The Spine Journal

-

(2013)

-

Clinical Study

Interspinous device versus laminectomy for lumbar spinal stenosis: a comparative effectiveness study Chirag G. Patil, MD, MSa,*, J. Manuel Sarmiento, BAa, Beatrice Ugiliweneza, PhD, MSPHb, Debraj Mukherjee, MD, MPHa, Miriam Nu~ no, PhDa, John C. Liu, MDa, Sartaaj Waliaa, Shivanand P. Lad, MD, PhDc, Maxwell Boakye, MD, MPHb,d a

Department of Neurosurgery, Center for Neurosurgical Outcomes Research, Maxine Dunitz Neurosurgical Institute, Cedars-Sinai Medical Center, 8631 W. Third St, Suite 800E, Los Angeles, CA 90048, USA b Department of Neurosurgery, University of Louisville, 220 Abraham Flexner Way, Suite 1103, Louisville, KY 40202, USA c Division of Neurosurgery, Department of Surgery, Duke University Medical Center, 200 Trent Drive, Blue Zone- Room 4529, Durham, NC 27710, USA d Roblex Rex VA Medical Center, 800 Zorn Ave., Louisville, KY 40206, USA Received 8 January 2013; revised 16 July 2013; accepted 25 August 2013

Abstract

BACKGROUND CONTEXT: Currently no studies directly compare effectiveness between interspinous devices (IDs) and laminectomy in lumbar spinal stenosis (LSS) patients. PURPOSE: To compare reoperations, complications, and costs between LSS patients undergoing ID placement versus laminectomy. STUDY DESIGN: Retrospective comparative study. PATIENT SAMPLE: The MarketScan database (2007–2009) was queried for adults with LSS undergoing ID placement as a primary inpatient procedure. OUTCOME MEASURES: Reoperation rates, complication rates, and costs. METHODS: Each ID patient was matched with a laminectomy patient using propensity score matching. Reoperations, complications, and costs were analyzed in patients with at least 18 months postoperative follow-up. The authors did not receive funding from any external sources for this study. RESULTS: Among 498 inpatients that underwent ID placement between 2007 and 2009; the average age was 73 years. The cumulative reoperation rates after ID at 12 and 18 months were 21% and 23%, respectively. The average inpatient hospitalization lasted 1.6 days with an associated cost of $17,432. Two propensity-matched cohorts of 174 patients that had undergone ID versus laminectomy were analyzed. Longer length of stay was observed in the laminectomy cohort (2.5 days vs. 1.6 days, p!.0001), whereas ID patients accrued higher costs at index hospitalization ($17,674 vs. $12,670, p5.0001). Index hospitalization (7.5% vs. 3.5%, p5.099) and 90-day (9.2% vs. 3.5%, p5.028) complications were higher in the laminectomy cohort compared with the ID cohort. The ID patients had significantly higher reoperation rates than laminectomy patients at 12 months follow-up (12.6% vs. 5.8%, p5.026) and incurred higher cumulative costs than laminectomy patients at 12 months follow-up ($39,173 vs. $34,324, p5.289). CONCLUSIONS: Twelve-month reoperation rates and index hospitalization costs were significantly higher among patients who underwent ID compared with laminectomy for LSS. Ó 2013 Elsevier Inc. All rights reserved.

Keywords:

Lumbar spinal stenosis; Neurogenic claudication; Interspinous device; Decompressive surgery; Laminectomy; Reoperation

FDA device/drug status: Not applicable. Author disclosures: CGP: Nothing to disclose. JMS: Nothing to disclose. BU: Nothing to disclose. DM: Nothing to disclose. MN: Nothing to disclose. JCL: Nothing to disclose. SW: Nothing to disclose. SPL: Nothing to disclose. MB: Nothing to disclose. CGP and JMS contributed equally to this manuscript. 1529-9430/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.spinee.2013.08.053

Disclaimer: The content of this article does not represent the views of the Department of Veterans Affairs or the United States Government. * Corresponding author. Department of Neurosurgery, Center for Neurosurgical Outcomes Research, Cedars-Sinai Medical Center, 8631 W. Third St, Suite 800E, Los Angeles, CA 90048, USA. Tel.: (310) 423 7900; fax: (310) 967 8592. E-mail address: [email protected] (C.G. Patil)

2

C.G. Patil et al. / The Spine Journal

Introduction Lumbar spinal stenosis (LSS) refers to a medical condition in which there is anatomical narrowing of the spinal canal resulting in compression of the thecal sack containing the cauda equina nerve roots. LSS is the most common indication for lumbar spine surgery and its annual incidence is estimated to be 5 cases per 100,000 individuals [1]. The most common and specific symptom associated with LSS is neurogenic claudication, which consists of cramping lumbar pain, numbness, and/or tingling that radiates into the buttocks, thighs, or legs [2]. This characteristic symptom is exacerbated by lumbar extension (hyperlordosis) because this posture causes additional narrowing of the spinal canal and improves after lumbar flexion (delordosis). LSS has multiple causes, such as facet osteoarthritis, disc degeneration and herniation, degenerative spondylolisthesis, spondylolisthesis secondary to spondylolysis (defect in the pars interarticularis), Paget disease, and thickening of the ligamentum flavum [3]. Treatment of spinal stenosis involves a multidisciplinary approach including both operative and nonoperative therapies. When nonoperative options have been exhausted, or if patients have worsening symptoms and/or functional impairment despite attempting conservative treatment for 3 to 6 months, then surgical interventions are considered. The traditional surgical interventions used for LSS are decompressive procedures; that is, an aim to decompress the entrapped neural elements [4]. These procedures include open or minimally invasive lumbar laminectomy or laminotomy along with possible discectomy or removal of hypertrophied facets or ligaments. Because LSS is generally a disease of elderly patients, a demographic associated with significant comorbidities, the decompressive procedures are associated with some morbidity [5]. Therefore, less invasive procedures involving the US Food and Drug Administration–approved interspinous device (ID) X-STOP, have been explored to provide patients age 50 or older relief from neurogenic intermittent claudication secondary to a confirmed diagnosis of LSS through an alternative, less invasive surgical approach (X-STOP IPD System Risk Statement; Medtronic, Minneapolis, MN, USA). IDs are implanted between the spinous processes, where they act to alleviate LSS symptoms by mechanically limiting lumbar extension and increasing flexion to enlarge the spinal canal and neural foramina [6]. Results from a relatively small (N5191), multicenter, randomized-controlled trial (RCT) showed that IDs improve clinical symptoms and function in patients with neurogenic claudication compared with epidural steroid injections and other conservative treatments [7]. In this study, there were three device-related complications in the ID group (N5100), and no ID procedures were converted to a laminectomy at the time of device placement. Furthermore, a 2010 cost-effectiveness analysis of laminectomy, ID, and nonoperative treatment for patients with symptomatic LSS determined that laminectomy was

-

(2013)

-

the most cost-effective treatment strategy for this patient population, followed by ID and nonoperative treatment [8]. Yet, a recent 2012 review reported highly unfavorable results among the collective ID literature, citing maximal complication rates of 38%, reoperation rates as high as 85%, poor outcomes in 77%, and high costs associated with the utilization and implantation of IDs [5]. There are currently no studies that directly compare the effectiveness between IDs and laminectomy in patients with LSS. Similarly, the health care utilization of patients undergoing these two treatment modalities has yet to be reported. The aim of this study was twofold: (1) to evaluate reoperation, complications, health care utilization, and overall costs for patients undergoing ID placement, and (2) to assess potential outcome differences between IDand laminectomy-treated patients using the MarketScan database.

Materials and methods Study design and dataset We conducted a retrospective analysis using Truven Health Analytics’ MarketScan, commercial claims and encounters, and Medicare supplemental and Medicaid databases from 2007 to 2009 [9]. This dataset contains information generated on inpatient admission as well as all outpatient services and outpatient pharmaceutical claims. The MarketScan database is a deidentified database that was deemed as exempt from review by the University of Louisville Institutional Review Board. Each patient has a unique encrypted ID that serves also as a linkage across the different clinical utilization, enrollment, costs, and detailed patient information tables. Patient selection To identify patients with LSS and the treatment they received, we used International Classification of Disease 9th revision, Clinical Modification (ICD-9-CM) diagnosis and procedure codes, along with Current Procedural Terminology, 2011 edition (CPT-2011), codes (Supplementary Appendix 1). More specifically, only patients with a primary diagnosis of lumbar spinal stenosis (ICD-9: 724.0, 724.02, 724.09) without concurrent spondylolisthesis (ICD-9: 738.4) were included in our study. The codes for ‘‘insertion or replacement of (cement) spacer’’ (ICD-9: 84.56), ‘‘insertion or replacement of interspinous process device(s)’’ (ICD-9: 84.80), or ‘‘insertion of posterior spinous process distraction device’’ (CPT: 0171T) were used to specify use of an ID. Each LSS patient undergoing ID was matched with a similar single-level laminectomy patient on the basis of age, gender, comorbidities, year of initial surgery, length of postoperative follow-up, and insurance status using propensity score matching. For this primary laminectomy

C.G. Patil et al. / The Spine Journal

comparison cohort, additional ICD-9 and CPT codes were used as outlined in Supplementary Appendix 1 to identify patients with a primary diagnosis of lumbar spinal stenosis who underwent initial laminectomy without concomitant ID or fusion. Among both the primary ID cohort as well as the primary laminectomy comparison cohort, codes for new ID, new laminectomy, new interbody fusion, and revision interbody fusion were used to determine the type of procedures performed at reoperation during the follow-up period (Supplementary Appendix 1). The index operation and subsequent reoperations were evaluated in the inpatient tables. MarketScan data are constructed from hospital claims that report procedures as ICD-9 codes and often as CPT codes as well. Our model was designed to extract eligible patients only once that met our inclusion criteria, even if they had multiple ICD-9 and CPT codes representing the same operative procedure. In such cases, these patients were included in the study on the basis of counting only one procedure code. This prevented double-counting patients whose single operation might have been reported as both an ICD-9 procedure code and a CPT procedure code. Any patient that was 18 years of age or older was included in the data collection. Any patient that had a diagnosis of neoplasm, infection, spondylolisthesis, or trauma at the time of index hospitalization was excluded from analysis. Variables Patient age, gender, type of insurance (ie, commercial, Medicare, Medicaid), and comorbidities at the time of initial hospitalization were documented. Comorbidities were further cataloged via the Charlson index, which is a weighted measure of the burden from comorbidities [10]. We used Deyo’s adaptation of the Charlson index for administrative data to calculate the Charlson index for each record [11]. Data sources/measurements Follow-up data were amassed to compare the long-term consequences of ID versus laminectomy on reoperation, complications, health care utilization, and costs. Reoperation Among both the primary ID cohort as well as the primary laminectomy comparison cohort, codes for new ID, new laminectomy, new interbody fusion, and revision interbody fusion were used to determine the type of procedures performed at reoperation during the follow-up period (Supplementary Appendix 1). Revision interbody fusion was a necessary reoperation category to include because some patients who underwent primary ID placement had interbody spinal fusions from prior surgeries. The ‘‘all-type’’ reoperation category reflects the number of instances in which a patient underwent any type of

-

(2013)

-

3

reoperation during the follow-up period. For example, some patients undergo a combination of operative modalities during a single reoperation, such as having a laminectomy with an interbody fusion; in such instances, both the laminectomy and the interbody fusion are counted as separate procedures, despite being done during at the same reoperation. This potential overlap is the reason why the aggregate of each individual’s reoperation approaches does not equate to the number of ‘‘all-type’’ reoperations. Postoperative complications Complications from deep vein thrombosis/pulmonary embolism, pulmonary, systemic infections, wound infections, or renal, cardiac, pulmonary, or neurologic etiologies compose postoperative complications at index hospitalization, 30-day follow-up, and 90-day follow-up. Complications were quantified according to standard ICD-9-CM diagnostic coding for complications [12]. Health care resource utilization Hospital utilization was assessed primarily via hospital length of stay (LOS) measures. The length of the index hospitalization was accounted for in all patients. For those patients followed for at least 18 months, a cumulative inpatient LOS, excluding the hospital stay at index hospitalization, was calculated as well. Similarly, among those patients followed for at least 18 months, cumulative services in emergency departments (EDs) were tallied. Emergency department outpatient services is a subset of all outpatient services and represents the cumulative number of medical services aggregated by all patients within the ED. For instance, if a patient underwent a blood draw, radiograph, and magnetic resonance imaging over the course of two outpatient visits, these would be recorded as three services. Costs Determining the financial burden in terms of the total payment absorbed by patients (insurer) takes into consideration charges submitted, eligible charges, deductibles, copayments, and co-insurance coverage for each of services received in the inpatient and outpatient setting. Hospital, physician, and third party that pertain to a hospital admission or the professional fees are accessible in this database but are not directly reported in this study. To evaluate the financial variable, the total gross payment to all providers associated with an admission (for inpatient analysis) and the gross payment to a provider for a specific service (for outpatient analysis) were considered. These payments represent the amount eligible for payment after applying pricing guidelines (fee schedules and discount) before accounting deductibles, copayment, and Coordination of Benefits and are available in the dataset. Thus, costs here are from the payer’s perspective. We incorporated payments from Medicare claim records, which included paid claims for fee-forservice plans. Inpatient and outpatient costs were analyzed for all patients followed for 18 months and either within

4

C.G. Patil et al. / The Spine Journal

the primary ID cohort or the primary laminectomy comparison cohort. Hospital costs were tallied both for the index hospitalization as well as cumulatively for the 18 months patients were followed. Cumulative costs at 12- and 18-month follow-up exclude costs from index hospitalization, except for overall costs, which is inclusive of index hospitalization charges. Overall costs comprised index hospitalization costs, postoperative outpatient services costs, and outpatient medication costs. Similarly, cumulative ED costs, medication costs, and overall costs were calculated for all primary ID cohort patients and all primary laminectomy comparison cohort patients followed for 18 months. Because hospital costs may vary from hospital to hospital depending on different contracts and arrangements hospitals may have with payers, we reported average costs and have included the standard deviation of the costs. All payments were adjusted to 2009 dollars using the medical care component of the consumer price index [13]. Statistical analysis We summarized continuous variables such as age, postoperative follow-up days, index hospitalization length of stay, cumulative hospital days, ED services, and costs using means and standard deviations. Categorical variables such as female, Charlson index, insurance type, reoperation, and complications were summarized using the chi-square test. We did not encounter any missing data in our analysis. To compare ID with laminectomy, we used multivariate regression models and adjusted for age, female, insurance type, and Charlson index. For continuous outcomes such as index hospitalization length of stay, postoperative hospital days, and costs, we used linear regression on logtransformed variables. For count outcomes such as number of outpatient services and number of outpatient medications, we used negative binomial regression. For categorical outcomes such as reoperation and complication, we used logistic regression. We used two-sided tests, and p values less than or equal to .05 were considered to be statistically significant. All statistical analyses were performed in Statistical Analysis Software (SAS, Cary, NC, USA) 9.3. Results

-

(2013)

-

Table 1 Characteristics for all inpatients that underwent ID placement between 2007 and 2009 Characteristics at the time of procedure

Total (N5498)

Age, mean (SD) Postoperative follow-up days, mean (SD) Females, N (%) Charlson index, N (%) 0 1 2 3 Insurance type, N (%) Commercial Medicaid Medicare

73 (9) 453 (279) 241 (48.39) 343 123 21 11

(68.88) (24.70) (4.22) (2.21)

86 (17.27) 14 (2.81) 398 (79.92)

ID, interspinous device; SD, standard deviation.

of follow-up data. This cohort had a mean age of 73 years, 53.9% of patients were women, and the mean number of postoperative follow-up days was 758 days (2.1 years) (Table 2). The majority of these patients had no comorbidities (75.8%, Charlson index50) and most were covered under Medicare health insurance (78.0%). Reoperation rates, health care utilization, and costs The overall reoperation rate was 22.1% for all patients that had an initial ID placed (Table 3). The mean length of stay for all patients in the inpatient setting was 1.53 days, and the mean hospital cost incurred by these patients was $19,988. The most common type of reoperation was laminectomy (12.1%), followed by new ID placement (10.4%), revision interbody fusion (6.0%), and new interbody fusion (3.6%). Complications occurred in 16 (3.2%) patients at index hospitalization, 51 (10.2%) patients within 30 days, and in 51 (11.2%) patients within 90 days. In patients with at least 18 months follow-up, the cumulative 12- and 18-month reoperation rates were 21.4% and 23.1%, respectively (Table 4). The most common type of reoperation at 18 months was laminectomy (14.8%), Table 2 Characteristics for inpatients with at least 18 months postoperative followup that underwent ID placement between 2007 and 2008

Demographics

Characteristics at the time of procedure

Total (N5182)

The MarketScan data revealed 498 patients that received an inpatient ID between the years 2007 and 2009. The mean patient age was 73 years, 48.4% of patients were female, and 79.9% of patients were insured by Medicare (Table 1). The mean number of postoperative follow-up days for this cohort was 453 days (1.2 years). Most (68.9%) patients who underwent an ID procedure had no comorbidities (ie, Charlson index50). None of the patients had any missing data to report. Of the initial 498 patients that received an ID between 2007 and 2009, 182 (36.5%) patients had at least 18 months

Age, mean (SD) Postoperative follow-up days, mean (SD) Females, N (%) Charlson index, N (%) 0 1 2 3 Insurance type, N (%) Commercial Medicaid Medicare

73 (10) 758 (160) 98 (53.85)

ID, interspinous device; SD, standard deviation.

138 35 4 5

(75.82) (19.23) (2.20) (2.75)

35 (19.23) 5 (2.75) 142 (78.02)

C.G. Patil et al. / The Spine Journal Table 3 Postoperative outcomes for all inpatients that underwent ID placement between 2007 and 2009 Outcomes

Total (N5498)

Overall reoperation, N (%) All type New ID Laminectomy New interbody spinal fusion Revision interbody spinal fusion Index hospitalization Days Cost in 2009 dollars

110 52 60 18 30

(22.09) (10.44) (12.05) (3.61) (6.02)

1.53 (1.27) $19,988 ($24,087)

ID, interspinous device.

followed by new ID placement or revision (both 8.8%), and new interspinous device (2.2%), similar to trends at 12 months follow-up. The cumulative mean LOS excluding hospital stay at index hospitalization for patients in this cohort was 1.81 days at 12 months follow-up and 2.65 days at 18 months follow-up. Hospital costs for index hospitalization was $17,432 and the corresponding cumulative hospital costs (excluding costs incurred at index hospitalization) at 12 and 18 months follow-up were $8,471 and $13,289, respectively. ID- and laminectomy-matched cohorts A total of 348 patients with at least 18 months follow-up, half of which underwent direct decompression via Table 4 Postoperative outcomes for inpatients with at least 12 and 18 months postoperative follow-up that underwent ID placement between 2007 and 2008 Outcomes 12 months reoperation, N (%) All type New ID Laminectomy New interbody spinal fusion Revision interbody spinal fusion 18 months reoperation, N (%) All type New ID Laminectomy New interbody spinal fusion Revision interbody spinal fusion Hospital days, mean, (SD) Index hospitalization 12 months postoperation (cumulative, index hospitalization) 18 months postoperation (cumulative, index hospitalization) Hospital cost, mean, (SD) Index hospitalization 12 months postoperation (cumulative, index hospitalization) 18 months postoperation (cumulative, index hospitalization)

-

(2013)

-

5

laminectomy and the other half indirect decompression by ID, were matched using propensity score matching (Table 5). No statistically significant differences were observed between the two matched cohorts on the basis of age, post-operative follow-up days, gender, insurance carrier, Charlson index, and year of index surgery signaling a postmatch covariate balance. The postoperative outcomes of these two cohorts are provided in Table 6. Patients who underwent IDs were significantly more likely to undergo a reoperation compared with patients who underwent primary laminectomy for LSS symptoms at 12 months follow-up (12.6% vs. 5.8%, respectively, p5.026). Patients that originally underwent ID placement had significantly more laminectomy reoperations compared with the patients who initially had a laminectomy (10.9% vs. 2.3%, respectively, p5.001). However, patients who underwent laminectomy had significantly more 30and 90-day complications compared with patients that underwent ID placement (9.2% vs. 3.5% in both cases, both p5.028). Complications at index hospitalization did not significantly differ between the two groups (7.5% vs. 3.5%, p5.099). However, patients who underwent laminectomy spent significantly more days at index hospitalization than patients who underwent ID placement (2.49 vs. 1.58, p!.0001). Interestingly, despite having shorter index hospitalization LOS, the ID cohort incurred significantly higher costs at index hospitalizations ($17,674 ID vs. $12,670, p5.0001), likely reflecting higher operative costs associated with instrumentation. The overall costs associated with laminectomy and ID patients at 12 and 18 months follow-up showed no significant differences. Discussion

Total (N5182)

Key results 39 16 23 4 12

(21.43) (8.79) (12.64) (2.20) (6.59)

42 16 27 4 16

(23.08) (8.79) (14.84) (2.20) (8.79)

excluding

1.57 (1.39) 1.81 (4.82)

excluding

2.65 (6.71)

excluding

$17,432 ($19,559) $8,471 ($28,124)

excluding

$13,289 ($40,986)

ID, interspinous device; SD, standard deviation.

Our study evaluated the reoperation rates, health care utilization, and costs of 498 patients diagnosed with LSS who Table 5 Demographics of matched patients with at least 18 months postoperative follow-up in an inpatient setting Characteristics Age, mean (SD) Postoperative follow-up days, mean (SD) Female, N (%) Insurance, N (%) Commercial Medicaid Medicare Charlson index, N (%) 0 1 2 3þ

Laminectomy (N5174)

ID (N5174)

p Value

73 (10) 754 (161)

73 (10) 762 (160)

.9528 .6571

99 (56.90)

92 (52.87)

37 (21.26) 10 (5.75) 127 (72.99)

35 (20.11) 5 (2.87) 134 (77.01)

127 44 1 2

133 35 2 4

.4508 .3848

.5644 (72.99) (25.29) (0.57) (1.15)

ID, interspinous device; SD, standard deviation.

(76.44) (20.11) (1.15) (2.30)

6

C.G. Patil et al. / The Spine Journal

-

(2013)

-

Table 6 Postoperative outcomes of matched patients with at least 18 months postoperative follow-up in an inpatient setting Outcome 12 months reoperation, N (%) All type Laminectomy New interbody spinal fusion Revision interbody spinal refusion ID 18 months reoperation, N (%) All type Laminectomy New interbody spinal fusion Revision interbody spinal refusion ID Complications, N (%) Index hospitalization 30-day complications 90-day complications Index hospitalization LOS, days Cost in 2009 US$ 18-month hospital use, mean (SD) Cumulative LOS in days (excluding index hospitalization) Cost in 2009 US$ (cumulative, excluding index hospitalization) 18-month outpatient ED services Cumulative services (excluding index hospitalization) Cost in 2009 US$ (cumulative, excluding index hospitalization) Overall costs in 2009 US$ 12 months (including index hospitalization) 18 months (including index hospitalization)

Laminectomy (N5174)

ID (N5174)

p Value

10 4 6 4 0

(5.75) (2.30) (3.45) (2.30) (0.00)

22 19 4 8 1

(12.64) (10.92) (2.30) (4.60) (0.57)

.0260 .0012 .521 .2399 1

17 7 10 6 0

(9.77) (4.02) (5.75) (3.45) (0.00)

26 23 4 12 1

(14.94) (13.22) (2.30) (6.90) (0.57)

.1426 .0022 .1017 .1464 1

13 (7.47) 16 (9.20) 16 (9.20)

6 (3.45) 6 (3.45) 6 (3.45)

.0986 .0276 .0276

2.49 (2.77) $12,670 ($9,883)

1.58 (1.41) $17,674 ($19,959)

!.0001 .0001

3.28 (11.20) $12,331 ($23,944)

2.69 (6.84) $13,559 ($41,803)

.473 .499

2.14 (7.36) $272 ($851)

1.84 (3.52) $322 ($688)

.0757 .0593

$34,324 ($33,209) $44,834 ($39,670)

$39,173 ($45,608) $51,255 ($62,959)

.289 .6846

ED, emergency department; ID, interspinous device; LOS, length of stay; SD, standard deviation.

underwent ID placement in an inpatient setting between 2007 and 2009, at least 174 (36.5%) of which had at least 18 months postoperative follow-up. In our propensitymatched comparative analysis between laminectomy and ID cohorts, ID patients had significantly higher index hospitalization costs and had more than twice the rate of reoperation as laminectomy patients at 12 months follow-up (12.6% vs. 5.8%). Patients in the laminectomy cohort did, however, have significantly more postoperative complications and had longer LOS compared with ID patients. ID versus nonoperative management There are various IDs available in the market, each one varying in design and composition from the others [14]. Despite these inherent differences, their underlying mechanism of lodging between adjacent spinous processes and limiting lumbar extension remain the same. In 2004, Zucherman et al. reported their results from a study designed to compare placement of the US Food and Drug Administration– approved ID X-STOP versus nonoperative management for LSS patients with neurogenic claudication [15]. This RCT enrolled 200 patients, 100 of which were treated with a total of 136 (64 single-level devices and 36 doublelevel devices) X-STOP implants. The primary outcomes measurement was the Zurich Claudication Questionnaire, a patient-completed validated tool that captures data in three

different domains: symptom severity, physical function, and posttreatment patient satisfaction. Treatment was defined as successful if the patient was at least ‘‘somewhat satisfied’’ (corresponding with an average score of 2.5) and had ‘‘at least a 0.5 improvement in both symptoms severity and physical function.’’ [15] At the 1-year follow-up, the X-STOP cohort reported a success rate of 59% versus 12% in the nonoperative management cohort. The same pool of patients were evaluated at 2-year follow-up in a separate study in which the X-STOP group reported improved symptom severity by 45% from their baseline scores, whereas the control group reported only 7.4% improvement (p!.001) [7]. Similarly, physical function improved by 44% in the X-STOP group compared with 0.4% in the control group (p!.001). Furthermore, Medical Outcomes Study Short Form-36 (SF-36) scores at 2 years follow-up showed that the X-STOP group scored significantly higher than the control group in the domains of quality of life, physical functioning, bodily pain, mental health, the physical component summary, and social function, adding further evidence to support the efficacy of ID over nonoperative treatment in patients with LSS [12]. ID versus laminectomy At the time this article was written, there are no level 1 evidence studies (ie, RCTs) that directly compare the

C.G. Patil et al. / The Spine Journal

clinical effectiveness of ID placement with direct surgical decompression (ie, laminectomy) for LSS. In 2011, Chou et al. published a systematic review of RCTs separately comparing laminectomy and ID with nonoperative management using a modified network analysis approach [16]. In this review, an indirect comparison of ID placement with surgical decompression was performed after analyzing the two previously mentioned RCTs by Zucherman et al. (comparing ID placement to conservative treatment) [7,15] along with three RCTs that compared surgical decompression with conservative treatment [17–19]. This analysis concluded there was low-level evidence supporting improved disability and pain outcomes at 12 months with ID compared with laminectomy. Furthermore, there was low-level evidence showing no difference in walking distance improvement or complication rates between ID and laminectomy. The review authors indicate, however, that these results should be considered with caution because of the indirect comparison, heterogeneity of treatment groups, and short follow-up periods. Despite the paucity of RCTs comparing IDs with surgical decompression, there are various other lower grade studies comparing both treatment modalities that report mixed results. A 2010 prospective study by Sobottke et al. compared clinical outcomes in 8 patients who underwent ID placement with 21 patients who had microsurgical bilateral operative decompression. This study found no significant difference between groups in terms of improvement in back/leg pain or Oswestry Disability Index, SF-36 physical, and SF-36 mental component summaries [20]. Moreover, Kondrashov et al. reported their results from a retrospective study involving 18 patients who had X-STOP implantation and 12 patients who had laminectomy for intermittent neurogenic claudication secondary to LSS [21]. Twelve of 18 (67%) X-STOP patients were treated at one level, with 6 of 18 (33%) treated at two levels. In contrast, 3 of 12 (25%) laminectomy patients were treated at one level, with 9 of 12 (75%) treated at two levels. The average follow-up was 51 months in the X-STOP group (standard deviation 5.3 months) and 52 months in the laminectomy group (standard deviation 5.3 months). An absolute improvement of 15 Oswestry Disability Index points was used to define an individual patient’s success. Based on this criterion, 78% (14/18) X-STOP procedures were considered successful and 33% (4/12) laminectomy procedures were considered successful (p!.0243). However, in another recent study, Richter et al. showed that the additional placement of an ID does not further improve clinical outcome beyond laminectomy in 12- and 24-month follow-up intervals [22,23]. Our study similarly shows the mixed effectiveness of IDs, highlighting the findings of these relatively small series with direct comparisons but often incongruous results. Based on our results, IDs should be considered as a therapeutic option for patients who might have comorbidities limiting them from undergoing inpatient decompressive surgery and for patients that are high risk for postoperative complications

-

(2013)

-

7

(eg, those with previous episodes of deep venous thromboses). However, patients who are able to tolerate a more invasive procedure and have longer projected follow-up may be better suited for undergoing laminectomy. Such findings highlight the need for further prospective comparative studies in the field both to assess effectiveness and complications, reoperations, and overall costs. Complications A recent review of IDs by Epstein cites an overall devicerelated complication rate of 11.6% to 38%, notably higher than our complication rate of 3.45%, although our followup period was relatively short at 90-days postoperatively [5]. Complications reported in the literature have consisted mostly of spinous process fractures, but also included new radicular deficits, device dislocations, and bilateral foot drop after ID implantation. Kim et al. prospectively followed 38 patients in whom 50 IDs were placed; they discovered 11 nondisplaced, radiographically occult spinous process fractures in 11 (29%) patients after postoperative computed tomography imaging [24]. Most fractures caused little to no acute localized pain. Other studies report atraumatic fractures of the intervening spinous processes in patients with adjacent, double-level ID implantation (‘‘sandwich phenomenon’’) [25] and stress fractures of the bilateral posterior facets after single-level implantation of L4-L5 IDs [26]. Many complications and reoperations after ID, however, may be prevented with better patient selection (ie, by selecting out those without severe stenosis, grade I spondylolisthesis, adjacent level disease, and/or severe foraminal stenosis) [5]. In such cases, laminectomy likely affords better results than IDs [27]. Reoperations Reoperation rates for IDs range from 4.6% to as high as 85% in studies with long-term follow-up [5,20]. Within our complete dataset, we report reoperation rates following ID placement in patients with cumulative follow-up times of at least 12 months and 18 months (21.4% and 23.1%, respectively). Cumulative reoperation rates of similar magnitude (23%) have been reported in 88 patients with longterm follow-up after decompressive surgery for LSS (mean follow-up time was 8.1 years), although our study shows significantly lower reoperation rates among laminectomy patients within a short follow-up period [28]. Within our propensity-matched analysis, patients in our ID cohort underwent significantly more reoperations compared with their laminectomy counterparts after 12 months of followup (12.6% vs. 5.8%, p5.03). This trend persisted in patients with at least 18 months of follow-up, but the greater rise in patients necessitating postlaminectomy reoperations compared with those requiring post-ID reoperations during this time rendered this difference not statistically significant (9.8% vs. 15%, p5.14). Thus, it appears that IDs do have

8

C.G. Patil et al. / The Spine Journal

a relatively high reoperation rate, although a majority of these reoperations take place relatively early within the follow-up period, whereas laminectomy patients may require reoperation relatively later. Cost-effectiveness Using SF-36 scores from studies providing levels I to III evidence as a surrogate marker for utility and 2008 Medicare national average reimbursement rates for costs, Burnett et al. developed a cost-effectiveness model to evaluate the optimal treatment strategy for patients with LSS at a 2-year time horizon [8]. The treatment arms consisted of conservative treatment, decompressive laminectomy, and X-STOP placement. Both surgical procedures were more costly than conservative treatment. For a single-level stenosis, laminectomy was found to be both more effective and more costly than XSTOP; however, laminectomy became the cost-effective choice between the two therapies if society was willing to pay more than $100,000 for purchasing an additional quality-adjusted life year. When a two-level stenosis was involved, laminectomy became the cost-effective choice at a willingness to pay threshold less than $60,000. In another recent study, Epstein reported the charges of implanting 31 X-STOP devices in 16 patients at a single institution in 2010 [5]. The average charge for X-STOP devices ranged from $17,000 for one-level procedures to $57,201 for three-level procedures. The total charge for placing all 31 devices was $576,407, excluding operating room and recovery room charges. These estimates may be placed in further context with Kondrashov et al.’s retrospective analysis, which found that a single-level X-STOP device was approximately $29,000 less expensive than a single-level laminectomy and a double-level X-STOP device was approximately $21,000 less expensive than a double-level laminectomy [21]. This difference in direct hospital costs was attributed to laminectomies requiring extended hospital stays, general anesthesia, and more surgical time. Our study demonstrates divergent results because the cumulative overall costs for ID placement was approximately $4,800 and $6,400 more expensive than laminectomy in patients with at least 12 and 18 months follow-up, respectively, favoring the use of laminectomy from a strict cost perspective.

-

(2013)

-

purposes of clinical research. Administrative databases often lack information on neurological symptoms and examination findings, the severity of spinal stenosis, the number of spinal levels affected by LSS, functional status, pain status, device-specific complications, the precise reason for reoperation, and quality of life measures. Furthermore, we cannot determine if the laminectomy cohort underwent an open or minimally invasive procedure based on the limitations of the database. Identification of which laminectomy approach was used is important because both approaches can influence covariates such as LOS and postoperative complications (especially infection rates) differently. Finally, the retrospective design of our study may theoretically predispose to some selection bias inherent to all such studies.

Conclusion Using the MarketScan database, we determined that patients undergoing laminectomy had longer in-hospital stays and were more likely to experience postoperative complications within 90 days. However, laminectomy had lower costs at index hospitalization and significantly less longterm reoperation rates compared with patients undergoing ID. Similarly, various previous studies have also reported mixed results with relative clinical effectiveness varying, in part, on duration of follow-up and outcomes assessed. An RCT or large prospective cohort study comparing decompressive surgery alone versus ID for stable LSS is needed to conclusively determine the precise role of IDs in the treatment of LSS. In the interim, IDs may remain a viable, less invasive approach for some elderly patients with persistent LSS, albeit with a relatively high reoperation rate. Decompressive laminectomy remains the definitive, longer-lasting treatment for LSS patients that are able to tolerate the surgical and anesthetic risks.

Appendix Supplementary material Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.spinee.2013.08.053.

Advantages and limitations The strength of large administrative databases such as MarketScan lies in their ability to reduce the probability of committing type II errors and provide ample power to delineate true disease trends and outcomes. In addition, the MarketScan database is able to track patients in the outpatient setting and outlines actual payments made by insurers, instead of reporting hospital charges that are often inflated and inaccurate. In spite of these strengths, our study has many limitations inherent in the use of any administrative database for

References [1] Johnsson KE. Lumbar spinal stenosis. A retrospective study of 163 cases in southern Sweden. Acta Orthop Scand 1995;66:403–5. [2] Atlas SJ, Delitto A. Spinal stenosis: surgical versus nonsurgical treatment. Clin Orthop Relat Res 2006;443:198–207. [3] Katz JN, Harris MB. Clinical practice. Lumbar spinal stenosis. N Engl J Med 2008;358:818–25. [4] Siebert E, Pr€uss H, Klingebiel R, et al. Lumbar spinal stenosis: syndrome, diagnostics and treatment. Nat Rev Neurol 2009;5:392–403. [5] Epstein NE. A review of interspinous fusion devices: high complication, reoperation rates, and costs with poor outcomes. Surg Neurol Int 2012;3:7.

C.G. Patil et al. / The Spine Journal [6] Lindsey DP, Swanson KE, Fuchs P, et al. The effects of an interspinous implant on the kinematics of the instrumented and adjacent levels in the lumbar spine. Spine 2003;28:2192–7. [7] Zucherman JF, Hsu KY, Hartjen CA, et al. A multicenter, prospective, randomized trial evaluating the X STOP interspinous process decompression system for the treatment of neurogenic intermittent claudication: two-year follow-up results. Spine 2005;30:1351–8. [8] Burnett MG, Stein SC, Bartels RH. Cost-effectiveness of current treatment strategies for lumbar spinal stenosis: nonsurgical care, laminectomy, and X-STOP. J Neurosurg Spine 2010;13:39–46. [9] Hansen LG, Chang S. Health research data for the real world: the Marketscan databases. White Paper, Research and Pharmaceutical Division. Sacramento, CA: Truven Health Analytics, 2012. [10] Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373–83. [11] Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992;45:613–9. [12] Hsu KY, Zucherman JF, Hartjen CA, et al. Quality of life of lumbar stenosis-treated patients in whom the X STOP interspinous device was implanted. J Neurosurg Spine 2006;5:500–7. [13] United States Department of Labor Bureau of Labor Statistics. Consumer price index. Available at: http://www.bls.gov/cpi/cpiovrvw. htm#item5. Accessed October 26, 2012. [14] Kabir SM, Gupta SR, Casey AT. Lumbar interspinous spacers: a systematic review of clinical and biomechanical evidence. Spine 2010;35:E1499–506. [15] Zucherman JF, Hsu KY, Hartjen CA, et al. A prospective randomized multi-center study for the treatment of lumbar spinal stenosis with the X STOP interspinous implant: 1-year results. Eur Spine J 2004;13:22–31. [16] Chou D, Lau D, Hermsmeyer J, et al. Efficacy of interspinous device versus surgical decompression in the treatment of lumbar spinal stenosis: a modified network analysis. Evid Based Spine Care J 2011;2:45–56. [17] Malmivaara A, Sl€atis P, Heli€ovaara M, et al. Surgical or nonoperative treatment for lumbar spinal stenosis? A randomized controlled trial. Spine 2007;32:1–8.

-

(2013)

-

9

[18] Weinstein JN, Tosteson TD, Lurie JD, et al. Surgical versus nonoperative treatment for lumbar spinal stenosis four-year results of the Spine Patient Outcomes Research Trial. Spine 2010;35:1329–38. [19] Weinstein JN, Tosteson TD, Lurie JD, et al. Surgical versus nonsurgical therapy for lumbar spinal stenosis. N Engl J Med 2008;358: 794–810. [20] Sobottke R, R€ollinghoff M, Siewe J, et al. Clinical outcomes and quality of life 1 year after open microsurgical decompression or implantation of an interspinous stand-alone spacer. Minim Invasive Neurosurg 2010;53:179–83. [21] Kondrashov DG, Hanibal M, Hsu KY, et al. X-STOP versus decompression for neurogenic claudication: economic and clinical analysis. Internet J Minim Invasive Spinal Tech 2006;1. http://dx.doi.org/10. 5580/1a34. [22] Richter A, Halm HF, Hauck M, et al. 2-year follow-up after decompressive surgery with and without implantation of an interspinous device for lumbar spinal stenosis: a prospective controlled study. J Spinal Disord Tech 2012 May 24. [23] Richter A, Sch€utz C, Hauck M, et al. Does an interspinous device (Coflex) improve the outcome of decompressive surgery in lumbar spinal stenosis? One-year follow up of a prospective case control study of 60 patients. Eur Spine J 2010;19:283–9. [24] Kim DH, Tantorski M, Shaw J, et al. Occult spinous process fractures associated with interspinous process spacers. Spine 2011;36: E1080–5. [25] Barbagallo GM, Corbino LA, Olindo G, et al. The ‘‘sandwich phenomenon’’: a rare complication in adjacent, double-level X-stop surgery: report of three cases and review of the literature. Spine 2010;35: E96–100. [26] Chung KJ, Hwang YS, Koh SH. Stress fracture of bilateral posterior facet after insertion of interspinous implant. Spine 2009;34:E380–3. [27] Postacchini R, Ferrari E, Cinotti G, et al. Aperius interspinous implant versus open surgical decompression in lumbar spinal stenosis. Spine J 2011;11:933–9. [28] Katz JN, Lipson SJ, Chang LC, et al. Seven- to 10-year outcome of decompressive surgery for degenerative lumbar spinal stenosis. Spine 1996;21:92–8.