The research protocol was approved by the Institutional Review Board (IRB) of Soonchunhyang University Hospital (IRB number 2017-05-008) and performed in accordance with the Declaration of Helsinki. This was an anonymous observational study, thus the need for informed consent was waived.

Claims data from the Health Insurance Review and Assessment Service (HIRA; https://www.hira.or.kr) national database were used to identify a population of patients who received spinal treatments for LDD from 2007–2016. The claims data represent 46 million patients per year that account for 90% of the total population in South Korea [15]. The HIRA claim data are organized into five categories : 1) general information, 2) specific information regarding the health care provided, 3) diagnostic information, 4) outpatient prescription, and 5) health care provider information. The disease codes using the International Classification of Diseases, 10th version (ICD-10) were used.

The study cohort consisted of patients who were first diagnosed with LDD and received spinal surgery for LDD in 2009. To clarify the definition of new-onset, patients diagnosed with spinal surgery within the past 2 years (2007–2008) were excluded. In addition, patients who were diagnosed and received reoperation within 1 year were excluded. Reoperation was defined as repeated surgery performed in the lumbar spine, because we could not distinguish whether or not reoperation was performed on the same segment of the previous operation in this study. The patients were classified into three groups : spinal fusion group, discectomy group, and decompression group and the final number in the study cohort was 78241.

The Charlson comorbidity index (CCI) has been a useful tool for health researchers to measure comorbid disease status or case mix in health care databases [21]. The comorbidities were defined as numerous clinical conditions based on review of hospital charts and the relevance in the prediction of 1-year mortality assessed. In the present study, medical comorbidity was assessed according to the ICD-10 of the CCI. The CCI value was calculated using 17 comorbidities. Each comorbidity is assigned a score (weight) when computing the weighted Charlson index. Diabetes with complications, hemiplegia, renal disease, and cancer are assigned a score of 2; metastatic cancer and severe liver disease are assigned a score of 3; and human immunodeficiency virus is assigned a score of 6; the remaining comorbidities are assigned a score of 1 [21].

All statistical analyses in this study were performed using program R version 3.3.2 (www.r-project.org) and SAS Enterprise version 9.1.3 (SAS institute Inc., Cary, NC, USA). Survival analysis was used to compare the cumulative incidence of reoperations on patients and reoperation time was calculated as the number of years between the hospital admission day for the first surgery and second surgery. Censoring occurred when the patient reached the end of the follow-up period without a second surgery, which included patients who died outside the hospital or died of other causes during the follow-up period. Cox proportional hazard regression modeling was used to identify significant differences in sex, surgery, age, causative disease, and comorbidity. The hazard plots and tests of the Schoenfeld residuals were performed to determine whether the proportional assumptions of the Cox regression model were violated. A hazard ratio (HR) and a 95% confidence interval (CI) were calculated, and a Kaplan-Meier plot was performed for visualization. A two-tailed p-value <0.05 was considered statistically significant.

In 2009, 78241 patients underwent primary surgery for LDD; 46.7% were males and 53.3% were females. Among the patients, 26.0% were 65–77 years of age and 6.9% were 75 years of age or older. The percentage of patients who underwent spinal surgery for lumbar herniated intervertebral disc (L-HIVD) was 48.9%, for lumbar stenosis 35.9%, and for spondylolisthesis 12.8%. The surgical methods included decompression (69.5%), discectomy, (24.5%), and spinal fusion (5.9%; Table 1). The incidence of medical comorbidities was : pulmonary disease (22.3%), diabetes (20.3%), and peripheral vascular disease (17.6%; Table 2).

Among the 78241 patients who were followed up for 8 years, 10328 underwent reoperation for LDD (Table 1). The reoperation rate for LDD was 13.2%. Cumulative probability curve of reoperation was the concave type, which has the greatest reoperation rate in early follow-up, decreasing over time (Fig. 1). The percentage of males and females was 51.7% and 48.3%, respectively. The reason for reoperation was L-HIVD (51.8%), lumbar stenosis (37.1%), and spondylolisthesis (8.8%). The surgical methods were decompression (72.0%), discectomy, (23.2%), and spinal fusion (4.8%; Table 1).

Based on multivariate Cox regression analysis with HR estimates and CIs, sex, age, causative disease, and surgical method showed significant differences for LDD reoperation rate (Table 3). The reoperation rate in male patients was statistically higher (p<0.01) than in female patients (HR, 1.32; 95% CI, 1.27–1.37). Patients 55–74 years and 65–74 years of age had statistically higher reoperation rates (p<0.01) than patients under 35 years of age (55–64 years : HR, 1.41; 95% CI, 1.30–1.52; 65–74 years : HR, 1.39; 95% CI, 1.28–1.50). Lumbar stenosis and L-HIVD showed higher reoperation rate (p<0.01) than spondylolisthesis (lumbar stenosis : HR, 1.50; 95% CI, 1.40–1.61; L-HIVD : HR, 1.63; 95% CI, 1.52–1.76). Decompression and discectomy showed a higher reoperation rate (p<0.01) than fusion surgery (decompression : HR, 1.38; 95% CI, 1.26–1.51; discectomy : HR, 1.21; 95% CI, 1.10–1.33).

Based on univariate HR analysis, higher CCI score was associated with significantly higher reoperation rate (p<0.01; Fig. 2). Multivariate HR analysis of comorbidities indicated peripheral vascular disease (HR, 1.14; 95% CI, 1.08–1.20), pulmonary disease (HR, 1.20; 95% CI, 1.15–1.26), peptic ulcer (HR, 1.19; 95% CI, 1.09–1.31), diabetes (HR, 1.15; 95% CI, 1.10–1.21), and diabetes complications (HR, 1.17; 95% CI, 1.08–1.26), were associated with a significant increase in reoperation rate (p<0.01) (Table 4).

The reasons for LDD reoperation are surgery-related complications, worsening of surgical site, and adjacent segment degeneration (ASD). In previous studies, reoperation rates differed for each disease. The reported reoperation rates based on the disease were more than 20% in lumbar spondylolisthesis, 15–20% in lumbar stenosis, and less than 10% in lumbar disc herniation [10,12,14,19,20]. Generally, the reoperation rate was reportedly approximately 13% in patients with LDD over a 6-year period; reoperation rate for LDD was 13.2% over an 8-year period in the present study.

The risk factors for various reoperations have been reported in previous studies and include preexisting spinal stenosis, diabetes, fusion of two or more levels, blood loss, non-fusion surgery, decompression alone, preoperative scoliotic wedging/listhesis, use of antidepressants, no neurogenic claudication, higher body mass index (BMI), greater disc height, sex (male), facet degeneration, ASD, and Pfirrmann grading for disc degeneration [5,6,10,12,14,20,24].

In the present study, the reoperation rate for LDD was 1.32 times higher in males than in females. Sato et al. [20] suggested that male sex was the greatest independent risk factor for adjacent segmental disease in reoperation for degenerative spondylolisthesis. However, this result was inconsistent in other studies. The effects of sex on higher reoperation rates for LDD is not clearly understood. Assumedly, life activities associated with men's occupations cause theoretically increasing the risk of reoperation.

Surgical methods were also associated with reoperation rate. In the present study, the rate of reoperation for LDD was 1.21 times higher in discectomy and 1.38 times higher in decompression than in fusion. In previous studies, the difference in reoperation rate between decompression alone and decompression with fusion for LDD was controversial [4,5,20]. Decompression and discectomy are likely to increase the reoperation at the surgical site, and decompression with fusion is expected to increase reoperation for ASD.

Aging is a main risk factor for intervertebral disc degeneration (IDD) [2]. In previous studies, age was not always statistically significant as a risk factor for reoperation after lumbar surgery [6,24]. The reason may be because the shorter the observation period, the less effect of age, and the longer the observation period, the greater the likelihood of reoperation due to deterioration of degenerative change.

The comorbidity measures showed a significant relationship with increased risk of reoperation after total hip replacement [11]. In a study on the association between reoperation and comorbidity in LDD, only diabetes was evaluated [14]. The present study showed there was a significant relationship with increased risk of reoperation based on increased CCI scores, indicating a higher comorbidity burden measured using the CCI is associated with increased reoperation rate for LDD.

Risk factors for LDD reoperation may be associated with factors for spinal degeneration. In previous studies, a major cause of LDD reoperation was a preexisting and natural history of spine degenerative disease [6,19,24]. Therefore, the reoperation for LDD in the long term observation period is likely associated with risk factors for IDD. The associated factors for IDD have been reported as specific key genes, older age, obesity, BMI, smoking, diabetes, atherosclerosis, dyslipidemia, oxidative stress, inflammation, endplate defect, bone mineral density, occupation, and sporting activities [9,17,22]. Therefore, the long-term study of risk factors for LDD reoperation should be complemented with investigation of factors associated with spinal degeneration.

The claims data of the HIRA national database are useful for the generalization of the population due to representativeness of the total patient population in South Korea. Despite this strength, the present study had several limitations. First, the accuracy of diagnosis has been an issue due to the nature of claims data which are collected with the purpose of reimbursing healthcare services and not for clinical purposes [16]. The fee-for-service system may contribute to the high incidence of comorbidities because the presence of any comorbidity must be listed when submitting claims to HIRA for insurance-covered medical fees [14]. Second, data were insufficient to clarify the causes of reoperation for LDD such as complications of primary operation, recurrence of index level, and deteriorated degeneration at other levels. However, despite these limitations, this study provides additional information on reoperation for LDD that was not shown in previous studies.