Japanese 2011 nationwide survey on complications from spine surgery

This survey aimed to enroll all patients who underwent spine surgery in Japan during the one-year period from 1 January 2011 to 31 December 2011. The JSSR research team prepared a computerized questionnaire made up of two categories in order to capture clinicopathological information and surgical information. In January 2012, a recordable optical disc for data storage was sent to 1,105 surgeons who were certified for spine surgery by JSSR. This data was returned by the end of May 2012.

The clinicopathological variables that were investigated included patient information [age, gender, body weight, height, body mass index (BMI)], diabetes mellitus (DM), dialysis, corticosteroid use, disease-modifying anti-rheumatic drug (DMARD) therapy involving biological therapy, and Parkinson’s disease. Other requested information included the involved classification of spinal involvement, degeneration, deformity, ossified lesions, spondylolisthesis, inflammation, infection and tumors.

The requested surgical information included surgical approaches, intraoperative blood loss, operation time, surgical technology, decompression methods, fusion methods and instrumentation. The 22 reported items for intraoperative and postoperative complications in hospital are listed in Table 1. Information regarding the experience of the principal surgeon who operated on the patient was also collected. This was classified as 1–4 years of surgical experience after graduation, 5–9 years experience, 10–19 years experience, or 20 or more years of experience.

The above items were evaluated and compared with data from previous surveys. Complications were evaluated in relation to the surgical approach, instrumentation surgery, surgeon and BMI. Associations between intraoperative blood loss and the incidence of deep wound infection (DWI), epidural hematoma (EH), death and pulmonary embolism and/or thromboembolism (PE/TE) and operation time were also examined.

Statistical analyses were performed using Statcel 2. For categorical variables, cross-tabulations were made and a Chi square test was used for comparison of proportions. For continuous variables, statistical significance was assessed using the F test, Welch’s t test, Student’s t test for comparison of two means, and Spearman’s correlation coefficient for association between quantitative characteristics. A p value of <0.05 was considered significant in all the analyses. In order to evaluate factors associated with major intraoperative and postoperative complications, multiple logistic regression analyses were performed by use of a computer program: Statflex 6.0 (Artech Co., Ltd., Osaka, Japan. URL: http://​www.​statflex.​net/​).

This survey received approval from the institutional review board of Yamaguchi University Hospital.

The certified surgeons were distributed in 750 institutions nationwide and a response was achieved from 209 institutions (response rate 28 %). Of these, 63 (30 %) were university hospitals. Following the exclusion of cases that lacked clinicopathological or surgical information, data was available for a total of 31,380 patients, comprising 18,546 men, 12,747 women and 87 persons of unknown gender. This cohort was considerably larger than for both the 1994 (19,271 cases) and 2001 (16,157 cases) surveys [4, 5].

The mean age was 59.3 years (range 0–97 years) and the most frequent age was in the 70–79 year range (Fig. 1). This is older than the most common age reported in the 1994 and 2001 surveys [4, 5]. A total of 19,802 patients were aged 60 years or older, corresponding to 63.1 % of the overall cohort. This was higher than in the 1994 and 2001 surveys (37.3 and 49.0 %, respectively), demonstrating a marked increase in the proportion of elderly patients who underwent spine surgery (p < 0.05 for 2001 vs. 2011) [4, 5]. In 2001, just 3.8 % (617) of patients were aged 80 years or older compared to 10.0 % (3,136 patients) in 2011 (p < 0.05) [5].

In 2011, 0.8 % of operations were performed by surgeons with 1–4 years of experience after graduation, 8.8 % by surgeons with 5–9 years of experience, 42.1 % by surgeons with 10–19 years of experience and 48.2 % by surgeons with 20 or more years of experience. In all, 90.3 % (28,352) of patients were operated by surgeons with 10 or more years of experience, compared to 83.1 % (12,760) in 2001 (p < 0.05) [5].

DM was reported in 3,792 patients (12.1 %), dialysis in 437 (1.4 %), corticosteroid use in 701 (2.2 %), DMARD therapy involving biological therapy in 512 (1.6 %) and Parkinson’s disease in 198 (0.6 %).

Table 2 shows the categories and subcategories of diagnosis. The category of “degenerative disease” was the most frequent diagnosis, corresponding to 79.7 % of all cases. Amongst the subcategories, stenosis was the most common (40.5 %), followed by disc herniation (23.0 %). This compares with the 1994 survey in which disc herniation was the most common (38.5 %), and the 2001 survey in which stenosis was the most common (31.1 %) [4, 5]. Other frequent diagnoses were tumor (5.4 %), trauma (2.9 %), osteoporotic vertebral collapse (2.3 %), inflammation (2.9 %) and spinal deformity (6.8 %). The contribution of degenerative disease (79.7 %) was higher than that reported in the 1994 and 2001 surveys (78.7 and 78.2 %, respectively) [4, 5]. The frequencies of osteoporotic vertebral collapse and stenosis were higher than those reported in the 2001 survey (p < 0.05), reflecting the higher proportion of elderly patients undergoing spine surgery [5]. Spinal deformity (6.7 %) was more frequent than that reported in the 1994 and 2001 surveys (2.1 and 2.3 %, respectively; p < 0.05), reflecting the increased use of spinal instrumentation [4, 5]. The frequencies of disc herniation, ossification of ligaments, metastatic spine tumor and RA were lower than those reported in the 2001 survey (each p < 0.05) [5].

Table 3 shows details of the cases with degenerative disease. A total of 4,954 patients underwent spine surgery for cervical disease: 3,564 for cervical spondylotic myelopathy or radiculopathy, 641 for cervical disc herniation and 149 for ossification of the posterior longitudinal ligament (OPLL). A total of 21,338 patients underwent spine surgery for lumbar disease: 11,136 for lumbar spinal stenosis, 7,086 for lumbar disc herniation, 1,999 for degenerative spondylolisthesis and 270 for spondylolysis or isthmic spondylolisthesis.

Information on the surgical approach was available for 30,271 cases (96.5 % of total). Anterior surgery was indicated in 1,012 cases (3.2 %), posterior surgery in 28,909 cases (92.1 %) and combined anterior/posterior surgery in 350 cases (1.1 %). In the 1994 survey, anterior surgery was indicated in 13.2 %, posterior surgery in 84.7 % and combined anterior/posterior surgery in 2.2 %, while in the 2001 survey, anterior surgery was indicated in 9.1 %, posterior surgery in 88.6 % and combined anterior/posterior surgery in 1.8 % [4, 5]. Compared to the earlier surveys, the frequency of anterior surgery has decreased while the incidence of posterior surgery has increased (Fig. 2) [4, 5].

Table 4 shows the main surgical approaches for degenerative disease. In the 2011 survey, anterior surgery for patients with cervical spondylotic myelopathy or radiculopathy (CSM or CSR) and cervical disc herniation (CDH) had decreased to 7.1 % (248) and 47.9 % (297), respectively (each p < 0.05), compared to the 2001 survey [5]. In the 2001 survey, anterior surgery was indicated in 12.8 % (76) of patients with cervical OPLL, whereas in the 2011 survey this had decreased to 11.6 % (17; p = NS) [5]. In the 2011 survey, posterior surgery for patients with CSM or CSR and CDH increased to 91.8 % (3,208) and 49.7 % (308), respectively (each p < 0.05), compared to the 2001 survey [5]. In the 2001 survey, posterior surgery was indicated in 85.6 % (510) of patients with cervical OPLL, whereas in the 2011 survey this increased to 88.4 % (130; p = NS) [5].

Posterior surgery was indicated for almost all patients with lumbar degenerative disease: 99.8 % with lumbar disc herniation, 99.7 % with lumbar spinal stenosis, 99.6 % with degenerative spondylolisthesis, 99.8 % with spondylolysis and isthmic spondylolisthesis, and 100 % with ossification of ligaments. Posterior surgery for all degenerative diseases was more frequent than in both earlier surveys [4, 5].

Information on the operation time was available for 31,235 (99.5 %) cases. The operation time was less than 1 h in 3,899 cases (12.5 %), 1–2 h in 10,806 cases (34.6 %), 2–3 h in 8,060 cases (25.8 %), 3–4 h in 4,115 cases (13.2 %), 4–5 h in 2,070 cases (6.6 %), 5–6 h in 1,030 cases (3.3 %), 6–8 h in 851 cases (2.7 %), 8–10 h in 300 cases (1.0 %) and more than 10 h in 104 cases (0.3 %). Thus, in 72.9 % of patients, the operation time was 3 h or less.

Information on intraoperative blood loss was available for 31,188 (99.4 %) cases. This was 200 ml or less in 21,752 cases (69.7 %), 200–500 ml in 5,445 cases (17.5 %), 500–1,000 ml in 2,079 cases (6.7 %), 1,000–2,000 ml in 839 cases (2.7 %), 2,000 ml or more in 391 cases (1.2 %) and an unknown volume in 682 cases (2.2 %).

Information on surgical procedures was available for 4,665 (94.2 %) of the patients with cervical degenerative disease. Decompression was commonly indicated for cervical spondylotic myelopathy or radiculopathy (81.0 %) and for cervical OPLL (80.3 %). Decompression and fusion were indicated for cervical disc herniation (50.2 %). Information was available for 21,011 (98.5 %) cases with lumbar degenerative disease. Decompression was mostly indicated for lumbar spinal stenosis (68.7 %) and for lumbar disc herniation (93.2 %), while decompression and fusion were indicated for lumbar degenerative spondylolisthesis (68.4 %).

Table 5 shows the frequencies of use for the different surgical technologies. The number cases treated by endoscopic surgery in the 2011 survey was more than tenfold higher than in the 2001 survey, while the number of cases treated by microscopic surgery increased about fourfold [5]. In the 2001 survey, endoscopic and microscopic surgery were indicated in 2.6 and 11.1 % of cases, respectively, whereas in the 2011 survey these had increased to 12.9 and 20.5 %, respectively (p < 0.05) [5].

Spinal instrumentation was applied in 9,487 (30.2 %) of the 31,380 cases in the 2011 survey, compared to 5,210 (27.0 %) and 5,497 (34.0 %) cases in the 1994 and 2001 surveys, respectively [4, 5]. Thus, the actual number of cases with spinal instrumentation was about 4,000 more than in both previous surveys. A pedicle screw system was used in the majority (82.3 %) of patients (Table 6). This was higher than the frequency reported in the 1994 (63 %) and 2001 (54.6 %) surveys [4, 5]. Surgeons with less than 5 years experience performed spinal instrumentation in 16.0 % (40/251 cases), while surgeons with 5–9, 10–19 or more than 20 years experience performed spinal instrumentation at a similar frequency of 30.2 % (831/2,754 cases), 30.3 % (4,005/13,212 cases) and 30.4 % (4,606/15,140 cases), respectively. Table 7 shows the frequencies of instrumentation surgery used for different diseases. This procedure was commonly indicated for spinal deformity (74.1–81.1 %), lumbar spondylolysis and isthmic spondylolysis (77.4 %), and RA (75.4 %). Overall, these frequencies were similar to those reported in the two earlier surveys for the same conditions [4, 5].

Table 1 shows the intraoperative and postoperative items of complication. These were reported in 3,269 of the 31,380 cases (10.4 %) in the 2011 survey and were significantly higher than in the 1994 (8.6 %, 1,569/19,271) and 2001 (8.6 %, 1,383/16,157) surveys (both p < 0.05) [4, 5]. The mean patient age for patients with intraoperative and postoperative complications in the 2011 survey was 64.4 ± 17.0 years, while it was 58.9 ± 20.6 years for those without complications (p < 0.01).

Table 8 shows the incidence of intraoperative and postoperative complications according to diagnosis. Diseases accompanied by a high incidence of complications included intramedullary tumor, primary malignant tumor, osteoporotic vertebral collapse, inflammatory disease and spinal deformity.

Table 9 shows the frequencies of neurological and non-neurological complications in representative diseases. Neurological complications comprised spinal cord damage, nerve root damage and cauda equine damage. The incidence of neurological complications was very low for cases with disc herniation (0.6 %), stenosis (1.1 %), spondylolisthesis (1.1 %) and osteoporotic vertebral collapse (1.0 %), but was higher for cases with ossification of ligaments (3.7 %) and spinal deformity (2.1–2.8 %).

Complications occurred in 12.0 % of cases treated at the cervical level, in 14.8 % of those treated at the thoracic level and in 10.1 % of those treated at the lumbar level. The incidence of complications for all levels of surgically treated spine was higher than that observed in the 2001 survey [5].

Complications occurred in 10.2 % of patients treated by a posterior approach, 14.5 % of those treated by an anterior approach and 27.7 % of those treated by a combined anterior/posterior approach (p < 0.05). These findings were similar to those reported in the 2001 survey (Table 10) [5].

Table 11 shows the relationship between surgical approach, incidence of complications and level of the spine. The highest incidence of complication was observed for a combined anterior/posterior approach in the thoracic spine (31.0 %). Complications for the cervical, thoracic and lumbar spine levels were related to the surgical approach (each p < 0.05).

Table 12 shows the incidence of complications in all spine surgery in the 1994, 2001 and 2011 surveys. The highest incidence of complication was dural tear (2.1 %), followed by neurological complication (1.4 %), DWI (1.1 %) and EH (0.9 %). The incidence of neurological complications (1.4 %) was higher than in the 1994 survey (0.9 %, p < 0.05), but lower than in the 2001 survey (1.7 %, p < 0.05) [4, 5]. The incidence of DWI has increased progressively from 0.6 % in the 1994 survey to 0.9 % in the 2001 survey and 1.1 % in the 2011 survey (p < 0.05 for 1994 vs. 2011) [4, 5]. The incidence of mental disorders in the 2011 survey increased compared to both earlier surveys (each p < 0.05) [4, 5]. The incidence of death increased from 0.08 % in the 2001 survey to 0.2 % in 2011 (p < 0.02). The incidence of PE/TE increased from 0.1 % in the 2001 survey to 0.2 % in the 2011 survey (p < 0.05) [5]. The incidence of vascular injury decreased from 0.05 % in the 2001 survey to 0.02 % in 2011 (p = NS) [5].

Table 13 shows details of the neurological complications. Spinal cord damage occurred in 85 cases (0.3 %), nerve root damage in 297 cases (0.9 %) and cauda equina damage in 50 cases (0.2 %). Seven cases showed damage at two sites. The incidence of damage to the nerve root, spinal cord and cauda equina were all higher in the 2011 survey compared to the 1994 survey, but only the former reached significance (p < 0.05) [4].

Table 14 shows the incidence of complications according to the experience of the surgeon. Neurological complications occurred most frequently (2.4 %) in patients treated by surgeons with less than 5 years of experience. In the 2001 survey, the incidence of dural tear was clearly lower for patients operated by surgeons with longer experience in spine surgery [5]. However, in the present survey, dural tear occurred most often in patients treated by surgeons with 10–19 years of experience (incidence of 2.4 %), and was actually lower for surgeons with less experience (2.0 %). DWI occurred most frequently in patients treated by surgeons with 20 or more years of experience (incidence of 1.2 %) and was lowest in patients treated by surgeons with 5 or less years of experience (0.4 %). The incidence of representative complications (neurological complications, dural tear and DWI) was not significantly different between surgeons with different lengths of professional experience.

The incidence of complications for cases treated with instrumentation surgery (15.6 %, 1,480/9,487) was almost double that of cases treated by non-instrumentation surgery (8.2 %, 1,789/21,893; p < 0.05). This result was similar to the two earlier surveys (15.5 vs. 5.9 % in 1994 and 12.1 vs. 6.8 % in 2001; Table 15) [4, 5]. The incidence of complications with both instrumentation surgery and non-instrumentation surgery was higher in the 2011 survey compared to the two earlier surveys. For patients undergoing instrumentation surgery, this increase occurred regardless of the surgeon’s experience (Table 16). The incidence of complications with instrumentation surgery and with all surgery was highest in patients treated by surgeons with 10–19 years of experience (17.0 and 11.8 %, respectively).

Spinal cord damage was reported in 85 patients. This was due to a tumor in 21 cases, OPLL in 20 cases and cervical spondylotic myelopathy in 13 cases. Similar to the 1994 survey, an intramedullary tumor was the cause of many of the cases (9/21) with spinal cord damage due to a tumor [4]. Cervical surgery was indicated in 44 cases with spinal cord damage and thoracic surgery in 23 cases.

DWI occurred in 343 (1.1 %) of the 31,380 surveyed cases and was more frequent in those with complicated and invasive surgical procedures. DWI was threefold higher in cases treated with instrumentation surgery (2.0 %, 189/9,487 cases) compared to those treated with non-instrumentation surgery (0.7 %, 154/21,893 cases; p < 0.05). In the 1999 survey, DWI was fivefold higher in cases treated with instrumentation surgery (1.0 %, 53/5,210 cases) compared to those treated with non-instrumentation surgery (0.2 %, 27/14,061 cases [4]. The distribution of DWI was: DM (25.1 %), dialysis (2.6 %), corticosteroid use (5.5 %), DMARD therapy involving biological therapy (5.3 %) and Parkinson’s disease (1.8 %). The incidence of DWI was 2.2 % (86/3,792) in cases with DM, 2.1 % (9/437) in cases with dialysis, 2.7 % (19/701) in cases with corticosteroid use, 3.5 % (18/512) in cases with DMARD therapy involving biological therapy and 3.0 % (6/198) in cases with Parkinson’s disease. DWI was significantly associated with each of these conditions (each p < 0.05).

After exclusion of 145 cases with missing information on the operation time, a total of 337 cases with DWI, 288 cases with EH, 59 cases with PE/TE and 54 cases who died were evaluated (Table 17). DWI, EH and PE/TE showed significant correlations with the length of operation time (each p < 0.05; Spearman’s correlation coefficient).

After exclusion of 192 cases that lacked information on intraoperative blood loss, a total of 343 cases with DWI, 275 cases with EH, 59 cases with PE/TE and 54 cases who died were subjected to analysis (Table 18). DWI, EH, PE/TE and death all showed a significant correlation with volume of intraoperative blood loss (each p < 0.05; Spearman’s correlation coefficient).

The 54 patients who died were evaluated for the diagnosis and the cause of death (Tables 19, 20). There were 30 males and 24 females and the mean age was 69.9 years (range 38–91 years). The highest incidence of death was associated with dialysis (4.0 %, 4/101) followed by metastatic spine tumor (3.9 %, 15/389). With regard to the cause of death, tumor-associated death was the most common, followed by inflammation of the lungs.

Table 21 shows the relationship between the incidence of representative complications and the surgical technology used. The incidence of dural tear was low (1.5 %) for microscopic surgery, while the incidence of EH was high (1.1 %) for conventional surgery. The incidence of DWI (0.05 %) and the overall incidence of complications were very low for endoscopic surgery. Surgeons with longer experience in spine surgery used endoscopic surgery more often.

Table 22 shows the association between the incidence of representative complications and the surgical technology used for 3,415 patients with lumbar disc herniation at the L4/5 level. Of these, 1,352 were treated by conventional surgery, 1,002 by endoscopic surgery, 1,044 by microscopic surgery and 17 by percutaneous surgery. Excluding the latter 17 patients, the complications of nerve root damage, EH and DWI were significantly correlated with surgical technology (each p < 0.05), whereas the complications of cauda equina damage and dural tear showed no significant correlation with surgical technology.

A total of 24,427 patients with information on BMI were available for data analysis, of which 2,551 experienced intraoperative and postoperative complications. The mean BMI of patients with intraoperative and postoperative complications was 23.7 ± 4.1, compared to 23.7 ± 3.8 for those without complications (p = NS). The mean BMI of patients with dural tear and PE/TE was higher than that of patients without intraoperative or postoperative complications, and it was lower with cauda equina damage, respiratory disease and death (each p < 0.05).

The mean age of patients with nerve root damage, dural tear, implant failure, cauda equina damage, EH, DWI, mental disorder, PE/TE, circulatory disease, cerebral disease, respiratory disease and death was higher when compared to patients without intraoperative or postoperative complications (each p < 0.05), and was lower with implant failure (p < 0.05).

Logistic regression analyses were performed for exploring factors that are associated with intraoperative and postoperative complications. We focused on the complications that occurred in more than 200 cases (dural tear, WDI, neurological complications and EH). Table 23 showed factors found to be associated with these complications. There were no factors associated with EH.

Compared to the 1994 and 2001 surveys, the frequency of anterior surgery was lower in 2011, whereas that of posterior surgery increased [4, 5]. Posterior rather than anterior decompression surgery was increasingly used for CSM or CSR and for CDH. In the 2001 survey, complications were more frequent in cases treated by an anterior approach compared to cases treated by a posterior approach, possibly explaining the subsequent decrease in use of this practice [5]. In general, the posterior approach is safer, easier to perform and less invasive than the anterior approach [10]. It is likely that a posterior approach is preferred by many surgeons who treat increased numbers of elderly patients.

In the current survey, the incidence of complications was related to the surgical approach used for the cervical, thoracic and lumbar spine. Surgeons should be especially aware of complications that can arise when using an anterior approach for thoracic spine. The Scoliosis Research Society reported similar complication rates for anterior versus posterior approaches in the treatment of adolescent idiopathic scoliosis [7]. Combined anterior and posterior instrumentation and fusion leads to a significantly higher incidence of neurologic complications than anterior or posterior instrumentation and fusion alone. The incidence of infection with the posterior approach is higher than with the anterior approach [11]. Pull ter Gunne [12] reported that isolated anterior surgical approaches were associated with a 1.7 % risk of surgical-site infection (SSI), whereas any surgery that included a posterior spinal approach was associated with a minimum risk of infection of 4.4 %. Care should therefore be taken to prevent infection when using a posterior approach.

Endoscopic and microscopic surgery were indicated in 2.6 and 11.1 % of cases in the 2001 survey, respectively, but increased to 12.9 and 20.5 % in the 2011 survey (each p < 0.05) [5]. The overall incidence of complications was very low for endoscopic surgery (Table 21). However, because endoscopic and conventional surgeries are used for different pathologies, it is difficult to correlate the incidence of complications with the surgical technology used. In addition, surgeons with a longer experience in spine surgery tended to use endoscopic surgery more often. We therefore correlated the incidence of representative complications with the surgical technology in patients with lumbar disc herniation at L4/5. The incidence of complications from nerve root damage, EH and DWI was very low for both endoscopic and microscopic surgery. Gotfryd and Avanzi. [13] reported that conventional, microscopic and endoscopic posterior discectomy surgical techniques were all effective for the treatment of single level lumbar disc herniations in patients without degenerative vertebral deformities. Endoscopic and microscopic surgery were superior to conventional surgery with respect to the volume of blood loss, systemic repercussions and duration of hospital stay [14]. The incidence of neurological complications was significantly lower in the 2011 survey compared to the 2001 survey, whereas the incidence of mental disorders, death and PE/TE increased significantly [5].

The incidence of representative complications (neurological complications, dural tear and DWI) did not vary significantly according to the experience of the surgeon.

In the current survey, the incidence of complications in patients treated with instrumentation surgery was higher than in those treated with non-instrumentation surgery (p < 0.05) and was significantly higher than in the 2001 survey (p < 0.05) [5]. This can be explained by the fact that patients indicated for instrument surgery were older. Carreon et al. [15] have previously reported that the complication rate increases with older age, increased blood loss, longer operation time and the number of levels of the arthrodesis. The most common major complication in older patients was wound infection, with a prevalence of 10 % for posterior lumbar decompression and for arthrodesis. These authors suggested that attention should be paid to the control of blood loss, and to limiting the operative time. On the other hand, Cassinelli et al. [16] argued that age should not be used as a criterion to avoid decompression and fusion with or without instrumentation for the treatment of lumbar stenosis associated with instability. Elderly patients can undergo these procedures safely with a low risk of developing major perioperative complications.

It is important to be aware of the risk factors for infection, so that preventive measures can be taken and the surgical treatment optimized [17]. Pull ter Gunne and Cohen [12] reported that blood loss of >1 l, previous SSI and DM were independent risk factors for SSI, while obesity was an independent risk factor for superficial SSI, and DM, obesity, previous SSI and longer surgeries (>5 h) were independent risk factors for deep SSI. In the current survey, we reviewed the preoperative factors of DM, dialysis, corticosteroid use, DMARD therapy involving biologic therapy, Parkinson’s disease, age and BMI. The intraoperative factors reviewed were operation time and intraoperative blood loss. The incidence of DWI was significantly associated with the preoperative factors of DM, dialysis, corticosteroid use, DMARD therapy involving biologic therapy, Parkinson’s disease and age (each p < 0.05), as well as with the intraoperative factors of operation time and blood loss (each p < 0.05). Factors associated with DWI were instrumentation surgery, DM and DMARD therapy involving biological therapy, as results of multiple logistic regression analyses.

It is important to be aware of the risk factors for EH so that preventive measures can be taken and surgical treatment optimized. Postoperative spinal EH are very rare, with a reported incidence ranging from 0.1 to 1 % [18‐21]. Patients who undergo multilevel operative procedures, are > 60 years old and/or have preoperative coagulopathy have a significantly higher risk [18‐20]. In the current survey, we reviewed the preoperative factors of patient age and BMI, while the intraoperative factors reviewed were operation time and intraoperative blood loss. EH was significantly associated with age as a preoperative factor (p < 0.05) and with operation time and blood loss as intraoperative factors (each p < 0.05). Cabana et al. [22] and Amiri et al. [23] previously reported on the association between surgical time and neurological outcome. Evacuation of EH within 4 h resulted in better neurological recovery than surgery after 4 h. Seichi et al. [24] reported on neurological complications of cervical laminoplasty for patients with OPLL and found that the incidence of EH was 0.5 % (3/581 patients). The three cases with EH were diagnosed on the day of surgery, demonstrating that careful attention should be paid to the condition of patients at that time.

It is also important to know the risk factors for PE/TE so that preventive measures can be taken and surgical treatment optimized. The incidence of PE/TE in the current survey was 0.2 % (59/31,380) compared to 0.1 % (18/16,157) in the 2001 survey [5]. The preoperative factors reviewed in this survey were age and BMI, while the intraoperative factors reviewed were operation time and intraoperative blood loss. PE/TE was significantly associated with age and BMI (each p < 0.05), and with operation time and intraoperative blood loss (each p < 0.05). However, due to insufficient data, a standardized prophylactic regimen cannot be recommended. Elastic compression alone or combined with pharmacological prophylaxis appears to be effective. Schoenfeld et al. [25] reported that a BMI of 40 or more, age of 80 years or more, operative time exceeding 261 min and American Society of Anesthesiologist classification 3 or greater were each significant independent predictors of deep vein thrombosis. They also reported that a BMI of 40 or more, operative time > 261 min and male gender were associated with the development of PE. Prophylaxis with both measures is strongly recommended for high-risk patients [26]. Dearborn et al. [27] investigated the association between surgical procedures and PE in 318 major spinal reconstructive procedures. They reported seven cases of PE, of which six occurred amongst 97 patients undergoing combined anterior and posterior spinal procedures (6.1 %) and one in a patient undergoing a posterior procedure. The overall clinical PE rate with the combined approach was significantly higher than for patients who underwent the posterior approach only. Preventive measures include pharmacological, mechanical and combinations of these interventions. Mechanical interventions include elastic compression stockings and intermittent pneumatic compression devices. Pharmacological options include low-dose unfractionated heparin or a low molecular weight heparin (LMWH) such as enoxaparin, dalteparin, tinzaparin, certoparin or nadroparin [28]. Because of the possibility of EH, anticoagulation has not gained wide acceptance by spine surgeons [29]. A 2012 survey of British orthopedic spine surgeons revealed that only 31 % routinely used LMWH [30]. Strom et al. [28] reported on the safety and efficacy of prophylactic LMWH (e.g., 40 mg enoxaparin for normal renal function, 30 mg for impaired renal function) starting 24–36 h after multilevel laminectomy or laminectomy and fusion in 367 patients with degenerative disease. None of the patients developed an EH, superficial hematoma or persistent wound drainage. In the current survey, fatal PE was reported in three cases. Surgeons with a better understanding of venous thromboembolism following spine surgery can now weigh the risks and benefits of postoperative anticoagulation measures.

The response was achieved from 209 institutions in 750 institutions (response rate 28 %). Because the response rate was low, the results of this analysis might not reflect correctly the present situation for spine surgery in Japan.