Lumbar spinal stenosis (LSS) is the most common degenerative condition of the lumbar spine. It causes symptoms of neurogenic claudication and is the most frequent cause of back and leg pain in the aging population. The condition mainly affects individuals older than 50 years, impairing quality of life (QOL) and consuming large amounts of healthcare resources.¹ LSS is the most common indication for spine surgery. Decompressive lumbar laminectomy is conventionally considered as the first-line surgical option for LSS. However, due to the morbidity associated with this invasive surgical treatment and a higher complication rate in the elderly population, laminectomy is not universally accepted as the optimal treatment.^2–4 Interspinous distractor devices (IDDs) have been used in the management of LSS for over a decade.^5–7 Their use is controversial due to mixed reports on their success rates, cost, and high failure rates.^8,9 The X-Stop Interspinous Process Decompression System (Medtronic Spine LLC) (Fig. 1) was the first IDD to be approved by the US FDA for the treatment of LSS. This device is intended to provide relief of the symptoms of neurogenic claudication from LSS while being minimally invasive. The procedure time for insertion is short, with potentially fewer complications than a laminectomy, and the device can be removed if necessary. X-Stop use has become increasingly popular in the management of LSS.⁷ The safety of the device was confirmed by the FDA in the US and by the National Institute for Health and Care Excellence (NICE) in the United Kingdom (UK), and its clinical efficacy was found to be similar to that of laminectomy, with a reported 70% success rate for improvement in symptoms.^10,11 However, questions have arisen regarding its cost-effectiveness; we are not aware of any data for the cost-effectiveness of IDDs in the UK, and the long-term impact on QOL needs to be determined.⁴ The objective of this study was to determine whether the device is cost-effective when compared with the standard treatment of laminectomy and how the device influences QOL.

Schematic diagram of the X-Stop implant. A: The X-Stop is a titanium/polyether ether ketone (PEEK) IDD. B: The X-Stop interspinous device is inserted between the spinous processes of the affected spinal segments. Figure is available in color online only.

Methods

The Cost-Effectiveness and Quality of Life After Laminectomy or X-Stop (CELAX) trial was an open-label randomized controlled trial conducted in three centers in the UK. The trial was registered with the National Institute for Health Research (NIHR UK; clinical trial registration no. ISRCTN88702314, www.isrctn.com). The hypothesis of the trial was that there was no important clinical difference in cost-effectiveness between laminectomy and X-Stop in the treatment of patients with LSS. Ethics approval for the study was granted by the Charing Cross Research Ethics Committee, and ethical approval was obtained for each of the participating centers.

The CELAX trial consisted of two comparative surgical arms: the laminectomy group and the X-Stop group. Written consent was sought from patients before entering the trial. The duration of clinical follow-up was 2 years. Patients were screened for symptoms of LSS when referred to the neurosurgical outpatient department at the participating centers. MRI was performed in patients who reported symptoms of neurogenic claudication as routine standard of care. The recruitment period was between 2010 and 2014. Patients with LSS who met the eligibility criteria were invited to participate in the trial. The inclusion and exclusion criteria are shown in Table 1.

The laminectomy operation was performed according to the standard practice of each surgeon and was not dictated by inclusion in the trial—all were simple one- or two-level laminectomies with bilateral muscle strip and no instrumentation or laminoplasty. The X-Stop insertion procedure was performed according to the manufacturer’s instructions. Correct positioning and adequate spacing were confirmed by radiography, and postoperative CT and/or MRI was performed at the discretion of the operating surgeon. This trial was conducted in accordance with CONSORT guidelines.¹² The CONSORT flow diagram demonstrating stages of enrollment, allocation, follow-up, and analysis is shown in Fig. 2.

CONSORT flow diagram demonstrating stages of enrollment, allocation, follow-up, and analysis. Figure is available in color online only.

The primary outcome measures were cost and QOL. Analysis was performed using intention-to-treat and as-treated principles. Figure 3 depicts the trial flowchart. Cost was measured per patient episode, and details were provided by the finance department of each center. In the National Health Service in England, hospitals are reimbursed for procedures using national tariffs based on Healthcare Resource Group codes. These tariffs for reimbursements often differ significantly from the true hospital costs. Since these charges and fees do not reflect the real cost, it was decided to use the price for operating room time per minute and the price per day of admission to estimate the cost per patient. Treatment costs (operating time costs and admission costs, respectively) at each of the participating centers were as follows: center 1, £17.98/min ($22 [USD]) and £188.16/day ($230); center 2, £16.79/min ($20.5) and £175.61/day ($214); and center 3, £14.99/min ($18.3) and £156.80/day ($192). The prices obtained were for 2010 with the adjusted marketing force factor, which varies for different centers. An additional 2% per year inflation was used. The currency conversion of British pound sterling to 1 USD was 1:1.22.

Trial flowchart. GP = general practitioner. Figure is available in color online only.

QOL was measured using the EQ-5D (UK, time trade-off tariff). This was then used to calculate quality-adjusted life-years (QALYs) by taking measurements at baseline and then postoperatively at 6 months, 1 year, and 2 years.^13,14 The area-under-the-curve method was used to calculate QALYs. This was done by plotting the QOL utility index over the four study time points. The change in QALYs and the cost per QALY were calculated for each patient over the study period based on the EQ-5D response at 2 years. The cost-effectiveness ratio was then calculated by comparing the mean cost per QALY of the two operations. The SF-36 was also used at the same time points (baseline and 6, 12, and 24 months) for sensitivity analysis. Secondary outcome measures were disease-specific patient-reported outcome measures (PROMs). These were the Oswestry Disability Index (ODI), Zurich Claudication Questionnaire (ZCQ), and Quebec Back Pain Disability Scale (QBPDS).^15–19 These PROMs were chosen in concordance with the instruments recommended by the NICE guidelines to measure outcomes in LSS (ODI and ZCQ) and were based on published health economic studies in which QALYs are calculated (EQ-5D and SF-36).¹⁹ The other secondary outcome measures were complication rates and length of hospital stay.

Sample size calculation was based on a study by Katz et al.²⁰ This was based on a type I error estimate of 0.05 and type II estimate of 0.2. We estimated a standard deviation of 33% across measurement scales, and the study was powered to detect a difference between groups of 20%. Using these estimates, a sample size of 50 patients (25 per arm) was required, which included an attrition rate of 10%. An open-label randomized design was used with a 1:1 treatment allocation. A computer random number generator was used with a block size of 10. Two authors (A.B. and B.N.) were responsible for the randomization allocation sequence, patient enrollment, and assigning interventions. Significant variation (set at p < 0.05) between the baseline scores of the two groups was checked using the two-sample Wilcoxon rank-sum (Mann-Whitney) test. The paired Wilcoxon signed-rank test was used to analyze the difference in response at the various time points. All analyses were performed using Stata (version 16.0, StataCorp) and Microsoft Excel for Macintosh.

Results

Between 2010 and 2014, 56 patients who met inclusion criteria for the CELAX trial were invited to participate. Forty-nine patients were randomized and 7 patients were managed conservatively. Of the 49 patients, 27 were randomized to lumbar laminectomy and 22 were randomized to X-Stop insertion. One laminectomy patient was lost to follow-up, and an X-Stop patient withdrew after 6 months. Therefore, 26 patients underwent a laminectomy and 21 received the X-Stop device. During the study period, 3 patients died of causes unrelated to their LSS diagnosis or treatment. Recruitment was stopped when the 4-year study period was completed. Follow-up continued for 2 years after the last patient was recruited.

The mean patient age was 69 years (range 47–86 years). Eighteen patients were female, and 29 were male. Patient demographics and baseline characteristics of their clinical condition are shown in Table 2. Leg symptoms were present in 79% of patients, and 75% had back pain. Grade 1 spondylolisthesis was present in 6 patients (5 patients in the X-Stop group and 1 patient in the laminectomy group). No statistically significant intergroup differences were identified for patient symptoms, comorbidities, or number of levels treated.

Primary Outcomes

Both groups showed an overall improvement in mean QOL (as measured by the EQ-5D) at 6 months when compared with the preoperative baseline, but at 12 and 24 months the improvement only remained significant for the laminectomy group. For the X-Stop group, the mean increase from the preoperative EQ-5D scores to the 6-month scores was 0.25 (p < 0.01), and the mean increases at 12 and 24 months were 0.21 and 0.25, respectively (neither of which reached significance). For the laminectomy group, the mean increases from the preoperative EQ-5D scores to the scores at 6, 12, and 24 months were 0.18 (p < 0.05), 0.24 (p < 0.01), and 0.29 (p < 0.001), respectively. Table 3 shows a summary of the results of the intention-to-treat analysis with the mean scores for all questionnaires, including the secondary outcome measures (QOL and disease specific), for all study time points (baseline and 6, 12, and 24 months). Figure 4 displays box-and-whisker plots representing the median EQ-5D scores for the two surgical arms.

Comparison of EQ-5D scores between the laminectomy and X-Stop groups at baseline and 6, 12, and 24 months. The y-axis represents the EQ-5D score, which ranges between −0.594 (worst health) and 1 (full health). The horizontal line represents the median, the box represents the IQR, the whiskers represent the range, and the dots represent outliers.

There were 5 patients in the X-Stop group who crossed over into the laminectomy group. An as-treated analysis was conducted at the 2-year time point to assess the outcome compared with the intention-to-treat analysis by allocating the 5 crossover patients to the laminectomy group. This analysis showed a statistically significant improvement in the primary outcome measure (EQ-5D at 2 years) for both groups compared with their preoperative scores. At the last time point measured, using EQ-5D QOL scores, there were 7 (33%) X-Stop patients whose conditions deteriorated and 14 (67%) whose conditions improved, compared with 5 (19.2%) laminectomy patients whose conditions deteriorated and 20 (76.9%) who experienced improvement.

QALYs were calculated using the area-under-the-curve method.²¹ Figure 5 shows the mean QALY calculated using the EQ-5D time trade-off scores for each time point for both groups. The average QALYs were 0.81 for the X-Stop group and 0.92 for the laminectomy group. The difference in QALYs between the two groups (−0.11, 95% CI −0.349 to 0.078) was not significant (p = 0.77).

Mean EQ-5D time trade-off (TTO) scores over the different study time points. QALYs were calculated using the area-under-the-curve method. Figure is available in color online only.

Further details of the primary outcome measures, including comparative EQ-5D subcomponent scores, preoperative EQ-5D scores, and intention-to-treat and as-treated analyses, QALY scores, and cost results, are included in the Supplementary Materials (Supplementary Figs. 1–3 and Supplementary Tables 1–3).

Secondary Outcomes

The results of the secondary outcome measures, SF-36, ZCQ, ODI, and QBPDS, are displayed in Table 3. For the SF-36 questionnaire, the laminectomy group showed improvement in 5 of 8 domains, whereas the X-Stop group showed improvement in 7 of 8 domains. Using intention-to-treat analysis, this was statistically significant for the social functioning domain for the X-Stop group (Supplementary Table 4). Bodily pain and physical function domains did not show statistically significant improvement in either group. Overall, the results of the disease-specific questionnaires were similar between the two groups. The X-Stop group reached statistically significant improvement for the ZCQ physical function scale and QBPDS at 24 months. The laminectomy group reached statistically significant improvement for the ZCQ symptom severity score at 24 months.

The mean operative time for the laminectomy group was 122 minutes (SD 3 minutes, 95% CI 105–137 minutes), and for the X-Stop group it was 66 minutes (SD 21 minutes, 95% CI 56–75 minutes). The operative time was significantly longer for the laminectomy group (T-score = 6, unpaired Student t-test). Sixteen of the 21 X-Stop operations were conducted by an independent attending or consultant surgeon, and 5 were performed by neurosurgeons in training under supervision. Eight laminectomies were performed by a consultant and 18 by neurosurgeons in training under supervision. There were 16 one-level laminectomies and 8 two-level laminectomies. Two patients were considered as having undergone a three-level laminectomy, as the operating surgeon also performed a partial laminectomy of an adjacent moderately stenotic level. In the X-Stop group, there were 15 one-level and 6 two-level procedures. There was no statistically significant difference in EQ-5D scores or number of levels operated on for either group. All laminectomies were bilateral with bilateral muscle stripping.

The average hospital length of stay was 4.2 days for the X-Stop group (range 1–20 days) and 4.3 days for laminectomy (range 1–15 days). The distribution of length of stay did not show any significant difference between the groups. There was a single complication in each group that resulted in a prolonged length of stay. No significant difference was found for the length of stay between groups (p = 0.404, Fisher’s exact test). The total complication rate for the study was 15%. The laminectomy group had 5 (19%) complications, and there were 2 (10%) in the X-Stop group. In the laminectomy group, there were 4 intraoperative dural tears. Three resolved without additional treatment, and the fourth required a return to the operating room for repair. One patient had a postoperative myocardial infarction within 30 days of the operation. In the X-Stop group, there was 1 case of worsening back pain immediately postoperatively that required IDD removal and laminectomy after 6 months and 1 intraoperative fracture of the spinous process that required intraoperative IDD removal. Both of these were coded as complications. The former complication was also coded as a reoperation, but the latter was not, as it occurred in the same anesthesia session. An additional 3 IDD patients went on to have X-Stop removal and laminectomy at the same level; all procedures were performed to treat persistent symptoms during a second operation on a separate date (1 after 8 months and 2 after 12 months). These are treatment failures and were not coded as complications.

Further details of the secondary outcome measures, including comparative mean subgroup SF-36 scores (and intention-to-treat and as-treated analyses), multivariate analysis of potential predictors of outcome, and sensitivity analysis, are included in the Supplementary Materials (Supplementary Tables 7–10).

The incremental cost of the X-Stop was £2437 ($2980), and the incremental QALY was −0.11. These values were used as the basis to calculate the incremental cost-effectiveness ratio (ICER):^11,22 ICER = (total cost of X-Stop − total cost of laminectomy)/(QALY X-Stop − QALY laminectomy) = (£5148 − £2712)/(0.81 − 0.92) = −£22,145 (−$27,078).

The X-Stop had a lower gain in QALY and cost more than a laminectomy on average. This results in a negative ICER (QALY loss for increased cost). If the device cost is subtracted from the equation, the incremental cost would be less for the X-Stop. It would be −£912.8 (−$1116), and the ICER would change to £8298 ($10,138).

If the reoperation costs for the removal of four X-Stop devices and decompression are included, then the average additional cost for the subsequent surgery and admission would be £3147.35 ($3848), which raises the X-Stop average cost from £5148 ($6417) to £5747 ($7027). There was a single reoperation in the laminectomy group (dural tear repair) that resulted in an additional cost of £3442.71 ($4209), consisting of £1511 ($1848) for additional operating room time and £1931.71 ($2361) for the admission. This results in an increased average laminectomy cost from £2712 ($3315) to £2844 ($3477).

Discussion

This article concerns a prospective randomized controlled trial of the X-Stop IDD versus laminectomy for LSS. It is a pragmatic trial and hence compares the most common treatment for LSS (standard bilateral laminectomy) as performed in three UK centers as opposed to the advanced or minimally invasive techniques that some authors will argue are the gold standard. This trial sought to address cost and QOL as opposed to biomechanical or imaging-based outcomes.

In recent years, there has been increased interest in the economic evaluation of healthcare. The value of looking at both the clinical effect of treatment and its economic impact and financial feasibility is increasingly recognized. The economic importance of LSS is increasing in relevance due to increasing life expectancy and disease prevalence. This is the first randomized controlled trial in the UK investigating QOL and the cost of an IDD. The results of this study have important implications for the use of healthcare resources and decisions regarding the use of IDDs. The average cost for a laminectomy was £2712 ($3316), and that for X-Stop insertion was £5148 ($6295). Unsurprisingly, a significant element of the excess cost was the device itself. Cost reduction and mitigation is an important element of healthcare. If the X-Stop device cost were to be removed, it would cost less than a laminectomy, with the cost savings being mainly due to shorter operating times. The length of stay was similar between the two groups, and therefore the expected cost savings in hospital admission for the X-Stop group were not observed. This may be due to various factors; the lack of a standardized discharge protocol between the different centers, the novelty of the X-Stop insertion procedure, and the age and comorbidities of LSS patients mean that safe and independent ambulation takes longer to achieve.

Our results suggest that a lumbar laminectomy is more cost-effective than X-Stop insertion for the treatment of LSS. The primary reason for this is the additional device cost rather than clinical efficacy. The intention-to-treat and as-treated analyses both found that patients treated with the X-Stop had improved outcomes in all PROMs, but to a lesser degree than those treated with a laminectomy (Fig. 4). If the reoperation costs are included, then the X-Stop cost-effectiveness is reduced and the ICER becomes negative (−£22,145 [−$27,078]). This can mean either that the new treatment is cheaper, or it had a less effective health outcome (as was the case in this study). It should be noted that the cost of operating room time in the UK is similar to that in other European countries (around €12/min).²²

In this study, both groups improved and maintained improvement with surgery for the duration of follow-up (2 years). The overall complication rate was 14.9%, which is similar to that reported in other studies such as the SPORT (12%).³ However, the complication rate following laminectomy was higher, at 19.2% (5/26 patients). The 5 complications included 4 dural tears, one of which required reoperation for repair. The remaining 3 dural tears resulted in a longer stay in the hospital but no long-term morbidity. The other complication was a myocardial infarct that occurred immediately postoperatively. All complications occurred in cases performed by neurosurgeons in training under supervision. It should be noted that the reporting of complications varies between studies, as some authors do not include intraoperative dural tears that do not require repair as complications.

This study adds to the existing literature that compares the X-Stop device with conventional surgery. Three randomized controlled trials have investigated the use of the X-Stop versus standard decompression.^8,9,22 In these studies, the X-Stop reoperation rates were 26%, 29%, and 25%. All found a similar primary clinical outcome between the two groups, but due to the high reoperation rate in the X-Stop group, the procedure was less cost-effective. We also found a similar result, with a reoperation rate of 19% within the 2-year study period, which reduced the ICER for the X-Stop group and made laminectomy more cost-effective. In 2010, Azzazi and Elhawary compared the X-Stop device with surgical decompression and fusion.²³ In that study, the authors showed that the X-Stop device had better clinical outcomes and lower complication rates than fusion. Other retrospective comparative studies have compared other types of IDDs with laminectomies and, similar to the findings in our study, have shown them to be more costly and with a higher reoperation rate.²⁴

The study was limited by small numbers and challenging recruitment. A large number of assessed patients did not meet the inclusion criteria or declined to participate. There is a wide variation in the funding and organization of different healthcare systems, and so applicability of the cost-effectiveness findings to non-UK systems needs to be made with caution.

We found that, on the basis of cost and QOL outcomes, the X-Stop should not be used as an alternative to conventional lumbar decompression. However, the X-Stop procedure does have the advantage of a shorter operating time and may be performed under local anesthesia. Therefore, there remains a limited role for the X-Stop as a treatment option in patients who are medically unfit for conventional surgery under general anesthesia.

Conclusions

Treatment of LSS with an X-Stop IDD had a higher cost (primarily device related) and a lower improvement in QOL measures at 24 months compared with laminectomy. The complication rate was lower for X-Stop than for a laminectomy, but the reoperation rate was higher. IDD devices like the X-Stop are inferior to standard laminectomy but may still be appropriate in patients with severe comorbidities in whom an invasive operation or general anesthesia would be contraindicated.

Disclosures

Dr. Borg: clinical or research support for the study described from Medtronic.

Author Contributions

Conception and design: Borg, Nurboja, Critchley, Choi. Acquisition of data: Borg, Nurboja, Critchley. Analysis and interpretation of data: Hill, Borg, Nurboja, Choi. Drafting the article: Hill, Borg. Critically revising the article: Hill, Borg, Choi. Reviewed submitted version of manuscript: Hill, Borg, Nurboja, Critchley. Approved the final version of the manuscript on behalf of all authors: Hill. Statistical analysis: Borg, Choi. Study supervision: Critchley, Choi.

Supplemental Information