Reoperation Rates After Surgery for Degenerative Cervical Spine Disease According to Different Surgical Procedures: National Population-Based Cohort Study

CERVICAL SPINE

Reoperation Rates After Surgery for Degenerative Cervical Spine Disease According to Different Surgical Procedures: National Population-based Cohort Study

Moon Soo Park, MD, PhD, Young-Su Ju, MD,y Seong-Hwan Moon, MD,z Tae-Hwan Kim, MD, Jae Keun Oh, MD,§ Melvin C. Makhni, MD,{ and K. Daniel Riew, MD{

cervical surgeries from January 2009 to June 2014. We separated Study Design. National population-based cohort study. patients into three groups based on surgical procedures: dis- Objective. To evaluate reoperation rates of cervical spine cectomy or corpectomy with anterior fusion, laminoplasty, or surgery for cervical degenerative conditions utilizing a national laminectomy with posterior fusion. Age, sex, presence of population database. Summary of Background Data. There is an inherently low diabetes, osteoporosis, associated comorbidities, number of incidence of reoperation after surgery for cervical degenerative operated cervical disc levels, and hospital types were considered disease. Therefore, it is difficult to sufficiently power studies to potential confounding factors. Reoperation rates were analyzed detect differences between reoperation rates of different cervical over early and late periods. Results. The reoperation rate over the entire follow-up period surgical procedures. National population-based databases pro- was 3.31%. Overall, the reoperation rate was significantly vide large, longitudinally followed cohorts that may help over- higher after laminectomy with posterior fusion or laminoplasty come this challenge. Methods. We used the Korean Health Insurance Review and than after discectomy or corpectomy with anterior fusion. Assessment Service national database to select our study A similar pattern was seen during the late period. In the population. We included patients with diagnosis of cervical early period, rates were higher only after laminectomy spondylotic radiculopathy or myelopathy who underwent with posterior fusion than after discectomy or corpectomy with anterior fusion. Sex, presence of diabetes, associated comorbidities, and hospital types were noted to be risk factors From the Department of Orthopedic Surgery, Hallym University Sacred for reoperation. Heart Hospital, Medical College of Hallym University, Gyeonggi-do, Conclusion. The reoperation rate was higher after laminectomy y Republic of Korea; Department of Occupational and Environmental Medi- with posterior fusion or laminoplasty. Given clinical scenarios in cine, Hallym University Sacred Heart Hospital, Medical College of Hallym University, Gyeonggi-do, Republic of Korea; zDepartment of Orthopedic which either anterior or posterior approaches can be utilized, Surgery, Yonsei University College of Medicine, Seoul, Republic of Korea; risk of reoperation can be another variable to consider in § Department of Neurosurgery, Hallym University Sacred Heart Hospital, surgical planning and patient education. Medical College of Hallym University, Gyeonggi-do, Republic of Korea; and {Department of Orthopedic Surgery, Columbia University, The Spine Key words: cervical, nationwide database, reoperation, Hospital at NY-Presbyterian, Allen Hospital, New York, NY. spondylosis, surgical procedure. Acknowledgment date: October 19, 2015. First revision date: January 4, Level of Evidence: 3 2016. Second revision date: January 20, 2016. Third revision date: February Spine 2016;41:1484–1492 3, 2016. Acceptance date: March 2, 2016. The manuscript submitted does not contain information about medical device(s)/drug(s). everal studies have been performed analyzing reop- Hallym University Research Fund 2014 (HURF-2014-28) was received in eration rates for surgeries on the degenerative cervical support of this work. spine.1–4 However, because of the low incidence of Relevant financial activities outside the submitted work: royalties, stocks, S reoperation, many of these are likely underpowered to and travel/accommodations/meeting expenses. detect significant differences between types of procedures. Address correspondence and reprint requests to Moon Soo Park, MD, Department of Orthopedic Surgery, Hallym University Sacred Heart Hos- National population-based studies have evolved to over- pital, Medical College of Hallym University, 896, Pyeongchon-dong, Don- come this challenge of insufficient sample size. National gan-gu, Anyang-si, Gyeonggi-do, 431-070, Republic of Korea; databases have the benefits of including large numbers of E-mail: [email protected] patients that can be followed longitudinally through their DOI: 10.1097/BRS.0000000000001581 unique resident registration numbers, and they can contain a 1484 www.spinejournal.com October 2016 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. CERVICAL SPINE Reoperation Cervical Surgical Procedures Park et al

Figure 1. Cohort definition.

wide variety of data points including reoperation events. spondylotic radiculopathy (diagnosis codes: M4720, They are also less subject to selection or nonresponse bias M4722, M4723, M4729, M501, M502, M503, M508, than case-series studies. M519, M5410, M5411, M5412, M5413, and M5419) or Population-based studies have been performed about cervical spondylotic myelopathy (diagnosis codes: G951- reoperation rates after surgeries for the degenerative lumbar G959, G992, M4710-M4719, M4800-M4803, M4809, spine5–9 and for the degenerative cervical spine.9–13 How- M500, M510, M9951, and M9971). These subjects were ever, to the best of our knowledge none have compared selected if they had any of the following primary procedures reoperation rates for common surgical procedures per- between January 1, 2009 and June 30, 2014: (i) discectomy formed on the degenerative cervical spine such as anterior (procedure code: N1491) with anterior fusion (procedure cervical fusion after discectomy or corpectomy, posterior code: N2463), (ii) corpectomy (procedure code: N0451), laminectomy with fusion, and laminoplasty. (iii) laminectomy (procedure code: N1497) with/without The purpose of this study was to evaluate reoperation posterior fusion (procedure code: N2469), or (iv) lamino- rates of cervical spine surgery for cervical degenerative plasty (procedure code: N2491, N2492). The patients’ res- conditions utilizing a national population database. ident registration numbers were encrypted for privacy. We identified 13,191 patients who underwent cervical MATERIAL AND METHODS spine surgery in 2009 from the cohort of patients (Figure 1). This study was approved by the institutional review board Among them, we excluded patients who had died during the (IBR) at the institution of the corresponding author (IRB follow-up period (causes of death were not recorded) or if number: 2015-I023). they were younger than 20-years old. Patients were also excluded if they had history of cervical surgery within the Data Source preceding 4 years (2005–2008), had undergone multiple of The Korean Health Insurance Review and Assessment Service the above procedures, or those who had surgeries that were (HIRA) nationaldatabase is a national, prospectively collected not specified. In addition, patients were excluded from the database of roughly 51 million people enrolled in the Republic study if they had concomitant disease of cervical spine of Korea. It contains all inpatient and outpatient data reported (including trauma, tumor, infectious spondylitis, inflamma- by diagnosis and procedure codes. The diagnosis codes are tory spondylitis of gout, rheumatoid arthritis, ankylosing standardized according to the Korean Classification of Dis- spondylitis, ossification of posterior longitudinal ligament ease, 5th version, which follows the International Classifi- (OPLL), congenital anomaly of Arnold Chiari malfor- cation of Disease, 10th edition (ICD-10). mation, Klippel-Feil syndrome, and unspecified diagnosis). The final study population was 9071 patients (Figure 1). All Study Population Selection and Design patients in the final study population whose operations We searched the HIRA national database to identify patients occurred in 2009 were evaluated for a minimum of 4.5 who had a primary diagnosis of degenerative cervical years, until June 30, 2014. All patients in the current study

Spine www.spinejournal.com 1485 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. CERVICAL SPINE Reoperation Cervical Surgical Procedures Park et al had surgery in 2009 and were followed for at least 4.5 years during the follow-up period. Reoperations were identified through June 30, 2014 to account for rates of reoperation in by the presence of any of the aforementioned procedure this time period. codes recorded after the index procedure code (N1491, Subjects were separated into three groups based on the N0451, N2463, N1497, N2469, N2491, and N2492). respective procedure codes: discectomy or corpectomy with Therefore, reoperation included cervical operations per- anterior fusion (Group 1), laminectomy with posterior formed at both the original or different levels. We have fusion (Group 2), or laminoplasty (Group 3). Our goal excluded cases that had more than two reoperations within was to determine changes in the reoperation rate over time the relatively short follow-up period of 4.5 years; we did not and to compare the reoperation rates of different surgical feel that these represented the natural history after cervical procedures by adjusting for confounding variables. surgery and may likely have been because of other technical considerations. For the statistical analysis and technical Surgical Indications convenience, we used January 1, 2009, which is the first Nearly all hospitals in Korea follow the Korean National Health date in our data collection period, and June 30, 2014, which Insurance Corporation requirements for reimbursement is the last date. because hospitals could not get the reimbursement if they do The reoperation rates were analyzed with two different not follow its requirements.6 For cervical radiculopathy, these time intervals: <90 postoperative days (early period) and regulations require intractable pain despite nonsurgical treat- greater than or equal to 90 postoperative days (late period) ment for at least 6 weeks or associated neurologic deficit. The to help differentiate between likely etiologies for reopera- regulations for cervical myelopathy are neurologic deficits tions.6,18 Baseline characteristics of the subjects were com- attributed to this diagnosis. Therefore, these Korean National pared using x2 test or ANOVA. Statistical analysis for Health Insurance Corporation requirements were considered as comparison of surgical procedures was performed using the surgical indications for patients in this cohort. Cox proportional hazards regression modeling. We analyzed data with SAS software (ver.6.1; SAS Institute, Inc., Confounding Factors Cary, NC). The statistical significance level was set at Age, sex, presence of diabetes, osteoporosis, number of P < 0.05. operated cervical disc levels, hospital types, and medical comorbidities were considered potential confounding fac- RESULTS tors in our analysis. Comorbidities were assessed according Discectomy or corpectomy with anterior fusion was the most to the ‘‘ International Classification of Disease, Ninth edi- common procedure in our cohort (89.77%, Table 1). Mean tion, Clinical Modification (ICD-9-CM) and ICD-10 coding patient age was 52.28 11.12 years: 37% were female (Table algorithms for Charlson Comorbidities’’ proposed by Quan 1). Age, sex, presence of diabetes, comorbidity, number of et al.14 If there were more than four distinct hospitalization operated cervical disc levels, and hospital types were different primary or secondary diagnoses in 2009, the patients were between the three groups (P < 0.0001, Table 1). Laminec- regarded as having comorbidities.6,7 Diabetes is a known tomy with posterior fusion or laminoplasty at one cervical risk factor for reoperation that increases complication rates disc level was very rare compared with cervical discectomy and inhibits functional recovery.6,15–17 with anterior fusion at one cervical disc level (Table 1). They In Korea, hospital types are determined by law.6 General were performed in the limited patients with cervical pathol- hospitals have at least seven departments, such as internal ogy at C2-C3 or the patients with previous surgeries of medicine, general surgery, obstetrics and gynecology, anterior cervical soft tissue including thyroid gland. Comor- pediatrics, diagnostic radiology, anesthesiology, pathology, bidities were found in 30.29% of the study population. Liver and laboratory medicine. They also must have at least one disease was the most common comorbidity (15.73%), fol- board-certified doctor in each department with more than lowed by chronic pulmonary disease (7.34%), and myo- 99 beds. Tertiary-referral hospitals are distinguished from cardial infarction (4.98%). general hospitals by having at least 20 departments. In In the entire follow-up period, 3.31% of the study popu- addition to the characteristics of general hospitals, terti- lation underwent reoperations (Table 2). The cumulative ary-referral hospitals also have residency training programs, reoperation rate was 1.03% at 3 months, 1.55% at 1 year, at least five operating rooms, and a variety of diagnostic and 3.08% at 4 years (Table 2). The cumulative incidence of tools of computed tomography, magnetic resonance imag- reoperation was highest after laminectomy with posterior ing, electromyography, angiography, gamma camera radi- fusion (12.48%), followed by laminoplasty (7.93%), and ography, and Holter cardiac monitoring. Hospitals are discectomy or corpectomy with anterior fusion (2.48%) healthcare systems that do not have essential departments, (Table 3). or have between 30 and 99 beds. Private clinics have fewer than 30 beds. Reoperations in the Entire Follow-up Period The unadjusted reoperation rate was significantly higher Statistical Analysis after laminoplasty or laminectomy with posterior fusion A time to event (reoperation) survival analysis was per- than after discectomy or corpectomy with anterior fusion formed. The primary endpoint was any cervical reoperation (Table 4). Sex, presence of diabetes, associated comorbidities,

1486 www.spinejournal.com October 2016 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. CERVICAL SPINE Reoperation Cervical Surgical Procedures Park et al

TABLE 1. The Characteristics of Study Population Discectomy or Laminectomy Corpectomy With With Posterior All Patients Anterior Fusion Fusion Laminoplasty P Number (%) 9071 8143 (89.77%) 537 (5.92%) 391 (4.31%) Age (yrs) <0.0001 20–29 147 (1.62%) 138 (1.69%) 6 (1.12%) 3 (0.77%) 30–39 899 (9.91%) 863 (10.60%) 25 (4.66%) 11 (2.81%) 40–49 2856 (31.48%) 2687 (33.00%) 144 (21.23%) 55 (14.07%) 50–59 2846 (31.37%) 2577 (31.65%) 162 (30.17%) 107 (27.37%) 60–69 1630 (17.97%) 1370 (16.82%) 138 (25.70%) 122 (31.20%) 70 693 (7.64%) 508 (6.24%) 92 (17.13%) 93 (23.79%) Mean age (SD) 52.28 11.12 51.56 10.82 57.19 11.71 60.54 11.34 <0.0001 Female sex, n 3412 (37.61%) 3139 (38.55%) 155 (28.86%) 118 (30.18%) <0.0001 Diabetes, n 1566 (17.26%) 1345 (16.52%) 109 (20.30%) 112 (28.64%) <0.0001 Osteoporosis, n 35 (0.39%) 34 (0.42%) 1 (0.19%) 0 (0.00%) NS Comorbidity, n 2748 (30.29%) 2427 (29.80%) 176 (32.77%) 145 (37.08%) 0.0040 Cervical level One level 7987 (88.05%) 7549 (92.71%) 425 (79.14%) 13 (3.32%) <0.0001 Two level 1040 (11.47%) 581 (7.13%) 111 (20.67%) 348 (89.00%) Three level 44 (0.49%) 13 (0.16%) 1 (0.19%) 30 (7.67%) Hospital types Tertiary-referral 2456 (27.08%) 2149 (26.39%) 113 (21.04%) 194 (49.62%) <0.0001 hospital General hospital 2360 (26.02%) 2078 (25.52%) 149 (27.75%) 133 (34.02%) Hospital 4117 (45.39%) 3787 (46.51%) 268 (49.51%) 62 (15.86%) Clinic 138 (1.52%) 129 (1.58%) 7 (1.30%) 2 (0.51%) not significant.

and hospital types were detected to be significant confound- presence of diabetes, associated comorbidities, and hospital ing factors by Cox regression analysis (Table 4). After con- types were all found to significantly affect risk for reoperation trolling for these confounders, the adjusted reoperation rates (female sex: P < 0.0001, hazard ratio ¼ 0.456, 95% CI were still significantly higher after laminoplasty or laminec- 0.343–0.605; diabetes: P < 0.0001, hazard ratio ¼ 1.756, tomy with posterior fusion than after discectomy or corpec- 95% CI 1.366–2.258; comorbidities: P ¼ 0.0025, hazard tomy with anterior fusion (laminectomy with posterior ratio ¼ 1.435, 95% CI 1.136–1.813; Hospital: P ¼ 0.0352, fusion: P < 0.0001, hazard ratio ¼ 4.704, 95% confidence hazard ratio ¼ 0.728, 95% CI 0.542–0.978; Clinic: interval, CI 3.547–6.239; laminoplasty: P < 0.0001, hazard P ¼ 0.0387, hazard ratio ¼ 1.347, 95% CI 1.016–1.786, ratio ¼ 2.396, 95% CI 1.622–3.540, Table 5, Figure 2). Sex, Table 5).

TABLE 2. Reoperation Rates With Time Cumulative Data Postoperative Number of Reoperation Number of Reoperation Period Reoperations (n) Rate (%) Reoperations (n) Rate (%) <1 mo 69 0.76 69 0.76 1–2 mos 12 0.13 81 0.89 2–3 mos 12 0.13 93 1.03 3–6 mos 16 0.18 109 1.20 0.5–1 yr 32 0.35 141 1.55 1–2 yrs 46 0.51 187 2.06 2–3 yrs 46 0.51 233 2.57 3–4 yrs 46 0.51 279 3.08 4 yrs 21 0.23 300 3.31

TABLE 3. Cumulative Reoperation Rates Discectomy or Laminectomy Corpectomy With With Posterior Peroid n All Patients Anterior Fusion Fusion Laminoplasty <90 days Operations 9071 8143 537 391 Reoperations 93 (1.03%) 58 (0.71%) 27 (5.03%) 8 (2.05%) 90 days to 4 yrs Operations 8978 8085 510 383 Reoperations 186 (2.05%) 128 (1.57%) 36 (6.70%) 22 (5.63%) 4 yrs Operations 8792 7957 474 361 Reoperations 21 (0.23%) 16 (0.20%) 4 (0.74%) 1 (0.26%) Total period Operations 9071 8143 537 391 Reoperations 300 (3.31%) 202 (2.48%) 67 (12.48%) 31 (7.93%)

Early Reoperations (Less Than 90 Days After Index diagnosis of cervical spondylotic radiculopathy or myelop- Procedure) athy and evaluated rates of reoperation over time after The unadjusted reoperation rate was significantly higher after anterior discectomy or corpectomy with fusion versus laminoplasty or laminectomy with posterior fusion than after posterior laminectomy with fusion versus laminoplasty. discectomy or corpectomy with anterior fusion (Table 4). Sex, Overall, reoperation rates were higher after laminoplasty presence of diabetes, and associated comorbidities were and laminectomy with posterior fusion than after anterior detected to be significant confounding factors by Cox cervical discectomy or corpectomy and fusion. Male sex was regression analysis (Table 4). The adjusted reoperation rate a risk factor for reoperation in the early and late post- was still significantly higher after laminectomy with posterior operative periods, whereas the presence of diabetes fusion than after discectomy or corpectomy with anterior and having associated comorbidities correlated to early fusion (laminectomy with posterior fusion: P < 0.0001, hazard reoperation. ratio ¼ 5.622, 95% CI 3.528–8.959, Table 4, Figure 3). Sex, King et al10 analyzed the Washington State Inpatient presence of diabetes, and having associated comorbidities were Database and found that the reoperation rates after surgery all risk factors for early reoperation (female sex: P < 0.0001, for degenerative cervical disease from 1998 to 2012 was hazard ratio ¼ 0.185, 95% CI 0.090–0.384; diabetes: 5.6%. They found that reoperation rates were lower in P < 0.0001, hazard ratio ¼ 3.267, 95% CI 2.137–4.995; females and those patients undergoing anterior surgery comorbidities: P ¼ 0.002, hazard ratio ¼ 1.928, 95% CI and fusion surgery, which are in accordance with our 1.271–2.927, Table 5). results.10 They also found higher reoperation rates in patients younger than 50 years old compared with those Late Reoperations (90 Days or Later Than Index older than 70 years old, and higher reoperation rates for Procedure) patients with the diagnosis of myelopathy.10 Surgical pro- The unadjusted reoperation rate was significantly higher after cedures analyses were ambiguous because they simply classi- laminectomy with posterior fusion or laminoplasty than after fied them into the presence of fusion and the direction of discectomy or corpectomy with anterior fusion (Table 4). Sex, surgical approach (ventral or dorsal).10 They did not evalu- presence of diabetes, associated comorbidities, and hospital ate the presence of comorbidities or the number of levels types were detected to be significant confounding factors by operated on or their interaction with surgical procedures to Cox regression analysis (Table 4). The adjusted reoperation influence the reoperation rates.10 In additon, they did not rate was still significantly higher after laminectomy with compare the differences between reoperation rates in the posterior fusion or laminoplasty than after discectomy or early and late periods.10 corpectomy with anterior fusion (laminectomy with posterior Adogwa et al11 evaluated reoperation rates of cervical fusion: P < 0.0001, hazard ratio ¼ 4.226, 95% CI 2.956– procedures using the MarketScan Database from 2000 to 6.041; laminoplasty: P < 0.0001, hazard ratio ¼ 2.812, 95% 2009. They found that the reoperation rate of cervical CI 1.780–4.442, Table 5, Figure 4). Sex was a risk factor laminectomy with fusion and cervical laminoplasty over for late reoperation (female sex: P ¼ 0.0015, hazard 2-year follow up was 9.77% and 7.36%, respectively.11 ratio ¼ 0.600, 95% CI 0.438–0.822, Table 5). This difference was not statistically significant.11 Although this serves as an interesting comparison between posterior DISCUSSION procedures, they could not evaluate the difference between There have been rare population-based studies analyzing the anterior and posterior procedures because of exclusion of differences between reoperation rates of different surgical patients who underwent anterior cervical surgical pro- procedures for the degenerative cervical spine. We analyzed cedures. Further, this study did not delve into the risk factors patients from a national population of patients with the for reoperation.

1488 www.spinejournal.com October 2016 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. CERVICAL SPINE Reoperation Cervical Surgical Procedures Park et al

TABLE 4. Comparison Between Surgical Procedures Based on Unadjusted Value of Cox Regression Analysis

Entire Period (n ¼ 9071) 1–89 days (n ¼ 9071) 90 days (n ¼ 8978)

P HR 95% CI P HR 95% CI P HR 95% CI

Surgical procedures Discectomy or corpectomy 1.000 1.000 1.000 with anterior fusion Laminectomy with <0.0001 5.322 (4.037, 7.015) <0.0001 7.157 (4.534, 11.3) <0.0001 4.544 (3.201, 6.450) posterior fusion Laminoplasty <0.0001 3.298 (2.260, 4.813) 0.005 2.879 (1.375, 6.029) <0.0001 3.447 (2.239, 5.400) Age (yrs) 20–29 1.000 1.000 1.000 30–39 0.319 0.568 (0.187, 1.727) 0.286 0.409 (0.079, 2.110) 0.685 0.728 (0.157, 3.370) 40–49 0.451 0.676 (0.245, 1.868) 0.061 0.232 (0.050, 1.073) 0.877 1.118 (0.271, 4.614) 50–59 0.326 1.646 (0.608, 4.454) 0.893 0.907 (0.218, 3.770) 0.224 2.386 (0.588, 9.685) 60–69 0.323 1.661 (0.607, 4.545) 0.550 1.546 (0.371, 6.433) 0.428 1.776 (0.429, 7.354) 70 0.416 1.543 (0.542, 4.388) 0.839 0.851 (0.181, 4.009) 0.278 2.234 (0.524, 9.527) Sex Male 1.000 1.000 1.000 Female <0.0001 0.417 (0.315, 0.552) <0.0001 0.155 (0.075, 0.321) 0.0003 0.560 (0.410, 0.765) Diabetes Yes <0.0001 2.381 (1.871, 3.031) <0.0001 4.742 (3.158, 7.120) 0.0016 1.656 (1.210, 2.267) No 1.000 1.000 1.000 Osteoporosis Yes 0.877 0.857 (0.120, 6.101) 0.979 0.000 (0.000, .) 0.829 1.241 (0.174, 8.852) No 1.000 1.000 1.000 Comorbidities Yes <0.0001 1.709 (1.359, 2.149) <0.0001 2.690 (1.790, 4.044) 0.024 1.382 (1.043, 1.832) No 1.000 1.000 1.000 Discectomy One level 1.000 1.000 1.000 Two level 0.220 0.601 (0.267, 1.355) 0.302 0.352 (0.049, 2.548) 0.433 0.700 (0.287, 1.709) Three level 0.978 0.000 (0.000, 0.989 0.000 (0.000, .) 0.982 0.000 0.000 1.73E227) Corpectomy One level 1.000 1.000 1.000 Two level 0.994 0.000 (0.000, .) 0.996 0.000 (0.000, .) 0.995 0.000 (0.000, .) Laminectomy with posterior fusion One level 1.000 1.000 1.000 Two level 0.004 0.228 (0.083, 0.627) 0.985 0.000 (0.000, .) 0.074 0.389 (0.139, 1.904) Three level 0.987 0.000 (0.000, .) 0.999 0.000 (0.000, .) 0.990 0.000 (0.000, .) Laminoplasty One level 1.000 1.000 1.000 Two level 0.925 1.100 (0.150, 8.076) 0.995 1229740 (0.000, .) 0.861 0.836 (0.113, 6.205) Three level 0.555 0.434 (0.027, 6.941) 0.995 2036600 (0.000, .) 0.991 0.000 (0.000, .) Hospital types Tertiary-referral hospital 1.000 1.000 1.000 General hospital 0.026 1.373 (1.038, 1.816) 0.964 1.047 (0.139, 7.867) 0.172 0.254 (0.035, 1.819) Hospital 0.006 0.669 (0.501, 0.893) 0.097 0.561 (0.284, 1.111) 0.025 0.695 (0.505, 0.956) Clinic 0.209 0.407 (0.100, 1.655) 0.964 1.047 (0.139, 7.867) 0.172 0.254 (0.035, 1.819)

Wang et al19 also performed a population analysis of years old compared with those older than 75-years old patients with degenerative cervical disease. The authors had higher complication rates.19 However, they did not used the National Impatient Sample Database of the United evaluate the reoperation rates. States to evaluate complication rates of surgery from 1992 The number of operated cervical disc levels was not a risk to 2001. As in our study, lower complication rates were factor in the current study. Similar to the current study, observed in anterior cervical procedures. Patients with the Hilibrand et al20 found the patients who had a multilevel diagnosis of myelopathy, and patients younger than 34- cervical arthrodesis were less likely to have symptomatic

TABLE 5. Comparison Between Surgical Procedures Based on Adjusted Value of Cox Proportional Hazards Regression Modeling

Entire Period (n ¼ 9071) 1–89 days (n ¼ 9071) 90 days (n ¼ 8978)

P HR 95% CI P HR 95% CI P HR 95% CI

Surgical procedures Discectomy or 1.000 1.000 1.000 corpectomy with anterior fusion Laminectomy with <0.0001 4.704 (3.547, 6.239) <0.0001 5.622 (3.528, 8.959) <0.0001 4.226 (2.956, 6.041) posterior fusion Laminoplasty <0.0001 2.396 (1.622, 3.540) 0.100 1.878 (0.886, 3.981) <0.0001 2.812 (1.780, 4.442) Age (yrs) 20–29 1.000 1.000 1.000 30–39 0.430 0.639 (0.210, 1.943) 0.285 0.409 (0.079, 2.108) 0.750 0.779 (0.168, 3.609) 40–49 0.571 0.745 (0.269, 2.061) 0.056 0.224 (0.048, 1.038) 0.803 1.198 (0.290, 4.948) 50–59 0.362 1.590 (0.587, 4.312) 0.628 0.702 (0.168, 2.938) 0.245 2.297 (0.565, 9.342) 60–69 0.602 1.309 (0.476, 3.600) 0.897 0.910 (0.215, 3.846) 0.610 1.450 (0.348, 6.034) 70 0.923 0.949 (0.331, 2.723) 0.274 0.416 (0.086, 2.004) 0.582 1.506 (0.350, 6.482) Sex Male 1.000 1.000 1.000 Female <0.0001 0.456 (0.343, 0.605) <0.0001 0.185 (0.090, 0.384) 0.0015 0.600 (0.438, 0.822) Diabetes Yes <0.0001 1.756 (1.366, 2.258) <0.0001 3.267 (2.137, 4.995) 0.062 1.361 (0.984, 1.882) No 1.000 1.000 1.000 Comorbidities Yes 0.0025 1.435 (1.136, 1.813) 0.002 1.928 (1.271, 2.927) 0.117 1.258 (0.944, 1.675) No 1.000 1.000 1.000 Hospital types Tertiary-referral 1.000 1.000 1.000 hospital General hospital 0.227 0.421 (0.104, 1.713) 0.289 0.824 (0.576, 1.179) Hospital 0.0352 0.728 (0.542, 0.978) 0.091 0.754 (0.543, 1.046) Clinic 0.0387 1.347 (1.016, 1.786) 0.187 0.265 (0.037, 1.908)

adjacent segmental disease than were those who had had a single-level arthrodesis. Reoperation rates were higher with laminoplasty, laminectomy, and posterior fusion than after anterior cervical discectomy or corpectomy and fusion in the current study. It may be that posterior cervical surgeries might have more surgery-related complications leading to reoperations, such as postoperative wound infection or hardware failure of rod-screw system than anterior cervical surgeries. Given clinical scenarios in which either anterior or posterior approaches can be utilized, risk of reoperation can be another variable to consider in surgical planning and patient education. As with any study, our investigation has several limitations. Clinical information about pain, quality of life, func- tion, and neurologic status was not available in this administrative dataset.10 Therefore, reoperation cannot be attributed solely to poor initial surgical outcome. Radiologic information and level of complexity of surgeries were also not available; these restrictions are inherent to administra- Figure 2. Adjusted cumulative reoperation rate of surgical pro- tive databases.21 However, our large sample size of our cedures during the entire follow-up period. The adjusted reoperation rate was significantly higher after laminectomy with posterior fusion cohort allows for estimation of average reoperation rates or laminoplasty than after anterior cervical discectomy or corpec- that are generalizable to the entire population. In addition, tomy and fusion. the fee-for-service reimbursement system in Korea may

1490 www.spinejournal.com October 2016 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. CERVICAL SPINE Reoperation Cervical Surgical Procedures Park et al

healthcare system is a modern one using the latest surgical techniques that are not appreciably different than those utilized in most first world countries. In addition, the diagnoses of degenerative cervical spondylotic radiculopathy and myelopathy may be risk factors for reoperations. Unfortunately, we could not perform the statistical analysis in the current study because of the limited number of reoperations in the subgroups. We are going to evaluate diagnosis as a risk factor for reoperation in the next study. In addition, smoking is a well-known risk factor for reoperation. However, the current national administrative database did not have information about smoking. Despite these limitations this study, to the best of our knowledge, represents the first population-based analysis of the reoperation rates after degenerative cervical spine surgeries according to different specific surgical procedures such as anterior discectomy or corpectomy with fusion, posterior Figure 3. Adjusted cumulative reoperation rate of surgical procedures laminectomy with fusion, and laminoplasty. during first 89 days after index operation. The adjusted reoperation rate was significantly higher after laminectomy with posterior fusion In conclusion, reoperation rates were higher after laminec- or laminoplasty than after anterior cervical discectomy or corpectomy tomy with posterior fusion or laminoplasty than after anterior and fusion during first 89 days after index operation. cervical discectomy or corpectomy and fusion for cervical spondylotic radiculopathy or myelopathy in the national population of patients in both the early and late periods. contribute to the high incidence of comorbidities in our cohort, as the presence of any comorbidity needs to be listed when submitting claims for insurance-covered medical fees Key Points 6 to HIRA. We have also excluded the patients who died Because of the low incidence of reoperation, during follow-up period; this might reduce the reported many clinical studies are likely underpowered to reoperation rates because mortality associated with surgery detect significant differences between different for degenerative cervical disease was 0.14%.19 Another cervical surgical procedures. limitation is that this was a population study of Koreans National population-based studies have evolved operated on by Korean surgeons and it may not be general- to overcome this challenge of insufficient izable to other countries or regions. However, the Korean sample size. None have compared reoperation rates for common surgical procedures performed on the degenerative cervical spine such as anterior cervical fusion after discectomy or corpectomy, posterior laminectomy with fusion, and laminoplasty in a national population-based studies. In a national population-based cohort study, the reoperation rate was higher after laminectomy with posterior fusion or laminoplasty than after discectomy or corpectomy with anterior fusion.

References 1. Coric D, Branch CL Jr, Jenkins JD. Revision of anterior cervical pseudoarthrosis with anterior allograft fusion and plating. J Neu- rosurg 1997;86:969–74. 2. Lee JC, Lee SH, Peters C, et al. Risk-factor analysis of adjacent- segment pathology requiring surgery following anterior, posterior, Figure 4. Adjusted cumulative reoperation rate of surgical pro- fusion, and nonfusion cervical spine operations: survivorship analysis of 1358 patients. J Bone Joint Surg Am cedures 90 days after index operation. The adjusted reoperation 2014;96:1761–7. rate was significantly higher after laminectomy with posterior fusion 3. Geisler FH, Caspar W, Pitzen T, et al. Reoperation in patients after or laminoplasty than after anterior cervical discectomy or corpec- anterior cervical plate stabilization in degenerative disease. Spine tomy and fusion 90 days after index operation. (Phila Pa 1976) 1998;23:911–20.

4. Singh K, Phillips FM, Park DK, et al. Factors affecting reoperations 13. Radcliff K, Zigler J, Zigler J. Costs of cervical disc replacement after anterior cervical discectomy and fusion within and outside of versus anterior cervical discectomy and fusion for treatment of a Federal Drug Administration investigational device exemption single-level cervical disc disease: an analysis of the Blue Health cervical disc replacement trial. Spine J 2012;12:372–8. Intelligence database for acute and long-term costs and compli- 5. Kim CH, Chung CK, Park CS, et al. Reoperation rate after surgery cations. Spine (Phila Pa 1976) 2015;40:521–9. for lumbar spinal stenosis without spondylolisthesis: a nation-wide 14. Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for cohort study. Spine J 2013;13:1230–7. defining comorbidities in ICD-9-CM and ICD-10 administrative 6. Kim CH, Chung CK, Park CS, et al. Reoperation rate after surgery data. Med Care 2005;43:1130–9. for lumbar herniated intervertebral disc disease: nationwide cohort 15. Takahashi S, Suzuki A, Toyoda H, et al. Characteristics of diabetes study. Spine (Phila Pa 1976) 2013;38:581–90. associated with poor improvements in clinical outcomes after 7. Hu RW, Jaglal S, Axcell T, et al. A population-based study of lumbar spine surgery. Spine (Phila Pa 1976) 2013;38:516–22. reoperations after back surgery. Spine (Phila Pa 1976) 1997;22: 16. Kim CH, Chung CK, Shin S, et al. The relationship between 2265–70. diabetes and the reoperation rate after lumbar spinal surgery: a 8. Malter AD, McNeney B, Loeser JD, et al. 5-year reoperation rates nationwide cohort study. Spine J 2015;15:866–74. after different types of lumbar spine surgery. Spine (Phila Pa 1976) 17. Boakye M, Patil CG, Ho C, et al. Cervical corpectomy: compli- 1998;23:814–20. cations and outcomes. Neurosurgery 2008;63:295–301. 9. Rajaee SS, Kanim LE, Bae HW. National trends in revision spinal 18. Martin BI, Mirza SK, Comstock BA, et al. Reoperation rates fusion in the USA: patient characteristics and complications. Bone following lumbar spine surgery and the influence of spinal fusion Joint J 2014;96:807–16. procedures. Spine (Phila Pa 1976) 2007;32:382–7. 10. King JT Jr, Abbed KM, Gould GC, et al. Cervical spine reoperation 19. Wang MC, Chan L, Maiman DJ, et al. Complications and rates and hospital resource utilization after initial surgery for mortality associated with cervical spine surgery for degenerative degenerative cervical spine disease in 12,338 patients in Wash- disease in the United States. Spine (Phila Pa 1976) 2007;32: ington State. Neurosurgery 2009;65:1011–22. 342–7. 11. Adogwa O, Huang K, Hazzard M, et al. Outcomes after cervical 20. Hilibrand AS, Carlson GD, Palumbo MA, et al. Radiculopathy and laminectomy with instrumented fusion versus expansile lamino- myelopathy at segments adjacent to the site of a previous anterior plasty: a propensity matched study of 3185 patients. J Clin Neuro- cervical arthrodesis. J Bone Joint Surg Am 1999;81: sci 2015;22:549–53. 519–28. 12. Cole T, Veeravagu A, Jiang B, et al. Usage of recombinant human 21. Campbell PG, Malone J, Yadla S, et al. Comparison of ICD-9- bone morphogenetic protein in cervical spine procedures: analysis based, retrospective, and prospective assessments of perioperative of the MarketScan longitudinal database. J Bone Joint Surg Am complications: assessment of accuracy in reporting. J Neurosurg 2014;96:1409–16. Spine 2011;14:16–22.