significant differences in utility at any of the follow-up points. A notable difference in utility existed at baseline (0.35 for surgery, 0.41 for rehabilitation). Adjusting for such a difference (using a regression based approach with trial arm and baseline score as explanatory variables) and recalculating the area under utility frontiers specific to patients produced a mean QALY difference in favour of surgery of 0.068 (–0.02 to 0.156, P = 0.13; mean 1.004 (SD 0.405) for surgery and 0.936 (SD 0.431) for rehabilitation).
Fig 1 Mean utility levels (with 95% confidence intervals) generated by applying the EuroQol EQ-5D social tariff to patients' self reported health state descriptions
Cost utility
The incremental cost per QALY of using a policy of immediate surgery was estimated to be ￡48 588 (–￡279 883 to ￡372 406). Figure 2 shows the cost effectiveness acceptability curve. Reading off from the curve shows that if decision makers are willing to pay ￡30 000 for a QALY (the value above which the National Institute for Clinical Excellence is less likely to accept a technology as cost effective23), at two years, the chance that surgery will be cost effective is less than 20%.
Fig 2 Cost effectiveness acceptability curve showing the probability that surgery is cost effective for different ceilings of willingness to pay
Sensitivity analysis
Although uncertainty surrounds several trial variables, alternative assumptions for some would not affect the baseline conclusion. For example, replacing unit costs provided by the lead investigating centre with national averages had they been available would make little difference. Similarly, alternative discount rates will have little effect over a two year time horizon.
We used sensitivity analysis to examine uncertainty surrounding the use of different surgical techniques for spinal stabilisation. Assuming any patient in the trial receiving surgery underwent posterolateral fusion, the least costly technique at ￡6170 (￡5638 to ￡6803), reduced the total cost in the surgery group to ￡6655 and in the rehabilitation group to ￡4252. The incremental cost per QALY fell to ￡35 338 (–￡188 876 to ￡410 404). Alternatively had all patients undergone 360° fusion, the most costly technique at ￡9279 (￡8632 to ￡9917), then the mean cost difference would have increased to ￡4132 (￡3065 to ￡5199) and the incremental cost per QALY to ￡60 765 (–￡420 210 to ￡617 081).
If the difference in utility observed at 24 months (0.566 for surgery and 0.532 for rehabilitation after adjustments for baseline) was maintained for a further two years, the incremental cost per QALY at four years would fall to ￡25 398 (￡13 121 to ￡75 916).
We also examined the impact of patients receiving other treatments subsequent to their allocated therapy. At two years, 45 patients (38 in the rehabilitation group and seven in the surgery group) had received both treatments under comparison. Holding all else constant and assuming patients in each arm would continue to receive both treatments in years three, four, and five at the rates observed in years one and two, the cost difference is reduced to ￡1144 (–￡312 to ￡2600) and the cost per QALY to ￡16 824 (–￡156 358 to ￡138 911). If the trend continued but at half the rate observed in years one and two, the excess cost of the surgery arm at five years would fall to ￡2165 (￡904 to ￡3425) and the cost per QALY to ￡31 838 (–￡407 056 to ￡283 783).
Discussion
Maniadakis N, Gray A. The economic burden of back pain in the UK. Pain 2000;84: 95-103.
BUPA. Health information, ABC of health, back pain. http://hcd2.bupa.co.uk/fact_sheets/mosby_factsheets/backpain.html (accessed 8 Apr 2005).
Fritzell P, H?gg P, Wessburg P, Nordwall A. 2001 Volvo award winner in clinical studies: lumbar fusion versus nonsurgical treatment for chronic low back pain. A multicenter randomized controlled trial from the Swedish lumbar spine group. Spine 2001;26: 2521-34.
Moller H, Hedlund R. Surgery versus conservative management in adult isthmic spondylolisthesis—a prospective randomized study: part 1. Spine 2000;25: 1711-5.
Brox JI, Sorensen R, Friis A, Nygaard O, Indahl A, Keller A, et al. Randomized clinical trial of lumbar instrumented fusion and cognitive intervention and exercises in patients with chronic low back pain and disc degeneration. Spine 2003;28: 1913-21.
Fairbank J, Frost H, Wilson-MacDonald J, Yu L, Barker K, Collins R, for the Spine Stabilisation Trial Group. The MRC spine stabilisation trial: a randomised controlled trial to compare surgical stabilisation of the lumbar spine versus an intensive rehabilitation programme for patients with chronic low back pain. BMJ 2005;330: (in press, May 28 issue).
Fairbank JC, Couper J, Davies JB, O'Brien JP. The Oswestry low back pain disability questionnaire. Physiotherapy 1980;66: 271-3.
Fairbank JC, Pynsent PB. The Oswestry disability index. Spine 2000;25: 2940-52.
EuroQol Group. EuroQol - a new facility for the measurement of health-related quality of life. Health Policy 1990;16: 199-208.
Netten A, Curtis L. Unit costs of health and social care. Canterbury: Personal Social Services Research Unit, University of Kent, 2003.
MASS Group. Multicentre aneurysm screening study (MASS): cost-effectiveness analysis of screening for abdominal aortic aneurysms based on four year results from a randomised controlled trial. BMJ 2002;325: 1135-41.
Dion P. The cost of anaesthetic vapours. Can J Anaesthesia 1992;39: 633.
Chartered Institute of Public Finance and Accountancy. The health service financial database and comparative tool. Croydon: Institute of Public Finance, 2000.
Mallender Hancock Associates (MHA). National average specialty treatment and hotel costs. London: Department of Health, 1998.
Information Services Division NHS Scotland. The cost book 2003. www.isdscotland.org/isd/info3.jsp?p_service=Content.show&pContentID=360&p_applic=CCC& (accessed 1 Jul 2004).
Department of Health. NHS reference costs 2003. Wetherby: DoH, 2004.
Bulman J. Patterns of pay: results of the 2003 new earnings survey. In: Labour Market Trends. Newport, Wales: Office for National Statistics, 2003: 601-12.
Dolan P, Gudex C, Kind P, Williams A. The time trade-off method: results from a general population study. Health Econ 1996;5: 141-54.
Great Britain H.M. Treasury. Green book, appraisal and evaluation in central government. London: Stationery Office, 2003.
Van Buuren S, Boshuizen H, Knook D. Multiple imputation of missing blood pressure covariates in survival analysis. Stat Med 2000;18: 681-94.
Briggs AH, Wonderling DE, Mooney CZ. Pulling cost-effectiveness analysis up by its bootstraps: a non-parametric approach to confidence interval estimation. Health Econ 1997;6: 327-40.
Van Hout BA, Al MJ, Gordon GS, Rutten FF. Costs, effects and the C/E ratios alongside a clinical trial. Health Econ 1994;3: 309-19.
Rawlins MD, Culyer AJ. National Institute for Clinical Excellence and its value judgments. BMJ 2004;329: 224-7.
Fritzell P, Hagg O, Jonsson D, Nordwall A. Cost-effectiveness of lumbar fusion and nonsurgical treatment for chronic low back pain in the Swedish lumbar spine study: A multicenter, randomized, controlled trial from the Swedish Lumbar Spine Study Group. Spine 2004;29: 421-34.
Skouen JS, Gradsal AL, Haldorsen EM, Ursin H. Relative cost-effectiveness of extensive and light multidisciplinary treatment programs versus treatment as usual for patients with chronic low back pain on long term sick leave: randomized controlled study. Spine 2002;27: 901-9.
Kuntz KM, Snider RK, Weinstein JN, Pope MH, Katz JN. Cost-effectiveness of fusion with and without instrumentation for patients with degenerative spondylolisthesis and spinal stenosis. Spine 2000;25: 1132-9.(Oliver Rivero-Arias, research officer1, )