- Email: [email protected]
Spinal Instrumentation With A Low Complication Rate
Spine
Spinal Instrumentation With A Low Complication Rate Scott A. Shapiro, M.D. and William Snyder, M.D. Section of Neurosurgery, Indiana University Medical Center, Indianapolis, Indiana
Shapiro SA, Snyder W. Spinal instrumentation with a low compliance rate. Surg Neurol 1997;48:566 –74. BACKGROUND
Spinal instrumentation has become an increasing part of the armamentarium of neurosurgery and neurosurgical training. For noncontroversial indications for spine fusion the arthrodesis rate seems to be better. For both noncontroversial and controversial indications, the reported complication rate with spinal instrumentation tends to be greater than that with noninstrumented spine surgeries. These reported complications include a 2–3% neurologic injury rate, 3– 45% reoperation rate for implant failure, and infection rates of 5–10%. Therefore, we report on 299 cases that have undergone spinal instrumentation placed exclusively by neurosurgeons with a very low complication rate. METHODS
Two hundred ninety-nine consecutive spinal instrumentation cases performed exclusively by neurosurgeons at Indiana University Medical Center were analyzed for complications related to spinal instrumentation. The spinal instrumentation placed consisted of 195 anterior cervical locking plates, 22 cases of posterior cervical instrumentation, 9 cases of combined anterior locking plates with posterior cervical instrumentation, 14 anterior thoracolumbar plates, 51 posterior thoraco-lumbar instrumentation cases, and 8 combined anterior/posterior thoracolumbar instrumentation cases. RESULTS
The mean follow-up is 40 months (6 –95). There was one perioperative death unrelated to the spinal instrumentation. There were no neurologic injuries and there has been no hardware infection to date. There were two dural tears, three superficial wound infections, and three minor wound breakdowns successfully treated. Hardware complications included three cervical plate/screw extrusions reoperated, one cervical plate fracture reoperated, one posterior cervical screw backout not reoperated, one case of broken pedicle screws not reoperated, one vertebral body failure not reoperated, and one posterior rod case reoperated for excessive rod length and protrusion. The overall complication rate attributable to placement of spinal instrumentation was 10/299 (3%) with a reoperation rate of 2%. The arthrodesis rate was 298/299 (99%). Address reprint requests to: Scott Shapiro, M.D., Room 323, East Outpatient Building, Wishard Memorial Hospital, 1001 W. 10th Street, Indiana University Medical Center, Indianapolis, IN, 46202. Received January 31, 1997; accepted April 24, 1997. 0090-3019/97/$17.00 PII S0090-3019(97)00296-6
CONCLUSION
The complication rate for using spinal instrumentation can be less than previously reported. Lessons learned and discussed should reduce the rate even more. Spinal instrumentation is a safe and useful adjunct to fusion in treating degenerative, traumatic, infectious, and neoplastic diseases of the spine. © 1997 by Elsevier Science Inc. KEY WORDS
Complications, hardware infection, spine fusion, spinal instrumentation.
pine surgery in general has a very low mortality rate in the range of 0.1– 0.3% which is usually attributable to the underlying illness and not the surgery [6,7]. Morbidity for elective herniated discs and non-instrumented spinal decompressions is also quite low in the modern era, but remains quantifiable. Lumbar decompressive, noninstrumented surgery is complicated by large vessel injury in 0.1%, cauda equina syndrome in 0.2%, nerve root damage in 0.2– 0.5%, infection in 1%, and dural tears in 1% [6 – 8,10,13,19,31]. Cervical spine surgery is complicated by 2% anterior bone graft failures, 0.4% CSF leaks, 0.2% root injury, 0.3% quadriplegia, and 0.7% mortality (Osenbach et al, unpublished data, 42). Internal fixation hardware to augment stability and improve spinal fusion for all levels of the spine has become increasingly popular, despite intense controversy surrounding specific constructs. Relatively noncontroversial indications include severe traumatic instability, reconstruction after corpectomy for trauma, neoplasm or spondylosis, scoliosis, and Grade III or more L5S1 spondylolisthesis [18,38,40,45–48]. Controversial indications include internally fixated fusion for degenerative disc disease without instability, degenerative listhesis above L5-S1, and long-level cervical laminectomy [5,23,37,38,45,46,52]. Numerous complications related to the placement of internal fixation hardware have been reported. These complications can occur intraoperatively or postoperatively. Intraoperative complications are primarily
S
© 1997 by Elsevier Science Inc. 655 Avenue of the Americas, New York, NY 10010
Few Complications with Spine Instrumentation
neurologic injuries, dural injuries, vascular injuries, and malplacement of the instrumentation. Postoperative complications include infection and instrumentation breakage/dislodgment (with or without pseudoarthrosis and/or recurrent instability or deformity). Previous reports have documented a 2–3% neurologic injury rate, 3– 45% reoperation rate for implant failure, and infection rates of 5–10% [4,5,12, 13,15,16,20,22,29,33,50]. There is some variability depending on the type and location of instrumentation, but the general finding is one of more complications as compared with discectomy and noninstrumented decompression. This is an important concept as neurosurgery has become increasingly involved with internal fixation and is passing these techniques on to the next generation of neurosurgeons. Many of the intraoperative and some of the postoperative complications are avoidable and must be avoided to justify the use of the various systems. Thus, we report a large experience with spinal instrumentation with a low complication rate.
Patient Population Two hundred ninety-nine consecutive cases of patients with spinal internal fixation placed exclusively by neurosurgery are analyzed. The internal fixation analysis will be subdivided into following groups: anterior cervical internal fixation, posterior cervical internal fixation, combined anterior and posterior cervical internal fixation, anterior thoraco-lumbar internal fixation, posterior thoracolumbar internal fixation, and combined anterior and posterior thoraco-lumbar internal fixation. Anterior grafts under compression were cadaveric fibula or tibia in 99%. Posterior grafts for lateral mass fusions were always autologous bone. Allogeneic bone matrix has been used to augment 199 (66%) fusions. The overall mean follow-up for all patients is 40 months (6 –95).
Perioperative Management Universal to All Levels and Types of Internal Fixation Any patient requiring decompression or reduction of a spinal segment involving the spinal cord receives perioperative steroids and intravenous mannitol. All patients receive prophylactic intravenous antibiotics intra-operatively and for at least 48 h after surgery. All sterilized spinal instrumentation is
Surg Neurol 567 1997;48:566 –74
left covered until the moment it is to be placed. Traffic in and out of the room is reduced as much as possible. After placement of the instrumentation, aggressive hemostasis using electrocautery and gelfoam thrombin powder is always used, followed by extensive irrigation with 1–2 L of antibiotic solution. No drain is left in any posterior instrumentation case with the rare exception of a patient with coagulopathy. The only drain that is left in an anterior case is a chest tube for thoraco-lumbar anterior approaches. The rare posterior bulb suction drain that is left is removed within 24 h and an intraoperatively placed suture to close the drain site is tied down immediately to close the drain site. Intravenous antibiotics are continued as long as the chest tube is in place. The wound must be checked frequently by someone knowledgeable until healed completely. At the first sign of superficial infection, attention to local wound care and intravenous antibiotics are instituted.
Anterior Cervical Instrumentation (N 5 195) This group included 195 anterior locking cervical plates (Synthes) including 146 anterior cervical discectomy (ACD) fusions; 31 single-level vertebrectomies; and 18 multilevel vertebrectomies for spondylosis, trauma, and tumor. All trauma and vertebrectomy cases were managed in a rigid orthosis or halo and the nontraumatic ACD cases were managed in a soft collar. There have been no infections and no neurologic complications in any of the 195 cases. Of the 146 ACD-banked fibula locking plate fusions, there were two complications. One occurred in a patient treated for a facet fracture dislocation with disc disruption by a singlelevel ACD banked fibula, locking plate fusion at C5-C6. The patient’s radiculopathy improved completely. Three weeks after surgery, while intoxicated from alcohol abuse, the patient fell while in a rigid collar, suffering a fracture of his banked fibula graft with 2 mm extrusion of the upper part of the plate and 3– 4 mm subluxation. He presented with radicular recurrence. He was revised to a combined anterior-banked fibula graft and a posterior locking plate fusion with autologous bone and was placed in a halo. Despite numerous patient-induced problems with the halo, the patient went on to arthrodesis and remained neurologically normal. The second ACD-banked fibula locking plate complication also occurred because of alcohol abuse when the plate fractured from a high-speed motor vehicle accident 6 months after surgery, and was subse-
568 Surg Neurol 1997;48:566 –74
Shapiro and Snyder
dependent, asthmatic patient with a C5 burst fracture with severe spinal cord injury exsanguinated at 4 weeks from a catastrophic gastric stress ulcer. Before her demise, her neurologic condition had improved significantly. All patients went on to arthrodesis. Thus, the overall reoperation rate for anterior cervical plate placement was 5/195 (2.6%).
Posterior Cervical Instrumentation (N 5 22)
Lateral cervical radiograph of a two-level vertebrectomy demonstrating fracturing of the lower two plasma screws along with graft and plate extrusion.
1
quently removed with a good outcome. It was noted that the graft had fused. There were no single-level vertebrectomy complications. There were 3/18 (11%) multilevel vertebrectomy plate/screw complications. The first complication was a two-level vertebrectomy done before the availability of plates longer than 60 mm. The author used a shorter plate with abnormal screw positioning that fractured out of the lower body, with recurrent neck and arm pain 2 days postoperatively. The patient was revised without instrumentation and did well. A two-level vertebrectomy with a locking plate and plasma screws presented at 8 weeks with fracturing of the lower screws, partial extrusion of the lower plate and graft, and dysphagia (Figure 1). She was revised and has done well. A third complication was in a twolevel vertebrectomy in which a locking screw was inadvertently not placed and the screw backed out 2 days postoperatively. This required reoperation and was easily corrected. One obese, steroid-
There were 12 lateral mass plates (Axis, Danek, Inc) placed for trauma (unilateral/bilateral facet dislocation), rheumatoid arthritis, or neoplasia. All had placement of autologous bone in the facet and lateral masses. There were seven cases where C1-C2 transarticular screws (Aesculap) were placed along with an interspinous type fusion using braided cable (Danek) and autologous iliac crest for trauma or rheumatoid arthritis. There were three instrumented craniocervical fusions with autologous iliac crest, two of which used lateral mass plates with C2 pedicle screws in addition to lateral mass screws at C3 and C4 and braided cables for the occiput. There were no infections, nerve, or vertebral artery injuries. There has been one posterior cervical plate complication where a single screw partially backed out which was not reoperated, because the patient was asymptomatic. Twenty patients were kept in rigid cervical collars and two were kept in halos, and all maintained reduction and went on to arthrodesis.
Combined Anterior/ Posterior Cervical Instrumentation (N 5 9) There have been nine banked fibula, locking plate, anterior cervical fusions with posterior lateral mass plates and autologous bone lateral mass fusions performed for fracture dislocations with disc disruption. There have been no complications in this group and all have gone on to arthrodesis. All but one were managed with rigid orthoses.
Anterior Thoracolumbar Instrumentation (Z-plate or TSRH) (N 5 14) Fourteen patients were instrumented for trauma or tumor including 13 Z-plates (Danek) and 1 Synthes
Few Complications with Spine Instrumentation
Surg Neurol 569 1997;48:566 –74
plate. There were no neurologic complications, vascular complications, or dural injuries. There have been no infections and only one minor wound breakdown successfully treated with dressing changes. There have been no screw or plate complications and all have fused. All were managed with thoraco-lumbar-sacral orthoses.
Posterior Thoracolumbar Instrumentation (CD or TSRH) (N 5 51 cases) The primary system used was TSRH (Danek) in 48 cases and Cotrel-Dubousset (Sofamor) in three cases. Indications included trauma (n 5 21), degenerative listhesis (n 5 8), spondylolisthesis (n 5 6), tumor (n 5 13), and degenerative disc disease (n 5 3). A total of 76 pedicle screws have been placed. Other constructs have made use of the claw technique using primarily pedicle hooks and transverse process hooks. We rarely use laminar hooks. There were no neurologic injuries or pedicle fractures. There were two small dural tears caused by pedicle screw slippage during insertion that were repaired primarily without leakage. There have been three superficial wound infections treated early with local wound care and intravenous antibiotics followed by oral antibiotics. There were no deep infections. There were two minor wound breakdowns successfully treated with dressing changes. An 82-year-old woman who suffered a traumatic severe T-12 burst fracture with resultant severe pain, paraparesis, and sphincter impairment underwent a T-12 vertebrectomy, banked tibia vertebral body replacement, and posterior pedicle screw/rod construct and lateral mass autologous rib fusion placed from T-11 to L-1. The patient was neurologically normal and ambulating in a brace by 2 weeks. The banked tibia graft angulated asymptomatically somewhere between 4 and 12 weeks after surgery and was noted on routine radiographic follow-up (Figure 2). Nothing was done other than periodic observation, and the patient returned to work after 6 months. A 57-year-old white female with metastatic breast cancer had severe destruction of three contiguous thoracic bodies with severe spinal cord compression and paraparesis leaving her bedridden. She had a successful three-level thoracic corpectomy with banked tibia anterior fusion and posterior TSRH hook/rod distraction instrumentation with autologous rib lateral mass fusion that led to the
Lumbar radiograph after L1 vertebrectomy demonstrating asymptomatic angulation of a tibial graft that had been initially been vertically oriented.
2
return of ambulation. Seven weeks after the fusion, the upper bordering body that also had cancer failed, with subsequent telescoping of the tibial graft through the body; the posterior construct remained unchanged. This patient became paraplegic acutely from this host/graft failure and died from liver failure caused by multiple metastases within 3 weeks. Screw threads of 1–2 mm in dimension outside of the lateral or medial aspect of the pedicle identified on follow-up CT scan occurred in 8%, though this had no effect on the operative result. There was 1 case out of 17 where the upper pedicle screws broke. This occurred 14 months after surgery in a tobacco- and alcohol-abusing patient with a L1 burst, flexion distraction injury who remained neurologically intact. The patient was instrumented in distraction with pedicle screws, with improvement in the kyphosis but no improvement in the compression of the L1 body, along with a lateral mass autologous iliac crest fusion. The patient developed low back pain 14 months postoperatively. Lumbar
570 Surg Neurol 1997;48:566 –74
Shapiro and Snyder
Combined Anterior/Posterior Thoracolumbar Instrumentation and Fusion (N 5 8) The indications were trauma in five, tumor in two, and infection in one. All cases had evidence for three column involvement. Seven consisted of anterior Z-plates with posterior TSRH instrumentation and one consisted of anterior and posterior TSRH. There were no complications in this group of eight patients and all have gone on to arthrodesis.
Discussion
Thoracic radiograph demonstrating inadequate trimming of the lower rods bilaterally that led to painful protrusion in this lymphoma patient, requiring a second procedure to trim the rods.
3
radiographs and CT scan demonstrated that the upper screws had broken and the patient had reangulated to the same degree as before surgery, but remained neurologically intact and has very little chronic pain. There have been no other cases of instrumentation breakage, dislodgment, or progressive deformity to date. Two additional cancer patients died from distant disease before 1 year with no spinal complications. Radiographs in patients alive at 1 year showed arthrodesis. Five additional cancer related deaths have occurred without failure of the spinal construct at 13–29 months. One lymphoma case was reoperated 18 months after posterior TSRH rod/hook placement to trim the inferior rods that were left too long and caused painful protrusion that interfered with upright posture and walking (Figure 3). The patient dramatically improved after rod trimming. Thus the reoperation rate is 1/51 (2%) with no cases of hardware removal to date.
In most large series of spinal instrumentation cases there continues to be a reported rate of neurologic injury that ranges from 0.4% to 22% [4,5,12,15,16, 23,29,37,38,45,46,50,51]. Neurologic injury, above all other complications, should not occur. Causes may be multifactorial and occasionally without explanation, but most can be explained. With reference to pedicle screws, the majority of neurologic injuries reported seem to be attributable to faulty placement of too large a diameter screw. In a series of 617 cases of pedicle screw placement, faulty placement occurred in at least 5.2% of the cases [15]. CT scan evaluation of pedicle screw placement has documented as high as 30% malplacement that ranges from catastrophic canal penetration to slight medial penetration of the pedicle screw [17]. There is no data to suggest that slight medial penetration by 1–2 mm adversely affects the outcome. Pedicle fracturing by the screw occurred in 2.2% of the cases [15]. Selection of screw diameter size should be the largest screw that will safely fit in the pedicle. Axial CT scans provide accurate measurements to help guide the choice of screw diameter and angulation. Table 1 lists the recommended screw diameter per level. Just because it is recommended that a 6.5–7.5 mm screw should be used at L5 does not mean that a 5.5 mm screw will not work. Many surgeons do not use intraoperative fluoroscopy and rely on anatomic landmarks. Presently we use fluoroscopy (A-P and lateral) in addition to CT scans and intraoperative anatomy to guide every pedicle screw placement, and find the radiographic guidance indispensable. It is the authors’ opinion that the placement of pedicle screws without radiographic guidance is substandard surgical practice and will eventually lead to an unacceptable and avoidable neurologic injury. Triggered EMG moni-
Few Complications with Spine Instrumentation
1
Recommended Screw Diameter
LEVEL T11 T12 L1 L2 L3 L4 L5 S1 S2
SCREW DIAMETER (MM) 5.0–5.5 5.0–5.5 5.5–6.0 5.5–6.0 6.5 6.5–7.5 6.5–7.5 6.5–7.5 6.5
toring can also be used to indicate root irritation when placing the screw [28]. We have used this on two smaller patients but have yet to have a root irritation sign. There is no question that fluoroscopy slows down the case significantly as one swings the machine from A-P to lateral in guiding the precise trajectory of the drilling and screw placement. Speed is not the goal of the operation. Neurologic decompression, restoration of stability and/or prevention of further slippage is the goal. Follow-up CT scan is often used to assess our pedicle screw placement. It seems that frameless stereotaxis may improve our ability to precisely cannulate the pedicle [34]. Another cause of neurologic injury appears to be excessive manipulation to attain anatomic alignment [14,37]. Perfect anatomic alignment is often unattainable or unsustainable and certainly not worth the risk of serious neurologic deficit. When easily obtained it should be a goal. Neurologic decompression, partial reduction, and stabilization without perfect anatomic alignment can lead to excellent results in Grade III and IV listhesis cases regardless of the level, and there is no absolute data that suggests otherwise [35,36]. The same applies to the management of traumatic fracture dislocations, especially in the cervical and thoracic spine [41,44]. Evoked potentials may help if the surgeon feels compelled to attain perfect alignment via aggressive reduction. In the context of trauma, there is very little data to support that perfect anatomic reduction leads to less chronic local spine pain [41,44]. Again, the main goal is neurologic decompression. The author’s philosophy is not to risk neurologic function to make x-rays look perfect. One other problem primarily seen with pedicle screws is screw penetration of the exposed dural sac. CSF leakage has been reported in 2% of the cases [15]. If the screw will not enter the drilled hole easily, do not force it as sudden slippage can
Surg Neurol 571 1997;48:566 –74
lead to penetration of the dural sac as our experience shows. Only rapid reflexes prevented the screw from penetrating the cauda equina on two instances. We now know to make the entry hole into the pedicle large enough to prevent slippage. We also have an assistant to hold an instrument in such a fashion as to block the screw from going towards the exposed dura. Meticulous dural closure, fibrin glue, and lumbar subarachnoid drainage, as indicated, should prevent CSF leakage. Though perfection is impossible, there is no such thing as an acceptable neurologic complication in spinal surgery. There is essentially a zero risk with respect to nerve root or cord injury in placing the screws of anterior cervical or thoraco-lumbar plates [1,16,42]. The 0.3%– 0.5% risk of neurologic injury after anterior cervical, thoracic or lumbar surgery was attributable to the decompression or excessive manipulation [16,29]. The reported incidence of neurologic injury with posterior cervical lateral mass plates varies. One report on 44 patients used a technique that we have also followed [18]. We drill no deeper than 10 mm and use screws no longer than 14 –16 mm. Another report on 78 patients had an incidence of nerve root injury of 7%, 2 spinal cord deficits, 1 vertebral artery injury and 1 cerebellar infarct [23]. The same report did not specifically mention screw length, but many radiographs showed screws of at least 18 –20 mm in length. A longer screw, though capable of bicortical lateral mass purchase and increased biomechanical strength, appears to significantly increase the likelihood of a nerve root injury or a vertebral artery injury. Vascular injury is fortunately extremely uncommon. Posterior C1-C2 transarticular screw placement can lead to vertebral artery injury. Preoperative planning with CT scan in the sagittal plane through the foramen transverserium can demonstrate an aberrant course of the vertebral artery with reference to the C1-C2 articulation, thus precluding placement of a screw and avoiding vertebral artery injury (Apfelbaum, unpublished observations). This procedure has precluded us from placing one screw to date. The screws of a posterior cervical lateral mass plate led to vertebral artery injury in 0.2% of patients in the report in which screws longer than 16 mm were used [23]. There have been no vertebral artery injuries in reports using screws of 16 mm or less. The risk of vascular injury with pedicle screw placement in one large series was 0.1% [15]. The risk of vascular injury with anterior instrumentation seems to be near zero [16,42]. Isolated reports of aortic injury or iliac artery injury were usually attributable to the expo-
572 Surg Neurol 1997;48:566 –74
sure and decompression and not the placement of screws [16]. Spine fusion, internal fixation, surgeries of longer than 5 hours, drains, obesity, and diabetes are all known risk factors for infection [22,33,43,46,49]. Acute deep infections occur in 1.5–7% or more [4, 12,15,16,29,33,37,49]. Reported reoperation for deep infections occurred in 4.2% [15,33,46]. Lumbar pedicle screws have a reported rate of deep infection of 1– 4.2% [15,38,45,46,49]. Delayed infections occurred in 10% of a series of 103 posterior TSRH instrumentation cases at a mean of 2 years after surgery [37]. We have seen no acute or delayed deep infections. Because the exposure and decompression part of a trauma or tumor surgery can take time, the instrumentation is never exposed to air until right before placement. Reports concerning ventricular shunt and hip prosthesis infections have demonstrated airborne bacterial contamination of the hardware [2,32]. After placement of the instrumentation and fusion, we irrigate extensively with 1–2 L of bacitracin. Additionally, broad spectrum intravenous perioperative antibiotics and postoperative antibiotics should be given for at least 48 h, and we always continue the antibiotics while a chest tube is in place. Numerous reports support prophylactic antibiotics in reducing the risk of infection of implanted foreign bodies such as internal fixation hardware [21,25,27,39]. Meticulous hemostasis with Bovie cautery and gelfoam/thrombin paste is always used so that both wound hematomas and drains are avoided that adversely affect infection rates. Close follow-up by the surgeon is mandatory for wound inspection. Do not rely on physicians or nurses at outside hospitals to treat the wound correctly. At the first sign of superficial infection or breakdown, aggressive local care and antibiotics are instituted under the surgeon’s direct supervision. After 30 days, the most common complication is instrumentation dislodgment, breakage, or both [4, 5,12,15,18,20,23,24,29,38,45,46,50,51]. Most of these problems are attributed to a failure of arthrodesis. Biomechanically inferior instrumentation such as the plasma screw or inadequate constructs can also lead to failure [1,9,24]. Biologic reasons for weaker constructs can be osteoporosis and vertebral body tumor involvement adjacent to the operative pathology making for weak bone/metal interfacing or host bone and graft interfacing. We encountered one case of catastrophic host bone (vertebral body) failure attributable to osteoporosis and tumor involvement. The upper body failed, allowing the graft to telescope through the body despite no dislodgment of the posterior instrumentation. An
Shapiro and Snyder
anterior plate may have added support preventing this sudden collapse. This case highlights the ethical dilemma spine surgeons face: To perform radical decompressive and instrumentation procedures on patients with a high risk of complications and little chance of long-term survival. Though the patient and family were grateful for the additional 6 weeks of ambulation gained in this case, in no way can one justify the cost from an economic viewpoint. Asymptomatic breakage can occur but most spinal instrumentation breakage is symptomatic. Reported rates of instrumentation breakage, loosening, or dislodgment range from 2% to 25% [4,5,12, 15,18,20,23,24,29,38,45,46,50,51]. Broken pedicle screws at the bone-screw interface is the most common reported instrumentation failure, but broken rods, rod screw dislodgment, and screw backout can occur. It is felt that failure of arthrodesis will ultimately lead to instrumentation failure in most instances. Our 5% incidence of pedicle screw breakage is similar to most reports. Recent reports have documented a high incidence, approaching 65%, of pedicle screw fracturing/bending after shortsegment pedicle screw instrumentation for thoracolumbar burst fractures such as our case [3,11,30]. Very recent biomechanical laboratory data and clinical data suggest that the addition of supplemental offset hooks to the short segment pedicle screw construct will reduce the bending moments on the screws, reducing the likelihood of screw failure [9]. By staying current on construct outcome and biomechanics, hopefully improvements will continue. When in doubt, it may be best to perform a 360° instrumentation as we did in 6% of the cases, which carries a high success rate and low morbidity rate [2,48]. Reoperation rates for instrumented spinal fusions are not always easy to determine. Reported rates have ranged from 7% to 45%, with most in the 15%–25% range [4,5,12,15,18,20,23,24,29,38,45,46,50, 51]. Our 2% reoperation rate is lower than the vast majority of reports. Most of the problems requiring reoperation had ready explanations that should be avoidable in the future. There have been no anterior cervical screw fractures since abandoning the biomechanically inferior fenestrated plasma screw. It was inexcusable to not place the locking screw initially and that mistake should never recur. Because of the availability of longer plates along with nonfenestrated screws, we have seen no complications in the last 22 vertebrectomies. It was inexcusable to not trim the rods that were too long after compression of the lower claw and this should be avoided in the future. In the other two reoperations
Few Complications with Spine Instrumentation
in alcoholics, it is the authors’ opinion that the presence of hardware prevented catastrophe. Not all instrumentation breakage is catastrophic. Breakage with minimal symptoms and no progressive instability can be left in place or easily removed. There has been no complication related to hardware removal. It is difficult to reliably determine fusion radiographically with reported rates of accuracy of only 68% [26]. Our rate of arthrodesis/stability seems to be 99%. Removal of posterior thoraco-lumbar instrumentation after successful arthrodesis after 1 year and before breakage is advocated by some, but not always done. The authors have had the experience of seeing loss of reduction in other surgeons’ cases following elective removal of hardware. We leave the hardware in place and have yet to regret this decision. Because of this, we suspect that the longer instrumentation is left in place the greater the rate of instrument breakage or delayed infection will be. In conclusion, a very low long-term complication rate with spinal instrumentation is achievable. There were no neurologic injuries and the deep infection rate was zero. The reoperation rate was 2%. There has been one pseudoarthrosis with progressive kyphosis without neurologic compromise or reoperation. Spinal instrumentation is a safe and valuable adjunct to spinal surgery and we will continue to analyze our experience, learn from our mistakes, and do as well as possible with this challenging area of spine surgery.
Surg Neurol 573 1997;48:566 –74
9.
10.
11.
12. 13. 14. 15.
16.
17.
18. 19. 20.
REFERENCES 1. An HS, Lim TH, You JW, Hong JH, Eck J, McGrady L. Biomechanical evaluation of anterior thoracolumbar spinal instrumentation. Spine 1983;20:1979 – 83. 2. Bayston R, Lari J. A study of the sources of infection in colonized shunts. Dev Med Child Neurol 1974; 16(Suppl 32):16 –22. 3. Benson DR, Burkus JK, Montesano PX, Sutherland TB, McClain RF. Unstable thoracolumbar and lumbar burst fractures treated with the AO fixateur interne. J Spinal Disord 1992;5:335– 43. 4. Blumenthal S, Gill K. Complications of the Wiltse pedicle screw fixation system. Spine 1993;18:1867–71. 5. Boos N, Marchesi D, Aebi M. Survivorship analysis of pedicular fixation systems in the treatment of degenerative conditions of the lumbar spine. A comparison of Cotrel-Dubousset instrumentation and the AO internal fixator. J Spinal Disord 1992;5:403–9. 6. Bosacco SJ, Love-Mignogna S. Complications of lumbar disc surgery. Part I. ONA 1978;5:13–15. 7. Bosacco SJ, Love-Mignogna S. Complications of lumbar disc surgery. Part II. ONA 1978;5:20 –2. 8. Boyd DP, Farha GJ. Arteriovenous fistula in isolated vascular injuries secondary to intervertebral disc sur-
21. 22. 23. 24.
25.
26.
27.
gery. Report of 4 cases and review of the literature. Ann Surg 1965;161:524 –31. Chiba M, McLain RF, Yerby SA, Moseley TA, Smith TS, Benson DR. Short-segment pedicle instrumentation. Biomechanical analysis of supplemental hook fixation. Spine 1996;21:288 –94. Choudhury AR, Taylor JC, Whitaker R. Paraplegia complicating lumbar disc surgery. J Coll Surg Edinb 1979;24:167–9. Crutcher JP, Anderson PA, King HA, Montesano PX. Indirect spinal canal decompression in patients with thoracolumbar burst fractures treated by posterior distraction rods. J Spinal Disord 1991;4:39 – 48. Davne S, Meyers D. Complications of lumbar spinal fusion with transpedicular instrumentation. Spine 1992;17:S184 –9. DeSaussure RL. Vascular injury coincident to disc surgery. J Neurosurg 1959;16:222–9. Dick WT, Schnebel B. Severe spondylolisthesis: reduction and internal fixation. Clin Orthop 1988;232: 70 –9. Esses SI, Sachs B, Dreyzin V. Complications associated with the technique of pedicle screw fixation. A selected survey of ABS members. Spine 1993;18: 2231–9. Faciszewski T, Winter RB, Lonstein JE, Denis F, Johnson L. The surgical and medical perioperative complications of anterior spinal fusion surgery in the thoracic and lumbar spine in adults. A review of 1223 procedures. Spine 1995;20:1592–9. Farber G, Place HM, Mazur RA, Jones DE, Damiano TR. Accuracy of pedicle screw placement in lumbar fusions by plain radiographs and computed tomography. Spine 1995;20:1494 –9. Fehlings M, Cooper P, Errico T. Posterior plates in the management of cervical instability: long-term results in 44 patients. J Neurosurg 1994;81:341–9. Freeman DG. Major vascular complications of lumbar disc surgery. West J Surg 1961;69:175–7. Griffiths HJ. Orthopedic complications. Radiol Clin North Am 1995;33:401–10. Haines SJ, Walters BC. Antibiotic prophylaxis for cerebrospinal fluid shunts: a meta-analysis. Neurosurgery 1995;36:1057– 8. Harle A, van Ende R. Management of wound sepsis after spinal fusion surgery. ACTA Orthopaedica Belgica 1991;57(Suppl 1):242– 6. Heller JG, Silcox DH, Sutterlin CE. Complications of posterior cervical plating. Spine 1995;20:2442– 8. Hollowell J, Reinartz J, Pintar F, Morgese V, Maiman DJ. Failure of Synthes anterior cervical fixation device by fracture of morscher screws: a biomechanical study. J Spinal Disorders 1994;7:120 –5. Josefsson G, Gudmundsson G, Kolmert L, Wijkstrom S. Prophylaxis with systemic antibiotics versus gentamicin bone cement in total hip arthroplasty. A five year survey of 1688 hips. Clin Orthop Rel Res 1990; 253:173– 8. Kant AP, Daum WJ, Dean SM, Uchida T. Evaluation of lumbar spine fusion. Plain radiographs versus direct surgical exploration and observation. Spine 1995;20: 2313–7. Langley JM, LeBlanc JC, Drake J, Milner R. Efficacy of antimicrobial prophylaxis in placement of cerebro-
574 Surg Neurol 1997;48:566 –74
28.
29. 30. 31. 32.
33. 34.
35.
36. 37.
38. 39.
40. 41. 42. 43.
spinal fluid shunts: meta-analysis. Clin Infect Diseases 1993;17:98 –103. Lenke LG, Padberg AM, Russo MH, Bridwell KH, Gelb DE. Triggered electromyographic threshold for accuracy of pedicle screw placement. Spine 1995;20:1585– 91. McAfee PC. Complications of anterior approaches to the thoracolumbar spine. Emphasis on Kaneda instrumentation. Clin Orthop Rel Res 1994;306:110 –9. McLain RG, Sparling E, Benson DR. Early failure of short-segment pedicle instrumentation for thoracolumbar fracture. J Bone Joint Surg 1993;75:162–7. McLaren AC, Bailey SI. Cauda equina syndrome. A complication of lumbar discectomy. Clin Orthop Rel Res 1986;204:143–9. Marotte JH, Lord GA, Blanchard JP, Guillamon JL, Samuel P, Servant JP, Mercier PH. Infection rate in total hip arthroplasty as a function of air cleanliness and antibiotic prophylaxis. A 10 year experience with 2,384 cementless Lord madreporic prostheses. J Arthroplasty 1987;2:77– 82. Massie JB, Heller JG, Abitbol JJ, McPherson D, Garfin SR. Postoperative posterior spinal wound infections. Clin Orthop Rel Res 1992;284:99 –108. Nolte LP, Zamorano LJ, Jiang Z, Wang Q, Langlotz F, Berlemann U. Image-guided insertion of transpedicular screws. A laboratory setup. Spine 1995;20:497– 500. Peek RD, Wiltse LL, Reynolds JB, Thomas JC, Guyer DW, Widell EH. In situ arthrodesis without decompression for grade III or IV isthmic spondylolisthesis in adults who have severe sciatica. J Bone Joint Surg 1989;71A:62– 8. Pizzutillo PD, Mirenda W, MacEwen GD. Posterolateral fusion for spondylolisthesis in adolescence. J Pediatr Orthop 1986;6:311– 6. Richards BS, Herring JA, Johnston CE, Birch JG, Roach JW. Treatment of adolescent idiopathic scoliosis using Texas Scottish Rite Hospital instrumentation. Spine 1994;19:1598 –1605. Roy-Camille R, Saillant G, Mazel C. Internal fixation of lumbar spine with pedicle screw plating. Clin Orthop 1986;203:7–17. Savitz M. The use of prophylactic antibiotics in neurosurgery. In: Morley TB, ed. Current controversies in neurosurgery. Philadelphia: WB Saunders Co. 1976: 648 –50. Schwab FJ, Nazarian DG, Mahmud F, Michelsen CB. Effects of spinal instrumentation on fusion of the lumbar spine. Spine 1995;20:2023– 8. Shapiro S. Banked fibula and the locking anterior cervical plate in anterior cervical fusions following cervical discectomy. J Neurosurg 1996;84:161–5. Shapiro S. Management of unilateral facet of the cervical spine. Neurosurgery 1993;33:832–7. Simchen E, Stein H, Sacks TG, Shapiro M, Michel J. Multivariate analysis of determinants of postoperative wound infection in orthopaedic patients. J Hosp Infect 1984;5:137– 46.
Shapiro and Snyder
44. Simpson AHRW, Williamson DM, Golding SH, Houghton GR. Thoracic spine translocation without cord injury. J Bone Joint Surg 1990;72B:80 –3. 45. Soini J, Laine T, Pohjolainen T, Hurri H, Alaranta H. Spondylodesis augmented by transpedicular fixation in the treatment of olisthetic and degenerative conditions of the lumbar spine. Clin Orthop 1993;297: 111– 6. 46. Steffee AD, Biscup RS, Sitkowski DJ. Segmental spine procedures with pedicle screw fixation: a new internal fixation device for disorders of the lumbar and thoracolumbar spine. Clin Orthop 1986;203:45–53. 47. Sundaresan N, Galicich JH, Lane JM. Treatment of neoplastic epidural cord compression by vertebral body resection and stabilization. J Neurosurg 1991; 63:650 –76. 48. Sundaresan N, Steinberger AA, Moore F, Sachdev VP, Krol G, Hough L, Kelliher K. Indications and results of combined anterior-posterior approaches for spine tumor surgery. J Neurosurg 1996;85:438 – 46. 49. Thalgott J, Cotler H, Sasso R, LaRocca H, Gardner V. Postoperative infections in spinal implants. Classification and analysis: a multi-center study. Spine 1991; 16:981– 4. 50. West J, Ogilvie J, Bradford D. Complications of the variable screw plate pedicle screw fixation. Spine 1991;16:576 –9. 51. Whitecloud TS III, Butler JC, Cohen JL, Candelora PD. Complications with the variable spinal plating system. Spine 1989;14:472– 6. 52. Zuckerman J, Hsu K, Picetti G, White A, Wynne G, Taylor L. Clinical efficacy of spinal instrumentation in lumbar degenerative disc disease. Spine 1992;17: 834 –7.
COMMENTARY
The authors have clearly demonstrated that with careful neurosurgical technique, the use of spinal instrumentation in the cervical, thoracic, and lumbar spine can be accomplished with a rather low complication rate. Given the observations that these instrumentations improve our ability to correct and stabilize spinal deformities and improve our rates of spinal arthrodesis, the authors’ data clearly demonstrate that segmental spinal instrumentation can be used with an acceptably low complication rate and that their benefit far exceeds their risk. George W. Sypert, M.D. Neurosurgeon Fort Myers, Florida
Spinal Instrumentation With A Low Complication Rate Scott A. Shapiro, M.D. and William Snyder, M.D. Section of Neurosurgery, Indiana University Medical Center, Indianapolis, Indiana
Shapiro SA, Snyder W. Spinal instrumentation with a low compliance rate. Surg Neurol 1997;48:566 –74. BACKGROUND
Spinal instrumentation has become an increasing part of the armamentarium of neurosurgery and neurosurgical training. For noncontroversial indications for spine fusion the arthrodesis rate seems to be better. For both noncontroversial and controversial indications, the reported complication rate with spinal instrumentation tends to be greater than that with noninstrumented spine surgeries. These reported complications include a 2–3% neurologic injury rate, 3– 45% reoperation rate for implant failure, and infection rates of 5–10%. Therefore, we report on 299 cases that have undergone spinal instrumentation placed exclusively by neurosurgeons with a very low complication rate. METHODS
Two hundred ninety-nine consecutive spinal instrumentation cases performed exclusively by neurosurgeons at Indiana University Medical Center were analyzed for complications related to spinal instrumentation. The spinal instrumentation placed consisted of 195 anterior cervical locking plates, 22 cases of posterior cervical instrumentation, 9 cases of combined anterior locking plates with posterior cervical instrumentation, 14 anterior thoracolumbar plates, 51 posterior thoraco-lumbar instrumentation cases, and 8 combined anterior/posterior thoracolumbar instrumentation cases. RESULTS
The mean follow-up is 40 months (6 –95). There was one perioperative death unrelated to the spinal instrumentation. There were no neurologic injuries and there has been no hardware infection to date. There were two dural tears, three superficial wound infections, and three minor wound breakdowns successfully treated. Hardware complications included three cervical plate/screw extrusions reoperated, one cervical plate fracture reoperated, one posterior cervical screw backout not reoperated, one case of broken pedicle screws not reoperated, one vertebral body failure not reoperated, and one posterior rod case reoperated for excessive rod length and protrusion. The overall complication rate attributable to placement of spinal instrumentation was 10/299 (3%) with a reoperation rate of 2%. The arthrodesis rate was 298/299 (99%). Address reprint requests to: Scott Shapiro, M.D., Room 323, East Outpatient Building, Wishard Memorial Hospital, 1001 W. 10th Street, Indiana University Medical Center, Indianapolis, IN, 46202. Received January 31, 1997; accepted April 24, 1997. 0090-3019/97/$17.00 PII S0090-3019(97)00296-6
CONCLUSION
The complication rate for using spinal instrumentation can be less than previously reported. Lessons learned and discussed should reduce the rate even more. Spinal instrumentation is a safe and useful adjunct to fusion in treating degenerative, traumatic, infectious, and neoplastic diseases of the spine. © 1997 by Elsevier Science Inc. KEY WORDS
Complications, hardware infection, spine fusion, spinal instrumentation.
pine surgery in general has a very low mortality rate in the range of 0.1– 0.3% which is usually attributable to the underlying illness and not the surgery [6,7]. Morbidity for elective herniated discs and non-instrumented spinal decompressions is also quite low in the modern era, but remains quantifiable. Lumbar decompressive, noninstrumented surgery is complicated by large vessel injury in 0.1%, cauda equina syndrome in 0.2%, nerve root damage in 0.2– 0.5%, infection in 1%, and dural tears in 1% [6 – 8,10,13,19,31]. Cervical spine surgery is complicated by 2% anterior bone graft failures, 0.4% CSF leaks, 0.2% root injury, 0.3% quadriplegia, and 0.7% mortality (Osenbach et al, unpublished data, 42). Internal fixation hardware to augment stability and improve spinal fusion for all levels of the spine has become increasingly popular, despite intense controversy surrounding specific constructs. Relatively noncontroversial indications include severe traumatic instability, reconstruction after corpectomy for trauma, neoplasm or spondylosis, scoliosis, and Grade III or more L5S1 spondylolisthesis [18,38,40,45–48]. Controversial indications include internally fixated fusion for degenerative disc disease without instability, degenerative listhesis above L5-S1, and long-level cervical laminectomy [5,23,37,38,45,46,52]. Numerous complications related to the placement of internal fixation hardware have been reported. These complications can occur intraoperatively or postoperatively. Intraoperative complications are primarily
S
© 1997 by Elsevier Science Inc. 655 Avenue of the Americas, New York, NY 10010
Few Complications with Spine Instrumentation
neurologic injuries, dural injuries, vascular injuries, and malplacement of the instrumentation. Postoperative complications include infection and instrumentation breakage/dislodgment (with or without pseudoarthrosis and/or recurrent instability or deformity). Previous reports have documented a 2–3% neurologic injury rate, 3– 45% reoperation rate for implant failure, and infection rates of 5–10% [4,5,12, 13,15,16,20,22,29,33,50]. There is some variability depending on the type and location of instrumentation, but the general finding is one of more complications as compared with discectomy and noninstrumented decompression. This is an important concept as neurosurgery has become increasingly involved with internal fixation and is passing these techniques on to the next generation of neurosurgeons. Many of the intraoperative and some of the postoperative complications are avoidable and must be avoided to justify the use of the various systems. Thus, we report a large experience with spinal instrumentation with a low complication rate.
Patient Population Two hundred ninety-nine consecutive cases of patients with spinal internal fixation placed exclusively by neurosurgery are analyzed. The internal fixation analysis will be subdivided into following groups: anterior cervical internal fixation, posterior cervical internal fixation, combined anterior and posterior cervical internal fixation, anterior thoraco-lumbar internal fixation, posterior thoracolumbar internal fixation, and combined anterior and posterior thoraco-lumbar internal fixation. Anterior grafts under compression were cadaveric fibula or tibia in 99%. Posterior grafts for lateral mass fusions were always autologous bone. Allogeneic bone matrix has been used to augment 199 (66%) fusions. The overall mean follow-up for all patients is 40 months (6 –95).
Perioperative Management Universal to All Levels and Types of Internal Fixation Any patient requiring decompression or reduction of a spinal segment involving the spinal cord receives perioperative steroids and intravenous mannitol. All patients receive prophylactic intravenous antibiotics intra-operatively and for at least 48 h after surgery. All sterilized spinal instrumentation is
Surg Neurol 567 1997;48:566 –74
left covered until the moment it is to be placed. Traffic in and out of the room is reduced as much as possible. After placement of the instrumentation, aggressive hemostasis using electrocautery and gelfoam thrombin powder is always used, followed by extensive irrigation with 1–2 L of antibiotic solution. No drain is left in any posterior instrumentation case with the rare exception of a patient with coagulopathy. The only drain that is left in an anterior case is a chest tube for thoraco-lumbar anterior approaches. The rare posterior bulb suction drain that is left is removed within 24 h and an intraoperatively placed suture to close the drain site is tied down immediately to close the drain site. Intravenous antibiotics are continued as long as the chest tube is in place. The wound must be checked frequently by someone knowledgeable until healed completely. At the first sign of superficial infection, attention to local wound care and intravenous antibiotics are instituted.
Anterior Cervical Instrumentation (N 5 195) This group included 195 anterior locking cervical plates (Synthes) including 146 anterior cervical discectomy (ACD) fusions; 31 single-level vertebrectomies; and 18 multilevel vertebrectomies for spondylosis, trauma, and tumor. All trauma and vertebrectomy cases were managed in a rigid orthosis or halo and the nontraumatic ACD cases were managed in a soft collar. There have been no infections and no neurologic complications in any of the 195 cases. Of the 146 ACD-banked fibula locking plate fusions, there were two complications. One occurred in a patient treated for a facet fracture dislocation with disc disruption by a singlelevel ACD banked fibula, locking plate fusion at C5-C6. The patient’s radiculopathy improved completely. Three weeks after surgery, while intoxicated from alcohol abuse, the patient fell while in a rigid collar, suffering a fracture of his banked fibula graft with 2 mm extrusion of the upper part of the plate and 3– 4 mm subluxation. He presented with radicular recurrence. He was revised to a combined anterior-banked fibula graft and a posterior locking plate fusion with autologous bone and was placed in a halo. Despite numerous patient-induced problems with the halo, the patient went on to arthrodesis and remained neurologically normal. The second ACD-banked fibula locking plate complication also occurred because of alcohol abuse when the plate fractured from a high-speed motor vehicle accident 6 months after surgery, and was subse-
568 Surg Neurol 1997;48:566 –74
Shapiro and Snyder
dependent, asthmatic patient with a C5 burst fracture with severe spinal cord injury exsanguinated at 4 weeks from a catastrophic gastric stress ulcer. Before her demise, her neurologic condition had improved significantly. All patients went on to arthrodesis. Thus, the overall reoperation rate for anterior cervical plate placement was 5/195 (2.6%).
Posterior Cervical Instrumentation (N 5 22)
Lateral cervical radiograph of a two-level vertebrectomy demonstrating fracturing of the lower two plasma screws along with graft and plate extrusion.
1
quently removed with a good outcome. It was noted that the graft had fused. There were no single-level vertebrectomy complications. There were 3/18 (11%) multilevel vertebrectomy plate/screw complications. The first complication was a two-level vertebrectomy done before the availability of plates longer than 60 mm. The author used a shorter plate with abnormal screw positioning that fractured out of the lower body, with recurrent neck and arm pain 2 days postoperatively. The patient was revised without instrumentation and did well. A two-level vertebrectomy with a locking plate and plasma screws presented at 8 weeks with fracturing of the lower screws, partial extrusion of the lower plate and graft, and dysphagia (Figure 1). She was revised and has done well. A third complication was in a twolevel vertebrectomy in which a locking screw was inadvertently not placed and the screw backed out 2 days postoperatively. This required reoperation and was easily corrected. One obese, steroid-
There were 12 lateral mass plates (Axis, Danek, Inc) placed for trauma (unilateral/bilateral facet dislocation), rheumatoid arthritis, or neoplasia. All had placement of autologous bone in the facet and lateral masses. There were seven cases where C1-C2 transarticular screws (Aesculap) were placed along with an interspinous type fusion using braided cable (Danek) and autologous iliac crest for trauma or rheumatoid arthritis. There were three instrumented craniocervical fusions with autologous iliac crest, two of which used lateral mass plates with C2 pedicle screws in addition to lateral mass screws at C3 and C4 and braided cables for the occiput. There were no infections, nerve, or vertebral artery injuries. There has been one posterior cervical plate complication where a single screw partially backed out which was not reoperated, because the patient was asymptomatic. Twenty patients were kept in rigid cervical collars and two were kept in halos, and all maintained reduction and went on to arthrodesis.
Combined Anterior/ Posterior Cervical Instrumentation (N 5 9) There have been nine banked fibula, locking plate, anterior cervical fusions with posterior lateral mass plates and autologous bone lateral mass fusions performed for fracture dislocations with disc disruption. There have been no complications in this group and all have gone on to arthrodesis. All but one were managed with rigid orthoses.
Anterior Thoracolumbar Instrumentation (Z-plate or TSRH) (N 5 14) Fourteen patients were instrumented for trauma or tumor including 13 Z-plates (Danek) and 1 Synthes
Few Complications with Spine Instrumentation
Surg Neurol 569 1997;48:566 –74
plate. There were no neurologic complications, vascular complications, or dural injuries. There have been no infections and only one minor wound breakdown successfully treated with dressing changes. There have been no screw or plate complications and all have fused. All were managed with thoraco-lumbar-sacral orthoses.
Posterior Thoracolumbar Instrumentation (CD or TSRH) (N 5 51 cases) The primary system used was TSRH (Danek) in 48 cases and Cotrel-Dubousset (Sofamor) in three cases. Indications included trauma (n 5 21), degenerative listhesis (n 5 8), spondylolisthesis (n 5 6), tumor (n 5 13), and degenerative disc disease (n 5 3). A total of 76 pedicle screws have been placed. Other constructs have made use of the claw technique using primarily pedicle hooks and transverse process hooks. We rarely use laminar hooks. There were no neurologic injuries or pedicle fractures. There were two small dural tears caused by pedicle screw slippage during insertion that were repaired primarily without leakage. There have been three superficial wound infections treated early with local wound care and intravenous antibiotics followed by oral antibiotics. There were no deep infections. There were two minor wound breakdowns successfully treated with dressing changes. An 82-year-old woman who suffered a traumatic severe T-12 burst fracture with resultant severe pain, paraparesis, and sphincter impairment underwent a T-12 vertebrectomy, banked tibia vertebral body replacement, and posterior pedicle screw/rod construct and lateral mass autologous rib fusion placed from T-11 to L-1. The patient was neurologically normal and ambulating in a brace by 2 weeks. The banked tibia graft angulated asymptomatically somewhere between 4 and 12 weeks after surgery and was noted on routine radiographic follow-up (Figure 2). Nothing was done other than periodic observation, and the patient returned to work after 6 months. A 57-year-old white female with metastatic breast cancer had severe destruction of three contiguous thoracic bodies with severe spinal cord compression and paraparesis leaving her bedridden. She had a successful three-level thoracic corpectomy with banked tibia anterior fusion and posterior TSRH hook/rod distraction instrumentation with autologous rib lateral mass fusion that led to the
Lumbar radiograph after L1 vertebrectomy demonstrating asymptomatic angulation of a tibial graft that had been initially been vertically oriented.
2
return of ambulation. Seven weeks after the fusion, the upper bordering body that also had cancer failed, with subsequent telescoping of the tibial graft through the body; the posterior construct remained unchanged. This patient became paraplegic acutely from this host/graft failure and died from liver failure caused by multiple metastases within 3 weeks. Screw threads of 1–2 mm in dimension outside of the lateral or medial aspect of the pedicle identified on follow-up CT scan occurred in 8%, though this had no effect on the operative result. There was 1 case out of 17 where the upper pedicle screws broke. This occurred 14 months after surgery in a tobacco- and alcohol-abusing patient with a L1 burst, flexion distraction injury who remained neurologically intact. The patient was instrumented in distraction with pedicle screws, with improvement in the kyphosis but no improvement in the compression of the L1 body, along with a lateral mass autologous iliac crest fusion. The patient developed low back pain 14 months postoperatively. Lumbar
570 Surg Neurol 1997;48:566 –74
Shapiro and Snyder
Combined Anterior/Posterior Thoracolumbar Instrumentation and Fusion (N 5 8) The indications were trauma in five, tumor in two, and infection in one. All cases had evidence for three column involvement. Seven consisted of anterior Z-plates with posterior TSRH instrumentation and one consisted of anterior and posterior TSRH. There were no complications in this group of eight patients and all have gone on to arthrodesis.
Discussion
Thoracic radiograph demonstrating inadequate trimming of the lower rods bilaterally that led to painful protrusion in this lymphoma patient, requiring a second procedure to trim the rods.
3
radiographs and CT scan demonstrated that the upper screws had broken and the patient had reangulated to the same degree as before surgery, but remained neurologically intact and has very little chronic pain. There have been no other cases of instrumentation breakage, dislodgment, or progressive deformity to date. Two additional cancer patients died from distant disease before 1 year with no spinal complications. Radiographs in patients alive at 1 year showed arthrodesis. Five additional cancer related deaths have occurred without failure of the spinal construct at 13–29 months. One lymphoma case was reoperated 18 months after posterior TSRH rod/hook placement to trim the inferior rods that were left too long and caused painful protrusion that interfered with upright posture and walking (Figure 3). The patient dramatically improved after rod trimming. Thus the reoperation rate is 1/51 (2%) with no cases of hardware removal to date.
In most large series of spinal instrumentation cases there continues to be a reported rate of neurologic injury that ranges from 0.4% to 22% [4,5,12,15,16, 23,29,37,38,45,46,50,51]. Neurologic injury, above all other complications, should not occur. Causes may be multifactorial and occasionally without explanation, but most can be explained. With reference to pedicle screws, the majority of neurologic injuries reported seem to be attributable to faulty placement of too large a diameter screw. In a series of 617 cases of pedicle screw placement, faulty placement occurred in at least 5.2% of the cases [15]. CT scan evaluation of pedicle screw placement has documented as high as 30% malplacement that ranges from catastrophic canal penetration to slight medial penetration of the pedicle screw [17]. There is no data to suggest that slight medial penetration by 1–2 mm adversely affects the outcome. Pedicle fracturing by the screw occurred in 2.2% of the cases [15]. Selection of screw diameter size should be the largest screw that will safely fit in the pedicle. Axial CT scans provide accurate measurements to help guide the choice of screw diameter and angulation. Table 1 lists the recommended screw diameter per level. Just because it is recommended that a 6.5–7.5 mm screw should be used at L5 does not mean that a 5.5 mm screw will not work. Many surgeons do not use intraoperative fluoroscopy and rely on anatomic landmarks. Presently we use fluoroscopy (A-P and lateral) in addition to CT scans and intraoperative anatomy to guide every pedicle screw placement, and find the radiographic guidance indispensable. It is the authors’ opinion that the placement of pedicle screws without radiographic guidance is substandard surgical practice and will eventually lead to an unacceptable and avoidable neurologic injury. Triggered EMG moni-
Few Complications with Spine Instrumentation
1
Recommended Screw Diameter
LEVEL T11 T12 L1 L2 L3 L4 L5 S1 S2
SCREW DIAMETER (MM) 5.0–5.5 5.0–5.5 5.5–6.0 5.5–6.0 6.5 6.5–7.5 6.5–7.5 6.5–7.5 6.5
toring can also be used to indicate root irritation when placing the screw [28]. We have used this on two smaller patients but have yet to have a root irritation sign. There is no question that fluoroscopy slows down the case significantly as one swings the machine from A-P to lateral in guiding the precise trajectory of the drilling and screw placement. Speed is not the goal of the operation. Neurologic decompression, restoration of stability and/or prevention of further slippage is the goal. Follow-up CT scan is often used to assess our pedicle screw placement. It seems that frameless stereotaxis may improve our ability to precisely cannulate the pedicle [34]. Another cause of neurologic injury appears to be excessive manipulation to attain anatomic alignment [14,37]. Perfect anatomic alignment is often unattainable or unsustainable and certainly not worth the risk of serious neurologic deficit. When easily obtained it should be a goal. Neurologic decompression, partial reduction, and stabilization without perfect anatomic alignment can lead to excellent results in Grade III and IV listhesis cases regardless of the level, and there is no absolute data that suggests otherwise [35,36]. The same applies to the management of traumatic fracture dislocations, especially in the cervical and thoracic spine [41,44]. Evoked potentials may help if the surgeon feels compelled to attain perfect alignment via aggressive reduction. In the context of trauma, there is very little data to support that perfect anatomic reduction leads to less chronic local spine pain [41,44]. Again, the main goal is neurologic decompression. The author’s philosophy is not to risk neurologic function to make x-rays look perfect. One other problem primarily seen with pedicle screws is screw penetration of the exposed dural sac. CSF leakage has been reported in 2% of the cases [15]. If the screw will not enter the drilled hole easily, do not force it as sudden slippage can
Surg Neurol 571 1997;48:566 –74
lead to penetration of the dural sac as our experience shows. Only rapid reflexes prevented the screw from penetrating the cauda equina on two instances. We now know to make the entry hole into the pedicle large enough to prevent slippage. We also have an assistant to hold an instrument in such a fashion as to block the screw from going towards the exposed dura. Meticulous dural closure, fibrin glue, and lumbar subarachnoid drainage, as indicated, should prevent CSF leakage. Though perfection is impossible, there is no such thing as an acceptable neurologic complication in spinal surgery. There is essentially a zero risk with respect to nerve root or cord injury in placing the screws of anterior cervical or thoraco-lumbar plates [1,16,42]. The 0.3%– 0.5% risk of neurologic injury after anterior cervical, thoracic or lumbar surgery was attributable to the decompression or excessive manipulation [16,29]. The reported incidence of neurologic injury with posterior cervical lateral mass plates varies. One report on 44 patients used a technique that we have also followed [18]. We drill no deeper than 10 mm and use screws no longer than 14 –16 mm. Another report on 78 patients had an incidence of nerve root injury of 7%, 2 spinal cord deficits, 1 vertebral artery injury and 1 cerebellar infarct [23]. The same report did not specifically mention screw length, but many radiographs showed screws of at least 18 –20 mm in length. A longer screw, though capable of bicortical lateral mass purchase and increased biomechanical strength, appears to significantly increase the likelihood of a nerve root injury or a vertebral artery injury. Vascular injury is fortunately extremely uncommon. Posterior C1-C2 transarticular screw placement can lead to vertebral artery injury. Preoperative planning with CT scan in the sagittal plane through the foramen transverserium can demonstrate an aberrant course of the vertebral artery with reference to the C1-C2 articulation, thus precluding placement of a screw and avoiding vertebral artery injury (Apfelbaum, unpublished observations). This procedure has precluded us from placing one screw to date. The screws of a posterior cervical lateral mass plate led to vertebral artery injury in 0.2% of patients in the report in which screws longer than 16 mm were used [23]. There have been no vertebral artery injuries in reports using screws of 16 mm or less. The risk of vascular injury with pedicle screw placement in one large series was 0.1% [15]. The risk of vascular injury with anterior instrumentation seems to be near zero [16,42]. Isolated reports of aortic injury or iliac artery injury were usually attributable to the expo-
572 Surg Neurol 1997;48:566 –74
sure and decompression and not the placement of screws [16]. Spine fusion, internal fixation, surgeries of longer than 5 hours, drains, obesity, and diabetes are all known risk factors for infection [22,33,43,46,49]. Acute deep infections occur in 1.5–7% or more [4, 12,15,16,29,33,37,49]. Reported reoperation for deep infections occurred in 4.2% [15,33,46]. Lumbar pedicle screws have a reported rate of deep infection of 1– 4.2% [15,38,45,46,49]. Delayed infections occurred in 10% of a series of 103 posterior TSRH instrumentation cases at a mean of 2 years after surgery [37]. We have seen no acute or delayed deep infections. Because the exposure and decompression part of a trauma or tumor surgery can take time, the instrumentation is never exposed to air until right before placement. Reports concerning ventricular shunt and hip prosthesis infections have demonstrated airborne bacterial contamination of the hardware [2,32]. After placement of the instrumentation and fusion, we irrigate extensively with 1–2 L of bacitracin. Additionally, broad spectrum intravenous perioperative antibiotics and postoperative antibiotics should be given for at least 48 h, and we always continue the antibiotics while a chest tube is in place. Numerous reports support prophylactic antibiotics in reducing the risk of infection of implanted foreign bodies such as internal fixation hardware [21,25,27,39]. Meticulous hemostasis with Bovie cautery and gelfoam/thrombin paste is always used so that both wound hematomas and drains are avoided that adversely affect infection rates. Close follow-up by the surgeon is mandatory for wound inspection. Do not rely on physicians or nurses at outside hospitals to treat the wound correctly. At the first sign of superficial infection or breakdown, aggressive local care and antibiotics are instituted under the surgeon’s direct supervision. After 30 days, the most common complication is instrumentation dislodgment, breakage, or both [4, 5,12,15,18,20,23,24,29,38,45,46,50,51]. Most of these problems are attributed to a failure of arthrodesis. Biomechanically inferior instrumentation such as the plasma screw or inadequate constructs can also lead to failure [1,9,24]. Biologic reasons for weaker constructs can be osteoporosis and vertebral body tumor involvement adjacent to the operative pathology making for weak bone/metal interfacing or host bone and graft interfacing. We encountered one case of catastrophic host bone (vertebral body) failure attributable to osteoporosis and tumor involvement. The upper body failed, allowing the graft to telescope through the body despite no dislodgment of the posterior instrumentation. An
Shapiro and Snyder
anterior plate may have added support preventing this sudden collapse. This case highlights the ethical dilemma spine surgeons face: To perform radical decompressive and instrumentation procedures on patients with a high risk of complications and little chance of long-term survival. Though the patient and family were grateful for the additional 6 weeks of ambulation gained in this case, in no way can one justify the cost from an economic viewpoint. Asymptomatic breakage can occur but most spinal instrumentation breakage is symptomatic. Reported rates of instrumentation breakage, loosening, or dislodgment range from 2% to 25% [4,5,12, 15,18,20,23,24,29,38,45,46,50,51]. Broken pedicle screws at the bone-screw interface is the most common reported instrumentation failure, but broken rods, rod screw dislodgment, and screw backout can occur. It is felt that failure of arthrodesis will ultimately lead to instrumentation failure in most instances. Our 5% incidence of pedicle screw breakage is similar to most reports. Recent reports have documented a high incidence, approaching 65%, of pedicle screw fracturing/bending after shortsegment pedicle screw instrumentation for thoracolumbar burst fractures such as our case [3,11,30]. Very recent biomechanical laboratory data and clinical data suggest that the addition of supplemental offset hooks to the short segment pedicle screw construct will reduce the bending moments on the screws, reducing the likelihood of screw failure [9]. By staying current on construct outcome and biomechanics, hopefully improvements will continue. When in doubt, it may be best to perform a 360° instrumentation as we did in 6% of the cases, which carries a high success rate and low morbidity rate [2,48]. Reoperation rates for instrumented spinal fusions are not always easy to determine. Reported rates have ranged from 7% to 45%, with most in the 15%–25% range [4,5,12,15,18,20,23,24,29,38,45,46,50, 51]. Our 2% reoperation rate is lower than the vast majority of reports. Most of the problems requiring reoperation had ready explanations that should be avoidable in the future. There have been no anterior cervical screw fractures since abandoning the biomechanically inferior fenestrated plasma screw. It was inexcusable to not place the locking screw initially and that mistake should never recur. Because of the availability of longer plates along with nonfenestrated screws, we have seen no complications in the last 22 vertebrectomies. It was inexcusable to not trim the rods that were too long after compression of the lower claw and this should be avoided in the future. In the other two reoperations
Few Complications with Spine Instrumentation
in alcoholics, it is the authors’ opinion that the presence of hardware prevented catastrophe. Not all instrumentation breakage is catastrophic. Breakage with minimal symptoms and no progressive instability can be left in place or easily removed. There has been no complication related to hardware removal. It is difficult to reliably determine fusion radiographically with reported rates of accuracy of only 68% [26]. Our rate of arthrodesis/stability seems to be 99%. Removal of posterior thoraco-lumbar instrumentation after successful arthrodesis after 1 year and before breakage is advocated by some, but not always done. The authors have had the experience of seeing loss of reduction in other surgeons’ cases following elective removal of hardware. We leave the hardware in place and have yet to regret this decision. Because of this, we suspect that the longer instrumentation is left in place the greater the rate of instrument breakage or delayed infection will be. In conclusion, a very low long-term complication rate with spinal instrumentation is achievable. There were no neurologic injuries and the deep infection rate was zero. The reoperation rate was 2%. There has been one pseudoarthrosis with progressive kyphosis without neurologic compromise or reoperation. Spinal instrumentation is a safe and valuable adjunct to spinal surgery and we will continue to analyze our experience, learn from our mistakes, and do as well as possible with this challenging area of spine surgery.
Surg Neurol 573 1997;48:566 –74
9.
10.
11.
12. 13. 14. 15.
16.
17.
18. 19. 20.
REFERENCES 1. An HS, Lim TH, You JW, Hong JH, Eck J, McGrady L. Biomechanical evaluation of anterior thoracolumbar spinal instrumentation. Spine 1983;20:1979 – 83. 2. Bayston R, Lari J. A study of the sources of infection in colonized shunts. Dev Med Child Neurol 1974; 16(Suppl 32):16 –22. 3. Benson DR, Burkus JK, Montesano PX, Sutherland TB, McClain RF. Unstable thoracolumbar and lumbar burst fractures treated with the AO fixateur interne. J Spinal Disord 1992;5:335– 43. 4. Blumenthal S, Gill K. Complications of the Wiltse pedicle screw fixation system. Spine 1993;18:1867–71. 5. Boos N, Marchesi D, Aebi M. Survivorship analysis of pedicular fixation systems in the treatment of degenerative conditions of the lumbar spine. A comparison of Cotrel-Dubousset instrumentation and the AO internal fixator. J Spinal Disord 1992;5:403–9. 6. Bosacco SJ, Love-Mignogna S. Complications of lumbar disc surgery. Part I. ONA 1978;5:13–15. 7. Bosacco SJ, Love-Mignogna S. Complications of lumbar disc surgery. Part II. ONA 1978;5:20 –2. 8. Boyd DP, Farha GJ. Arteriovenous fistula in isolated vascular injuries secondary to intervertebral disc sur-
21. 22. 23. 24.
25.
26.
27.
gery. Report of 4 cases and review of the literature. Ann Surg 1965;161:524 –31. Chiba M, McLain RF, Yerby SA, Moseley TA, Smith TS, Benson DR. Short-segment pedicle instrumentation. Biomechanical analysis of supplemental hook fixation. Spine 1996;21:288 –94. Choudhury AR, Taylor JC, Whitaker R. Paraplegia complicating lumbar disc surgery. J Coll Surg Edinb 1979;24:167–9. Crutcher JP, Anderson PA, King HA, Montesano PX. Indirect spinal canal decompression in patients with thoracolumbar burst fractures treated by posterior distraction rods. J Spinal Disord 1991;4:39 – 48. Davne S, Meyers D. Complications of lumbar spinal fusion with transpedicular instrumentation. Spine 1992;17:S184 –9. DeSaussure RL. Vascular injury coincident to disc surgery. J Neurosurg 1959;16:222–9. Dick WT, Schnebel B. Severe spondylolisthesis: reduction and internal fixation. Clin Orthop 1988;232: 70 –9. Esses SI, Sachs B, Dreyzin V. Complications associated with the technique of pedicle screw fixation. A selected survey of ABS members. Spine 1993;18: 2231–9. Faciszewski T, Winter RB, Lonstein JE, Denis F, Johnson L. The surgical and medical perioperative complications of anterior spinal fusion surgery in the thoracic and lumbar spine in adults. A review of 1223 procedures. Spine 1995;20:1592–9. Farber G, Place HM, Mazur RA, Jones DE, Damiano TR. Accuracy of pedicle screw placement in lumbar fusions by plain radiographs and computed tomography. Spine 1995;20:1494 –9. Fehlings M, Cooper P, Errico T. Posterior plates in the management of cervical instability: long-term results in 44 patients. J Neurosurg 1994;81:341–9. Freeman DG. Major vascular complications of lumbar disc surgery. West J Surg 1961;69:175–7. Griffiths HJ. Orthopedic complications. Radiol Clin North Am 1995;33:401–10. Haines SJ, Walters BC. Antibiotic prophylaxis for cerebrospinal fluid shunts: a meta-analysis. Neurosurgery 1995;36:1057– 8. Harle A, van Ende R. Management of wound sepsis after spinal fusion surgery. ACTA Orthopaedica Belgica 1991;57(Suppl 1):242– 6. Heller JG, Silcox DH, Sutterlin CE. Complications of posterior cervical plating. Spine 1995;20:2442– 8. Hollowell J, Reinartz J, Pintar F, Morgese V, Maiman DJ. Failure of Synthes anterior cervical fixation device by fracture of morscher screws: a biomechanical study. J Spinal Disorders 1994;7:120 –5. Josefsson G, Gudmundsson G, Kolmert L, Wijkstrom S. Prophylaxis with systemic antibiotics versus gentamicin bone cement in total hip arthroplasty. A five year survey of 1688 hips. Clin Orthop Rel Res 1990; 253:173– 8. Kant AP, Daum WJ, Dean SM, Uchida T. Evaluation of lumbar spine fusion. Plain radiographs versus direct surgical exploration and observation. Spine 1995;20: 2313–7. Langley JM, LeBlanc JC, Drake J, Milner R. Efficacy of antimicrobial prophylaxis in placement of cerebro-
574 Surg Neurol 1997;48:566 –74
28.
29. 30. 31. 32.
33. 34.
35.
36. 37.
38. 39.
40. 41. 42. 43.
spinal fluid shunts: meta-analysis. Clin Infect Diseases 1993;17:98 –103. Lenke LG, Padberg AM, Russo MH, Bridwell KH, Gelb DE. Triggered electromyographic threshold for accuracy of pedicle screw placement. Spine 1995;20:1585– 91. McAfee PC. Complications of anterior approaches to the thoracolumbar spine. Emphasis on Kaneda instrumentation. Clin Orthop Rel Res 1994;306:110 –9. McLain RG, Sparling E, Benson DR. Early failure of short-segment pedicle instrumentation for thoracolumbar fracture. J Bone Joint Surg 1993;75:162–7. McLaren AC, Bailey SI. Cauda equina syndrome. A complication of lumbar discectomy. Clin Orthop Rel Res 1986;204:143–9. Marotte JH, Lord GA, Blanchard JP, Guillamon JL, Samuel P, Servant JP, Mercier PH. Infection rate in total hip arthroplasty as a function of air cleanliness and antibiotic prophylaxis. A 10 year experience with 2,384 cementless Lord madreporic prostheses. J Arthroplasty 1987;2:77– 82. Massie JB, Heller JG, Abitbol JJ, McPherson D, Garfin SR. Postoperative posterior spinal wound infections. Clin Orthop Rel Res 1992;284:99 –108. Nolte LP, Zamorano LJ, Jiang Z, Wang Q, Langlotz F, Berlemann U. Image-guided insertion of transpedicular screws. A laboratory setup. Spine 1995;20:497– 500. Peek RD, Wiltse LL, Reynolds JB, Thomas JC, Guyer DW, Widell EH. In situ arthrodesis without decompression for grade III or IV isthmic spondylolisthesis in adults who have severe sciatica. J Bone Joint Surg 1989;71A:62– 8. Pizzutillo PD, Mirenda W, MacEwen GD. Posterolateral fusion for spondylolisthesis in adolescence. J Pediatr Orthop 1986;6:311– 6. Richards BS, Herring JA, Johnston CE, Birch JG, Roach JW. Treatment of adolescent idiopathic scoliosis using Texas Scottish Rite Hospital instrumentation. Spine 1994;19:1598 –1605. Roy-Camille R, Saillant G, Mazel C. Internal fixation of lumbar spine with pedicle screw plating. Clin Orthop 1986;203:7–17. Savitz M. The use of prophylactic antibiotics in neurosurgery. In: Morley TB, ed. Current controversies in neurosurgery. Philadelphia: WB Saunders Co. 1976: 648 –50. Schwab FJ, Nazarian DG, Mahmud F, Michelsen CB. Effects of spinal instrumentation on fusion of the lumbar spine. Spine 1995;20:2023– 8. Shapiro S. Banked fibula and the locking anterior cervical plate in anterior cervical fusions following cervical discectomy. J Neurosurg 1996;84:161–5. Shapiro S. Management of unilateral facet of the cervical spine. Neurosurgery 1993;33:832–7. Simchen E, Stein H, Sacks TG, Shapiro M, Michel J. Multivariate analysis of determinants of postoperative wound infection in orthopaedic patients. J Hosp Infect 1984;5:137– 46.
Shapiro and Snyder
44. Simpson AHRW, Williamson DM, Golding SH, Houghton GR. Thoracic spine translocation without cord injury. J Bone Joint Surg 1990;72B:80 –3. 45. Soini J, Laine T, Pohjolainen T, Hurri H, Alaranta H. Spondylodesis augmented by transpedicular fixation in the treatment of olisthetic and degenerative conditions of the lumbar spine. Clin Orthop 1993;297: 111– 6. 46. Steffee AD, Biscup RS, Sitkowski DJ. Segmental spine procedures with pedicle screw fixation: a new internal fixation device for disorders of the lumbar and thoracolumbar spine. Clin Orthop 1986;203:45–53. 47. Sundaresan N, Galicich JH, Lane JM. Treatment of neoplastic epidural cord compression by vertebral body resection and stabilization. J Neurosurg 1991; 63:650 –76. 48. Sundaresan N, Steinberger AA, Moore F, Sachdev VP, Krol G, Hough L, Kelliher K. Indications and results of combined anterior-posterior approaches for spine tumor surgery. J Neurosurg 1996;85:438 – 46. 49. Thalgott J, Cotler H, Sasso R, LaRocca H, Gardner V. Postoperative infections in spinal implants. Classification and analysis: a multi-center study. Spine 1991; 16:981– 4. 50. West J, Ogilvie J, Bradford D. Complications of the variable screw plate pedicle screw fixation. Spine 1991;16:576 –9. 51. Whitecloud TS III, Butler JC, Cohen JL, Candelora PD. Complications with the variable spinal plating system. Spine 1989;14:472– 6. 52. Zuckerman J, Hsu K, Picetti G, White A, Wynne G, Taylor L. Clinical efficacy of spinal instrumentation in lumbar degenerative disc disease. Spine 1992;17: 834 –7.
COMMENTARY
The authors have clearly demonstrated that with careful neurosurgical technique, the use of spinal instrumentation in the cervical, thoracic, and lumbar spine can be accomplished with a rather low complication rate. Given the observations that these instrumentations improve our ability to correct and stabilize spinal deformities and improve our rates of spinal arthrodesis, the authors’ data clearly demonstrate that segmental spinal instrumentation can be used with an acceptably low complication rate and that their benefit far exceeds their risk. George W. Sypert, M.D. Neurosurgeon Fort Myers, Florida