SPINAL STENOSIS

The etiology of spinal stenosis is most commonly either
congenital (developmental), acquired (degenerative), or a combination
of both (Table 147.1) (2).
The majority of cases of spinal stenosis are acquired, being caused by
degenerative changes occurring in the three-joint complex consisting of
the intervertebral disc and the two facet joints. In some cases, such
degenerative changes may be

superimposed
on a pre-existing congenital stenosis. Variations in the shape, as well
as the size, of the spinal canal may predispose the patient to spinal
stenosis, with a trefoil canal being associated with lateral recess
stenosis more commonly than a round or oval canal.

Comorbidity refers to the presence and severity of conditions other than the disorder under study or treatment (56).
The relationship between comorbidity and outcome is more commonly
applied to surgical than medical outcomes. Comorbidity typically
increases with age and is associated with a poor outcome for many
medical and surgical conditions (20,25,107).
Sick people have a higher mortality rate, a higher complication rate,
and a lower level of function than do healthy patients. It is
imperative to take this factor into account when assessing and
comparing outcomes between treatment groups. If such factors are not
taken into account, differences in outcome between treatment groups
could reflect differences in patient comorbidities rather than
differences as a result of treatment.

Rates of hospital morbidity and mortality following lumbar spinal surgery are greater with increasing age of the patient (25).
Complications are more frequent with advancing patient age, increasing
the complexity of both diagnosis and surgical treatment. The study by
Deyo et al. (25) reported an overall mortality
of 0.07% for 18,122 hospitalizations between 1986 through 1988. The
mortality increased with age, increasing to 0.6% (ninefold increase) in
patients older than 75 years of age. The overall complication rate of
9.1% increased to 17.7% in patients 75 years of age or older.

As would be expected, an increase in comorbidity is
often associated with more in-hospital complications and perioperative
mortality. This finding is independent of age alone. Oldridge et al. (86)
found an age-related increase in mortality only for patients older than
80 years of age. There was, however, a significant increase in
in-hospital and 1-year cumulative mortality associated with increasing
number of comorbidities.

In a retrospective review of 88 patients undergoing laminectomy for spinal stenosis, Katz (56) concluded that the long-term outcome was generally less favorable than had been previously reported (Table 147.6).
By 1 year after surgery, 6% of patients had a second operation and, by
the time of the last follow-up, 17% had a repeat surgery. Only 40% of
patients with the highest comorbidity score had a good outcome at the
time of final follow-up compared with 75% of patients who had the
lowest comorbidity score (P = 0.004). The most common comorbidities
were osteoarthritis (32%), cardiac disease (22%), rheumatoid arthritis
(10%), and chronic pulmonary disease (7%). Their data suggested that
the effect of comorbidities was additive, because no single comorbidity
was significantly associated with worse outcome. In a subsequent study
by the same authors, comorbidity was found to be the second most
important determinant of disability in lumbar canal stenosis, with
complaints of predominantly LBP (as opposed to leg pain) preoperatively
being the most important contributor to disability (55,58).

The gold standard surgical procedure for spinal stenosis is decompressive laminectomy. This procedure may involve either bilateral laminectomy or hemilaminectomy.
For bilateral laminectomy, the lamina and ligamentum flavum are removed
on both sides of the stenotic level or levels to the lateral recess.
Decompression begins at the most distal extent of neural compression
and proceeds in a caudal-to-cranial direction. Although the L5–S1 level
is rarely compressed centrally, owing to the capacity of the spinal
canal at that level, decompression is most safely initiated at that
level rather than at L4–L5, the most commonly involved level, which is
often severely stenotic. Perform decompression sequentially, from
medial to lateral.

Following midline and lateral decompression, check for
the presence or absence of a concomitant disc herniation, which might
contribute to neural compression. Such herniations may be located
either posterolaterally, foraminally, or extraforaminally. Unless the
disc is contributing to definite neural compression, it is generally
best to avoid discectomy in the presence of laminectomy because
subsequent instability is more likely to occur when both anterior and
posterior supporting structures are violated. When laminectomy is
accompanied by discectomy, consider performing an arthrodesis at the
time of surgery.

Because spinal stenosis is a global degenerative
process, encompassing multiple levels and involving nerve roots
bilaterally, multilevel bilateral laminectomy is commonly required.
There is, however, some debate as to whether it is more appropriate to
decompress only the symptomatic level and side, or whether all stenotic
levels should be decompressed. The argument against decompression of
asymptomatic root levels or sides is the risk of producing symptoms at
a previously asymptomatic level or side. On the other hand, failure to
decompress a stenotic but asymptomatic level or side risks progression
of the degenerative process with the development of more severe and
potentially symptomatic stenosis. In addition, the natural tendency for
degenerative changes to progress over time makes it possible that, in
time, asymptomatic stenotic levels will eventually become stenotic.
Indeed, several studies have reported long-term deterioration following
initially successful surgical decompression (18,19,56,60,90,91 and 92).

Hemilaminectomy involves
unilateral, rather than bilateral, removal of bone and ligamentum
flavum. Because the spinous processes, interspinous ligaments, and
supraspinous ligaments are preserved medially, normal stabilizing
structures are retained with less risk of development of postoperative
instability. Take care to preserve the pars interarticularis laterally
in order to minimize risk of postoperative instability (16,106).
Hemilaminectomy is appropriate for patients with unilateral symptoms
from stenosis. A disadvantage of this procedure is the relative
difficulty of performing contralateral decompression and also in
obtaining enough medial exposure to perform an adequate ipsilateral
decompression in patients with foraminal stenosis. The presence of an
intact spinous process and interspinous or supraspinous ligament
complex makes it difficult to angle the Kerrison rongeur laterally
enough to insert the jaw of the rongeur into the depths of the neural
foramen. Under such circumstances, removal of the midline spinous
process and interspinous or supraspinous ligament complex may be
necessary in order to allow the proper angulation of the rongeur to
perform the foraminal decompression. In addition to preserving midline
stabilizing structures, hemilaminectomy also avoids exposure of, and
potential injury to, the contralateral facet joint. Because the
integrity of the unexposed contralateral facet is maintained, more
aggressive decompression of a nerve root by partial, or even total,
ipsilateral

facetectomy need not necessarily be accompanied by a fusion.

Contralateral nerve root decompression may be
accomplished through a unilateral hemilaminectomy approach by tilting
the table away from the operating surgeon (Fig. 147.1).
Particularly when used in conjunction with an operating microscope,
which provides excellent illumination and which can be angled to
visualize the opposite side, contralateral decompression can be
accomplished without the need for removal of stabilizing midline
structures (spinous processes and interspinous or supraspinous
ligaments). The contralateral neural foramen can be visualized and
decompressed, and its more distal portion can be palpated with a long
bent probe such as a #4 Penfield elevator or a contoured probe.
Although offering the advantage of preserving normal, noncompressing
midline structures and minimizing scar tissue on the opposite side,
this technique is more demanding than bilateral laminectomy because
decompression is performed through a more limited exposure and the
determination of adequate foraminal patency is more dependent on feel
(palpation) than by direct visualization. In addition, there is a
greater potential for dural laceration from the Kerrison rongeur when
working through a small opening. Should such a dural tear occur, its
repair often necessitates complete (bilateral) laminectomy with
adequate exposure of the dural rent.

A recent review of the literature for spinal stenosis
surgery failed to identify even a single randomized trial comparing
surgery and conservative treatment (Table 147.7) (113). Turner et al. (113)
attempted a meta-analysis of the literature on surgical outcomes for
spinal stenosis, but the poor scientific quality of the literature
precluded the authors from conducting the intended meta-analysis. Even
using the authors’ own ratings, the average proportion of
good-to-excellent outcomes was only 72%. This study found no
statistically significant relationship between outcome and patient age,
gender, presence of prior back surgery or number of levels operated on.
In those studies reporting on only patients with degenerative
spondylolisthesis, the outcome was better. There was no statistically
significant difference in outcome between decompression with or without
associated fusion. This observation is particularly significant in
light of the reported increased morbidity associated with lumbar fusion
(114).

In contrast, the prospective cohort study of Atlas et al. (3)
reported the 1-year outcome of patients with spinal stenosis treated
surgically or nonsurgically in the state of Maine and found that at 1
year, 55% of the surgical patients reported definite improvement in
their predominant symptom, compared with only 28% of the nonsurgical
group. Surgery was found to increase the relative odds of “definite
improvement” 2.6 fold compared with nonsurgical treatment.

In a large retrospective series Katz et al. (60) followed 88 patients for a period of 2.8 to 6.8 years (Table 147.6).
Outcome assessment included a questionnaire in which the patients rated
their outcomes in terms of pain and function. The authors reported a
surprisingly high failure rate, with 11% of patients reporting a poor
outcome at 1 year and 43% reporting poor outcome at final follow-up.
Six percent of patients had repeat lumbar surgery

within
the first year and 17% had additional surgery by the time of last
follow-up. The authors concluded that the long-term outlook for
patients undergoing decompressive laminectomy for spinal stenosis is
guarded owing to progressive deterioration of the results over time.
They suggested that more extensive bone removal may be indicated at the
time of initial surgery.

The authors also reported a high initial success rate
following surgery, but by 5 years, the failure rate had reached 27%,
with a predicted failure rate of 50% within the anticipated life
expectancy of most patients. More than half (62%) of these failures
were due to subsequent neurologic symptoms, with an equal incidence of
recurrent stenosis at the same level and stenosis at a new level.
Because of the high rate of failure from recurrent stenosis, the
authors recommended that all levels of impending stenosis be
decompressed along with the symptomatic levels.

Despite the fact that some studies report deterioration
of standard bilateral decompressive laminectomy over time, more limited
alternatives to decompressive laminectomy and hemilaminectomy have been
espoused in order to avoid removal of normal, noncompressing structures
and thereby minimize risk of postoperative instability (18,54,60,91). Such procedures include hemilaminotomy, wide fenestration, and laminoplasty. Hemilaminotomy involves a more limited decompression than hemilaminectomy (Fig. 147.2).
Rather than removing an entire hemilamina, hemilaminotomy removes only
the ligamentum flavum and adjacent portions of two hemilaminae
responsible for neural compression. This procedure is more commonly
performed in younger patients with unilateral focal stenosis in whom
extensive laminectomy carries the risk of instability. It may also be
considered in older patients who do not have extensive global stenosis.

Wide fenestration is a
procedure described for central stenosis in which only the medial
portion of the inferior facets and adjacent ligamentum flavum is
removed (69,83,123).
Care is taken to remove only pathologic anatomy and to preserve the
interspinous or supraspinous ligament complex and spinous processes,
which make up the midline stabilizing structures. This may be performed
by using bilateral laminotomies at one or more segmental levels,
removing the ligamentum flavum (Fig. 147.2D). In a 5-year follow-up study of this procedure, 82% of patients had good or excellent early surgical outcomes, but results

deteriorated to 71% satisfactory by 4 years postoperatively (83).

Laminoplasty is an alternative to laminectomy that was originally advocated for active manual workers (76,77).
This procedure is similar to cervical laminoplasty and involves hinging
open the lamina on one side and inserting the excised spinous processes
into the open hinge in order to keep it patent. There is not sufficient
experience with this technique to provide outcomes assessment.

The role of fusion in the treatment of spinal stenosis
is somewhat controversial. For stenosis not associated with
degenerative spondylolisthesis or other deformity, most studies report
that simple decompression is the preferred method of surgical
treatment. For patients with associated degenerative spondylolisthesis,
concomitant fusion is generally recommended (30,44). The issue of using supplementary spinal instrumentation is yet unresolved (32,125).

In a recent prospective, randomized study of 45 patients
undergoing either decompression alone or decompression with fusion for
spinal stenosis without associated instability, there was no
significant difference in outcome between fused and unfused groups (Table 147.8) (40).
Overall, 78% of patient-reported and 80% of examiner-rated results were
rated very good or good. When broken down by type of procedure
performed, there were no significant differences in outcome between the
three groups with regard to pain relief. The authors concluded that
surgical decompression changed the natural history of spinal stenosis,
resulting in generally favorable outcome and improved quality of life
in the majority of patients. They further concluded that arthrodesis
was not justified in the absence of radiographically proven segmental
instability because there was no statistical difference in outcome
between the three treatment groups.

Degenerative spondylolisthesis was first described in 1930 by Junghanns who coined the term pseudospondylolisthesis to describe the presence of forward slippage of a vertebral body in the presence of an intact neural arch (54a).
The clinical and pathologic features of this entity were further
defined by Macnab, who described the condition as “spondylolisthesis
with an intact neural arch” (70). The term degenerative spondylolisthesis
was originally used by Newman and Stone and is the terminology most
commonly used to describe the anterior slippage of one vertebral body
on another in the presence of an intact neural arch (84a).

Degenerative spondylolisthesis may be a source of both
LBP and leg pain and may contribute to radicular or referred leg pain
in a characteristic pattern of neurogenic claudication (36).
The diagnosis is typically made on lateral radiographs, but it may have
a dynamic component to it such that the slip may reduce in the supine
position and, therefore, may be readily apparent only on stress
radiographs. Such radiographs may include standing lateral views,
sitting or standing flexion-extension views, or distraction compression
radiography (14,41).

As with spinal stenosis generally, little is known of the natural history of degenerative spondylolisthesis. Mardjetko et al. (74)
attempted a meta-analysis of the literature from 1970 to 1993. Only
three papers, reporting on 278 patients, described the natural history
of degenerative spondylolisthesis (33,78,96). Overall, 90 of these 278 patients (32%) achieved satisfactory results untreated (Table 147.9).

Matsunaga et al. (78)
presented a study of 40 patients who received no treatment and who were
followed for at least 5 years (range: 5 to 14 years.; mean: 8.25 year).
Progressive slip was noted in 12 patients (30%), although no
correlation was noted between slip progression and worsening of
symptoms. Only 4 of 40 patients (10%) showed clinical deterioration
over the course of the study, all of whom were in the group of 28
patients showing no slip progression over the follow-up period.
Interestingly, none of the 12 patients with slip progression
deteriorated clinically. Therefore, the majority of the patients in
this study showed a slight improvement in their clinical symptoms over
time, although only 1/3 were felt to have satisfactory function at
final follow-up.

Although current thinking generally favors concomitant
fusion for spinal stenosis associated with degenerative
spondylolisthesis, decompression without fusion is also a viable
therapeutic option (Table 147.10) (74).
Overall, 69% of patients from Mardjetko’s meta-analysis reported
satisfactory outcome with decompression without fusion, with 31% having
an unsatisfactory result and 31% having progression of their slip.
There was generally no correlation between clinical outcome and amount
of slip progression except in the study by Bridwell et al., which
showed a positive correlation between the two (15).

A recent report by Epstein and Epstein (29)
reviewed 290 patients undergoing decompression without fusion for
degenerative spondylolisthesis. Only patients with a “stable” slip, as
defined by a slip having less than 4 mm translation and less than 10°
to 12° angulation on dynamic lateral radiographs, were included. Two
hundred and fifty patients had one-level listhesis and 40 had a
two-level slip. Decompressive procedures included laminectomy in 249
patients and fenestration procedures in 41 patients. Fenestration
procedures typically involved bilateral laminotomy with partial medial
facetectomy and foraminotomy. At an average 10-year follow-up (range: 1
to 27 years), 69% of patients exhibited excellent, 13% good, 12% fair,
and 6% poor outcome. The authors concluded that 82% excellent or good
outcome was very acceptable in their elderly population (average age:
67 years old), in whom fusion is associated with higher morbidity and
mortality (26).

Poor surgical outcome following laminectomy without
fusion was reported in the prospective randomized study by Herkowitz
and Kurz (44) comparing decompression alone with combined decompression and noninstrumented fusion (Table 147.11).
In the decompression group, only 11 of 25 patients (44%) had a
satisfactory result. This group of patients was found to have
significantly more LBP and leg pain than their fused counterparts.
Furthermore, the mean slip increased from an average of 5.3 mm
preoperatively to 7.9 mm postoperatively. Other authors have reported a
similar experience (15,30).

The role of fusion in the surgical treatment of spinal
stenosis associated with degenerative spondylolisthesis is less
controversial than the role of fusion in the treatment of other
degenerative back conditions (80,115). Caputy and Luessenhop (18)
reported a retrospective review of 96 patients undergoing decompressive
surgery for spinal stenosis who were followed for at least 5 years. The
treatment failed in 16 patients because of recurrent neural
involvement, and it failed in 10 patients because of LBP (total
failures = 26). The authors concluded that because of the higher
incidence of recurrent symptoms in patients with pre-existing
degenerative spondylolisthesis, all patients with an associated slip
should undergo fusion of the listhetic level.

In their prospective and randomized study comparing
decompression alone with decompression and noninstrumented spinal
fusion in the treatment of degenerative spondylolisthesis with spinal
stenosis, Herkowitz and Kurz (44) reported superior results when concomitant fusion was performed with the decompression (Table 147.11).
The reported outcome for the arthrodesis group was excellent in 44% and
good in 52% (96% excellent or good total results), whereas in the
nonarthrodesis group, only 8% reported an excellent outcome and 36%
reported a good outcome (44% excellent or good total results) (P =
0.0001). There was a significant increase in the preoperative slip in
patients not receiving an arthrodesis compared with those undergoing
fusion (P = 0.002). Interestingly, 36% of those undergoing attempted
arthrodesis were noted to have a pseudarthrosis, all of whom had either
an excellent or a good result. This study concluded that the results of
surgical decompression with in situ
arthrodesis are superior to those of decompression alone. The authors
further concluded that the decision for concomitant arthrodesis should
be based purely on the presence or absence of a preoperative slip
rather than on other preoperative factors, such as the age or sex of
the patient or the disc height, or on intraoperative factors such as
the amount of bone resected during the decompression.

The prospective randomized study by Bridwell et al. (15)
included a subgroup of 11 patients undergoing decompression and
noninstrumented fusion. Of the 10 patients available for follow-up,
only 3 (30%) reported improved functional outcome and seven had an
increase in their preoperative spondylolisthesis.

Postacchini and Cinotti (90) reported on bone regrowth occurring an average of 8.6 years after surgical decompression for spinal stenosis (92).
Although all 16 patients with degenerative spondylolisthesis showed
some bone regrowth, the degree of regrowth was more severe in the six
patients who did not undergo arthrodesis. Furthermore, the proportion
of satisfactory results was significantly higher in patients who had
spinal fusion (Table 147.12). Although this
study was nonrandomized and retrospective, it suggested that
arthrodesis stabilizes the spine, resulting in less bone regrowth and
superior long-term results.

Zdeblick (125) reported a prospective and randomized study of 124 patients undergoing either instrumented or

noninstrumented fusion for a variety of diagnoses. The overall fusion
rate for the noninstrumented group was 65%; for the semirigid fixation
group, the fusion rate was 77%; and for the rigid fixation group, it
was 95%. A trend for better clinical outcome with increasing rigidity
of fixation was also observed. Seventy-one percent of the
noninstrumented patients, 89% of the semirigid group, and 95% of the
rigid group reported excellent or good results. For the subgroup of
patients with degenerative spondylolisthesis, 65% of the
noninstrumented patients fused compared with 50% of the semirigid
fixation group, and 86% of the rigid fixation group had a good or
excellent result. Subsequent studies have also reported superior
results with concomitant arthrodesis and decompression for spinal
stenosis with degenerative spondylolisthesis (18,44,60,90).

The historical cohort study of spinal fusion using pedicle screw fixation reported by Yuan et al. (124)
involved a retrospective, multicenter study of 2,684 patients with
degenerative spondylolisthesis. Solid radiographic fusion was noted in
89% of patients undergoing pedicle screw fixation compared with 70% of
those without instrumentation. Clinical outcome was also better in the
group of patients undergoing instrumented fusion.

Nork et al. (85) reported a
retrospective study of 30 patients undergoing decompression and
instrumented fusion for degenerative spondylolisthesis. Outcome was
determined by fusion rate, a functional questionnaire, and the SF-36
survey. Both the rate of fusion and patient satisfaction was 93%.
Thirteen patients (43%) had complications, including dural tears (three
patients), excessive blood loss (two patients), pseudarthrosis (two
patients), pulmonary embolus (PE) (one patient), deep infection (one
patient), urinary tract infections (3 patients), and unstable angina
(one patient).

A recent randomized prospective study of posterolateral
lumbar fusion, with and without pedicle screw instrumentation, for a
variety of conditions concluded that the addition of instrumentation
did not produce an incremental clinical benefit to that obtained from
noninstrumented fusion, although there was a slight nonsignificant
trend toward a higher fusion rate in the instrumented fusion group (34).
This study, involving a mean clinical follow-up of 40 months,
prospectively examined 71 patients undergoing posterolateral fusion for
either failed back surgery syndrome (FBSS), degenerative disc disease,
isthmic spondylolisthesis, or degenerative spondylolisthesis. For the
10 patients who had degenerative spondylolisthesis, five underwent
instrumented fusion and five underwent fusion in situ.
Eighty percent of the patients with degenerative spondylolisthesis
undergoing instrumented fusion achieved an excellent or good outcome,
compared with 40% of those without instrumentation. For the small
subgroup of 10 patients with degenerative spondylolisthesis, the
clinical outcome appeared to be better than that of the overall
population studied, although this subgroup was too small to establish
statistical significance.

There does not appear to be a clear consensus as to the
optimal way to treat the patient with degenerative spondylolisthesis.
Most studies suggest that patients undergoing concomitant fusion do
better when decompression is accompanied by fusion (44). It is less clear, however, whether or not the fusion should be augmented with instrumentation (32).
It would seem reasonable that if there is clear evidence of instability
on flexion-extension radiographs, the immediate stability provided by
instrumentation would warrant the additional time, expense, and
potential morbidity associated with its use. On the other hand, the
indication for its use in the patient with a collapsed disc space and
no motion at the spondylolisthetic level is less clear.

All lumbar spine surgery, and indeed all surgeries
generally, share certain broad groups of potential complications, which
can be thought of as occurring either preoperatively, intraoperatively, or postoperatively.

PE is the third most common cause of death in the United
States, accounting for up to 200,000 deaths annually. In hospitalized
patients, PE is the most common preventable cause of hospital death
with pulmonary emboli detectable in more than one quarter of all
routine autopsies. The etiology of pulmonary embolism includes the
immobilization associated with hospitalization as well as factors
related to surgery itself, which produce a hypercoagulable state. DVT
is the precursor to PE in 90% of cases and is common in hospitalized
patients. The risk of DVT following general surgery ranges between 5%
and 63% and is particularly high with certain orthopaedic conditions,
such as fracture of the hip, and following some orthopedic procedures,
particularly total hip and total knee arthroplasty, in which the
incidence of DVT following unprotected joint replacement is as high as
60% to 80%.

Although it is uncommon, DVT has been reported to occur
following scoliosis surgery. DVT following routine spinal
decompressions, however, was thought to be a rare occurrence. More
recently, DVT has been recognized following spinal surgery (108,119).
Using postoperative duplex scanning, the incidence of DVT in
unprotected patients undergoing posterior lumbar surgery has been
reported to be 14% (119). The use of elastic
compression stockings or intermittent pneumatic compression stockings
(PCS) has been shown to reduce the incidence of DVT diagnosed by duplex
scanning to 0.9% to 6% (108). Bell et al.
reported the incidence of venographically proven DVT following
unprotected surgery for lumbar disc herniation or spinal stenosis
performed under spinal anesthesia to be 25.8% (6A).
This rate is significantly higher than that reported using duplex
scanning as the method of diagnosis, thereby reflecting the greater
accuracy of venography in diagnosing DVT (30a,30b,119).
Prophylaxis with PCS reduced the incidence to 4.5% in patients
receiving spinal anesthesia. PCS seemed to provide no significant
protection from DVT in patients receiving general anesthesia,

in
whom the incidence of DVT was 13.6% in unprotected spinal surgery and
8.1% with PCS protection. This study suggested that the best
combination of type of anesthesia and DVT prophylaxis in terms of
prevention of DVT was spinal anesthesia with PCS. The worst combination
was spinal anesthesia without PCS. See Chapter 5 for more details as well as recommendations for treatment

Both laminectomy and laminotomy share common bony,
soft-tissue, and neural anatomy, and therefore, share a common list of
potential complications. These complications include inadequate neural
decompression, recurrent stenosis, incidental durotomy, neural injury,
epidural hematoma, neural compression from either fat grafts or other
barriers to scar formation, vascular injury, and late instability.

Although it is not a complication in the usual sense of
the term, failure to obtain symptomatic relief of radicular leg pain
that is not due to an error in surgical decision making should be
considered at least an adverse effect of surgery. Its avoidance
requires precise correlation of the preoperative imaging study with the
clinical picture and surgical anatomy, and demands that surgery be
continued until the offending neural compression is found. It also
requires a thorough knowledge of surgical anatomy and of the potential
sources and sites of neural compression, as described by MacNab (70).
In addition, it is imperative that the surgeon have a precise
understanding of the potential anatomic variations in the location of
disc herniations so that he will know precisely where to look for
neural compression, particularly when the predicted pathology is not
found (110).

It is important for the surgeon to recognize and look
for additional sites of neural compression that may account for
inadequate relief following decompression of only one site. This
condition is sometimes referred to as a “double crush phenomenon” and
is thought to be at least partially due to venous congestion of the
neural segment located between the two sites of compression resulting
in a compartment syndrome–like condition of the intervening segment.
Multiple sites of compression are common with spinal stenosis, which is
a global condition frequently involving multilevel, bilateral neural
compression. Sites of compression include central compression of the
cauda equina and lateral compression, either within the lateral recess,
within the neural foramen, or extraforaminally. It is important to
identify all clinically significant sites of neural compression and to
decompress those levels adequately.

Distinguishing recurrent symptoms due to neural
compression from those due to scar formation is a complex
decision-making process that requires a precise history and
high-quality radiographic imaging. Failure to obtain even temporary
pain relief following decompressive lumbar surgery suggests either
inadequate neural decompression, irreversible neural damage already
present at the time of surgery, or a nonspinal cause for the pain. A
short pain-free interval of less than 6 months suggests development of
scar formation as the cause of recurrent pain. Recurrence of pain
following a long pain-free interval of more than 6 to 12 months
suggests a new process such as a recurrent disc herniation or recurrent
stenosis.

Because spinal stenosis may occur in association with
disc herniation, recurrence of symptoms following decompression
involving discectomy could be due to recurrent disc herniation. The
overall reported incidence of recurrent disc herniation is
approximately 3% (38). Its incidence following
laminectomy or laminotomy associated with discectomy is unknown but
could be even greater if decompression involved destabilization from
facetectomy and resulted in instability (31).

Several studies have reported that the surgical results of spinal stenosis surgery deteriorate over time (18,56,60,91).
Although this may be due to many factors, such as associated
comorbidity and advanced patient age, surgical and pathologic factors
are also important. These factors include progression of degenerative
changes at unoperated levels (60), regrowth of bone at the operated levels (90,91), pre-existing instability (degenerative scoliosis or spondylolisthesis) (35),
and development of postoperative instability [for example, due to
resection of one or more facets at a single segmental level (1) or development of a facet fracture at the level of decompression during or following decompression (95)].
All of the above-mentioned factors can lead to recurrent LBP or
radicular leg pain following decompressive surgery for spinal stenosis.

Incidental violation of the dura (durotomy) is a
well-recognized complication of spinal surgery that has a reported
incidence of 0.3% to 13%. It most commonly occurs when the edge of a
biting instrument, such as a Kerrison rongeur, inadvertently grabs the
dura and produces either a punctate hole in the dura or a frank
laceration. Usually, the injury is noted immediately by the sudden
appearance of cerebrospinal fluid (CSF) within the wound. Occasionally,
however, the tear is not noted until

sometime
later by the clinical appearance of persistent spinal headache, the
presence of CSF drainage from the wound, or by the onset of an obvious
swelling in the patient’s back suggesting a pseudomeningocele (75). The incidence of the latter complication has been estimated between 0.07 and 2 per cent.

Neural injury may occur due to direct trauma to the
nerve root itself during surgery. This injury may be due to excessive
neural retraction, contusion, laceration, or electrocauterization (111).
The incidence of neurologic complications following lumbar spine
surgery has been estimated to be 0.2%. Such injury may be suspected
postoperatively by the presence of a new or increased objective
neurologic deficit, or by the onset of new parasthesias. This
condition, sometimes referred to as the “battered root,” may occur
either as an unavoidable consequence of severe neural compression or as
a result of indelicate surgery (11). Meticulous
surgical technique, therefore, is of paramount importance in order to
minimize such complications. Adequate surgical exposure is imperative
in order to minimize excessive neural retraction. In cases of a large
midline disc herniation or a disc herniation associated with spinal
stenosis, for example, a bilateral laminectomy rather than a keyhole
laminotomy may be required in order to remove the disc fragment safely.

It is important to perform an anteroposterior x-ray
study of the lumbar spine in order to identify bony anatomy that might
be of surgical significance. Note the presence of spina bifida occulta
or a pre-existing laminectomy defect, for example, on the preoperative
radiograph because the presence of either of these features mandates
cautious surgical exposure in order to minimize risk of damage to
underlying dura and nerve roots. Carefully examine other pre-operative
studies to identify other potentially significant anatomic variants,
such as anomalous nerve roots, that could be injured during surgical
decompression (111).

It is important always to visualize the lateral edge of
the nerve root during surgical decompression to be sure that exposure
is not inadvertently being performed in the axilla of a nerve root
where accidental dural laceration and neural injury could occur and
where repair of such injuries is particularly difficult. This is also
important during revision lumbar surgery, in which dissection should
usually be performed lateral to the root along the lateral edge of the
bony canal to avoid a potentially dangerous midline scar. When
performing lateral nerve root decompressions,

work parallel, rather than perpendicular, to the long axis of the nerve root in order to minimize risk of cutting across a root.

Although scarring is a common event following all
surgeries, it has been implicated as a potential cause for continued
pain following spinal surgery. Postoperative scar tissue may be located
either intradurally (arachnoiditis) or extradurally (epidural fibrosis). Arachnoiditis is an inflammation of the pia-arachnoid membrane that surrounds the cauda equina or spinal cord (17).
It can result in surgical failure and continued pain following
decompressive surgery. Its etiology is often unclear, but it has been
associated with many conditions, including oil-based myelographic
contrast agents and prior surgery (93). The
exact mechanism by which arachnoiditis occurs following surgery is not
completely understood, but it is thought to be more likely to occur
following dural laceration in which blood gains entry into the dural
sac and mixes with neural elements. It is also associated with
intraoperative trauma to neural structures. Arachnoiditis exists as a
spectrum of severity, from mild pia-arachnoid thickening to severe
scarring with complete blockage of the flow of contrast agents or
spinal fluid. Diagnosis can be made by water-soluble myelography, MRI,
or post-contrast CT in which the individual nerve roots of the cauda
equina appear clumped together rather than as well-defined structures.
Surgical treatment of arachnoiditis is not indicated because surgery
rarely produces any significant pain relief and may be complicated by
further damage to neural structures and more scarring (17).

Epidural fibrosis is extradural, rather than intradural,
scar tissue in which adhesive constrictions can form around neural
tissue. It commonly arises from contact with the paraspinal musculature
and is probably a relatively frequent event following spinal surgery.
Although such scar tissue can result in postoperative pain, symptoms
are relatively infrequent. When postoperative pain exists, the primary
differential diagnosis is between scar and recurrent disc herniation.
Radiographic distinction between these two conditions is best made with
gadolinium-enhanced MRI or post-contrast CT (97).

Efforts to prevent postoperative scar formation include
delicate surgical technique with adequate illumination and
magnification, meticulous hemostasis and drainage, and the use of some
form of an interposition membrane as a barrier to scar formation. These
barriers include a thin layer of fat or synthetic agents such as an
absorbable gelatin sponge. The use of a free fat graft has been
considered the gold standard interposition membrane, although use of
large grafts has been associated with postoperative cauda equina
syndrome (79).

Postoperative spine infections may be divided into either superficial or deep
infections. Although the treatment for both types of infections is
often similar (i.e., debridement and antibiotic therapy), it is useful
to make this distinction because duration of treatment (e.g.,
short-term antibiotics for superficial infections versus long-term
intravenous antibiotics for deep infections), morbidity, and long-term
outcome are often very different for the two types.

Superficial wound infections are located beneath the
dermis and subcutaneous tissue but superficial to the deep
thoracolumbar fascia and are characterized by tenderness and localized
erythema. They usually have associated drainage and fluctuance,
although in milder cases consisting only of cellulitis these may be
absent. Patients may be febrile but usually show no other systemic
signs of illness (67). Laboratory data usually
show elevation of the erythrocyte sedimentation rate (ESR) and
C-reactive protein (CRP), although the white blood cell count (WBC) is
usually within normal limits.

Treatment of superficial wound infections consists of
local wound care, ranging from simple packing of a small area of
localized infection, followed by a short course of oral antibiotics, to
more aggressive surgical debridement of necrotic tissue with short-term
parenteral antibiotics. In cases requiring surgery, the wound can
usually be closed primarily, although delayed closure is an option if
there is any question about the adequacy of the debridement. The use of
short-term suction-irrigation tubes is at the discretion of the
surgeon, although they are usually not required.

As opposed to superficial infections, in which the
diagnosis is usually readily apparent, deep infections may be difficult
to diagnose and a high index of suspicion is often required (67).
Because delay in diagnosis is common, the amount of tissue necrosis is
often extensive. Symptoms include disproportionate back pain or leg
pain. This may follow a relatively painless and uneventful immediate
postoperative period. The patient may feel and look ill and may exhibit
generalized malaise. Fever is often present but may be deceptively low
grade. If an epidural abscess is present, radicular leg pain and
neurologic deficit may occur. Although the patient may exhibit a
leukocytosis, elevation of the WBC count is frequently absent. The ESR
and CRP are usually elevated.

Radiographic imaging studies are usually required to
confirm the diagnosis. MRI provides the best and most useful
information by revealing both the presence and extent of a deep
abscess. Typically, an abscess is demonstrated by the presence of a
well-demarcated area of increased signal intensity on the T2-weighted
image. When

MRI
is not available, diagnosis may be confirmed radiographically by the
presence of a circumscribed area of fluid density visualized by CT. If
a deep abscess is strongly suspected, diagnosis may be confirmed by
aspiration, with subsequent culture and sensitivity of any fluid
obtained (61).

Treat deep wound infection aggressively with surgical
debridement of all necrotic tissues, followed by appropriate parenteral
antibiotics. Begin surgical exposure of the affected area with careful
sequential debridement, and irrigate each layer before proceeding to
the next deeper layer to avoid inadvertent contamination of potentially
unaffected deeper tissues. If the infection extends deeply into the
laminectomy site, take care to remove any fat graft or absorbable
gelatin sponge material. Following removal of infected or suspicious
tissues, thoroughly irrigate the wound with pulsatile lavage. Do not
remove rigid fixation and bone graft from an instrumented spinal fusion
because this may increase the risk of subsequent pseudarthrosis. Loose
hardware, on the other hand, no longer performs its function of
providing stability to the spine and, therefore, should be removed and
thoroughly debrided. Place a drain and close the wound meticulously.
Tightly close the deep fascia with interrupted absorbable suture
oversewn with a continuous running stitch. Close the wound primarily,
particularly in the presence of spinal fixation hardware, unless there
is infection from a particularly virulent organism. This may require
extensive undermining of wound margins in order to avoid tension on
friable wound edges. It is usually advisable to close the wound using
large throws of a sturdy, nonabsorbable suture rather than staples. Use
of suction-irrigation tubes for a few days may be considered, although
this is usually not necessary.

Epidural abscess, because of its risk of paresis or
frank paralysis, is one of the most feared complications of spinal
surgery. Fortunately, it is a rare occurrence following spinal surgery,
with only 16% of epidural abscesses resulting from postoperative
infection (4). Signs and symptoms are obvious and constitute a typical presentation (4).
Patients nearly always have significant back pain, and often present
with obvious neurologic findings such as nuchal rigidity and weakness
or paralysis of the lower extremities. The patient appears to be much
sicker than with either postoperative discitis or vertebral
osteomyelitis and typically has a fever. Both the WBC and acute phase
reactants are elevated. MRI is the diagnostic imaging modality of
choice and clearly visualizes the abscess as a discreet,
well-circumscribed entity within the subarachnoid space. It clearly
delineates the upper and lower extent of the abscess and, therefore, is
invaluable in the preoperative planning of the extent of decompression.

Treatment of an epidural abscess must be prompt and
decisive: surgical evacuation of the abscess and any adjacent necrotic
tissue, followed by parenteral antibiotics. The preferred surgical
approach is generally posterior, although an anterior approach may be
indicated in the presence of a significant kyphotic deformity in which
bony collapse has compromised the neural structures and simultaneous
anterior reconstruction and bone grafting are required to restore
stability.

Epidural hematoma causing symptomatic neurologic
compression is another devastating complication of spinal surgery.
Fortunately, the risk of this complication can be minimized by
meticulous attention to preoperative, intraoperative, and postoperative
detail. Preoperatively, advise the patient
to stop all nonsteroidal anti-inflammatory drugs (NSAIDS) for
approximately 1 week before surgery. In addition, it is important that
the patient is not hypercoagulable. When indicated, check the
prothrombin time (PT), partial thromboplastin time (PTT), bleeding
time, platelet count, and platelet function. Intraoperatively,
position the patient with the abdomen hanging freely in order to
minimize epidural venous congestion. Keep the blood pressure below 100
mm Hg systolic, if possible, in order to minimize bleeding. Use
electrocautery, and seal raw bone surfaces with bone wax to minimize
bleeding during the surgical exposure. Control epidural bleeding with
bipolar electrocoagulation. At the end of the surgery, when the deep
paraspinal muscle retractors are removed, check the muscle walls for
persistent bleeding, because prolonged muscle retraction may
temporarily occlude potentially significant muscle bleeders that could
begin bleeding after muscle layer closure. In general, I prefer to use
a drain postoperatively in order to minimize the formation of
postoperative hematoma. Postoperatively,
leave the drain in place for 24 to 48 hours or until the amount of
collected blood is less than approximately 30 ml per 8-hour shift. I do
not prescribe NSAIDS during the immediate postoperative period (during
the first 48 hours) in order to minimize bleeding from the fresh wound.

A characteristic clinical feature of epidural hematoma
is the presence of severe pain that appears out of proportion to what
is normally expected. This is usually associated with a progressive
neurologic deficit. Depending on the extent and location of the
surgical exposure and the magnitude of the hematoma, the neurologic
deficit may be focal and unilateral, or it may be widespread and may
involve multiple muscle groups in both legs.

The key to diagnosis of this condition is having a high
index of suspicion. Confirm the diagnosis with MRI, myelography, or CT.
Once the diagnosis is suspected, immediately return the patient to the
operating room for decompression and drainage of the hematoma.

Postoperative neurologic compression may be due to
structures other than epidural hematoma. The use of free fat grafts as
a barrier to scar formation has been associated with symptomatic
neurologic compression mimicking epidural hematoma (79).
Although this risk can be minimized with the use of a smaller (3 to 5
mm thick) piece of fat, the fear of epidural compression by fat graft
has led some surgeons to abandon fat grafts in favor of other synthetic
scar barrier substances. Even these substitutes, however, may cause
neural compression if proper care is not exercised. For multilevel
laminectomy requiring the use of a lengthy piece of scar barrier
material, I prefer an absorbable gelatin sponge material rather than
fat because the fat could theoretically become balled up and exert
focal compression on the underlying dura, and result cauda equina
syndrome.

Vascular injuries associated with posterior lumbar
spinal procedures are nearly always associated with surgical
discectomy, rather than laminectomy. Vascular injury occurs most
commonly at L4–L5, followed by L5–S1 (24).
Although to some extent this reflects the most common levels of spinal
surgery, regional differences in vascular anatomy of the lower lumbar
spine also play a role. Injury of a major abdominal vessel typically
occurs from aggressive use of the pituitary rongeur, with penetration
through the anterior annulus.

Such injuries may be recognized early or late. Brisk
bleeding from acute laceration of a major abdominal vessel presents
early as hypotension and abdominal distention and is associated with a
high mortality rate. The mortality rate from arterial injuries has been
reported to be 78%, whereas that for venous injuries is 89% (24).
Vascular injuries may be recognized late by the development of
high-output cardiac failure or abdominal bruits from formation of an
arteriovenous fistula. Arteriovenous fistula formation is the most
common result of a vascular injury. It occurs most commonly between the
right common iliac artery and vein (29.1%), between the left common
iliac artery and vein (25.5%), the right common iliac artery and the
IVC (21.8%), and the right common iliac artery and left common iliac
vein (12.7%). Late arteriovenous fistula formation is more compatible
with long-term survival, with mortality reported between 9% and 11% (24).

Instability following surgical decompression is usually
an iatrogenic complication of spinal surgery. Such instability can
occur in either the anteroposterior plane (spondylolisthesis), in the
mediolateral plane (lateral listhesis and scoliosis), or in both planes
simultaneously. In general, the risk of postoperative anteroposterior
instability can be minimized by maintaining the integrity of at least
one facet joint at the level decompressed (1).
In other words, if a unilateral complete facetectomy is performed on
one side, then the integrity of the opposite facet must be maintained.
Similarly, if half of one facet is removed during a surgical
decompression, then at least half of the contralateral facet joint
should be spared. If a total of more than one facet is removed during a
decompression, consider prophylactic fusion of that level.

When decompressing a stenotic level associated with a
degenerative spondylolisthesis, concomitant fusion should generally be
performed because surgical outcome has been shown to be better with
fusion than with decompression alone (44). This
seems to occur even in the absence of solid bony arthrodesis,
suggesting that even the presence of a stable pseudarthrosis results in
better clinical outcome than when no fusion is attempted. This is
thought to be due to a reduced risk of a subsequent increase in the
slip, although a direct relationship between increase in magnitude of a
subsequent slip and poorer surgical outcome has not been demonstrated.
Although the use of segmental spinal fixation with pedicle screws has
been shown to increase the rate of fusion compared with posterolateral
fusion without instrumentation, there is no convincing evidence that
such instrumentation leads to better clinical outcome (32).

Isthmic spondylolisthesis with frank fracture of the
pars interarticularis may also occur in the absence of prior slip as a
result of surgical decompression. In such cases, the patient presents
with evidence of a de novo
spondylolisthesis occurring either at one of the levels decompressed or
at a level above the level of decompression. The presumed mechanism is
either a mechanical stress fracture or perhaps an impairment of the
blood supply to the affected level (16,106). Instability may also occur as a result of fracture of the facet joint following decompressive surgery.