Subscribe to eTOC


The number of spinal fusions performed in the US has increased exponentially over the past decade, while the incidence of diseases for which spinal fusion is performed has not had a parallel increase. The rapid increase has coincided with the availability of new spinal-fixation devices, that is, “hardware” such as fusion cages and pedicle screws. About half of all spinal fusions are done for degenerative disk conditions associated with low back pain, which is the focus of this Special Report.

Concerns have been raised that spinal fusions are being performed inappropriately in patients with low back pain because the indications are unclear, the tests for decision-making are at best imperfect, and there is little evidence that spinal fusion improves outcomes. These concerns apply to both instrumented – using hardware – and non-instrumented spinal fusion. Neurology Today spoke with several experts, including outcomes researchers, neurologists, neurosurgeons, and orthopedic surgeons, who agreed that too many fusions are being done for low back pain.


Experts said that it is difficult, if not impossible, to tease out data on the numbers of spinal fusions performed for low back pain from use of spinal fusion for other indications.

The number of spinal fusions performed annually in the US is estimated to be more than 250,000, and this number is expected to increase steadily. As Richard Deyo, MD, Professor of Medicine and Health Services at the University of Washington in Seattle, pointed out in an editorial (N Engl J Med 2004; 350:722–766), the Agency for Healthcare Research and Quality estimates that the annual number of spinal fusion surgeries performed in the US increased by 77 percent between 1996 and 2001; by contrast, two common surgeries – hip replacement and knee arthroplasty – increased by about 13 percent during the same period. About 75 percent of spinal fusions are performed for spondylosis, disk disorders, and spinal stenosis exclusive of deformities.

These procedures are costly. The average hospital bill for spinal fusion is more than $34,000, excluding professional fees. The total costs per procedure can be higher, depending on the implants and devices used in instrumented spinal fusions. The market for implants and devices is estimated to be about $2 billion per year, with an annual growth rate of 18 to 20 percent (Orthop Network News 2002;13(4):7–8).


Financial incentives are one motivation for increased use of spinal fusion. Reimbursement for spinal procedures, especially instrumented procedures, is more favorable than reimbursement for most other procedures performed by orthopedic surgeons and neurosurgeons, according to Dr. Deyo and John Loeser, MD, a neurosurgeon with the Department of Neurological Surgery at the University of Washington in Seattle.

Dr. Loeser and others also said that training during surgical residency influences surgeons' attitudes toward spinal procedures, and procedures like spinal fusion may be advocated in the absence of good evidence to support their use.


These figures illustrate the various points for insertion of pedicle screws during spinal fusion: the starting point for the pedicle screw (figure 1), shown in a cross-section of line A bisection of transverse processes and line B lateral border of the pars interarticularis; a probe is at the pedicle entrance (figure 2) and angled at the correct superior/inferior and medial/lateral orientation to create a pedicle track; and the pedical screw is inserted medially and inferiorly close to the anterior cortex (figure 3) without penetrating it.

Dr. Loeser pointed out that there is a lack of evidence to show that uninstrumented spinal fusion (without hardware) improves outcomes in people with low back pain, and no evidence has demonstrated that using hardware is more effective than uninstrumented fusion in terms of functional outcomes.

“There is no question that hardware increases the success of the fusion itself, and that is a big step forward. This has led to an increase in the number of fusions being done based on the theoretical concept that hardware improves outcomes,” Dr. Loeser said.

Although there is a lack of data to support improved outcomes with spinal fusion in patients with low back pain, there is evidence to suggest that fusion is associated with increased morbidity. An older study demonstrated that spinal fusion in a Medicare population doubled the risk of complications and the risk of transfusions without improving outcome (Spine 1993;18(11):1463–1470). Dr. Loeser was a coauthor of that paper.


Dr. John Loeser said there is a lack of evidence to show that uninstrumented spinal fusion (without hardware) improves outcomes in people with low back pain, and no evidence has demonstrated that using hardware is more effective than uninstrumented fusion in terms of functional outcomes.


The lack of established indications for spinal fusion was cited by interviewees as a major factor responsible for the dramatic rise in the number of procedures performed each year for patients with low back pain. There was consensus that the main clear indication for spinal fusion for low back pain is measurable spinal instability associated with neurological symptoms. Examples include spinal stenosis and degenerative spondylolisthesis; degenerative scoliosis; repeated surgeries at the same level; and failed fusions, said William Abdu, MD, Medical Director of the Spine Center at Dartmouth Hitchcock Medical Center, and Associate Professor of Orthopedic Surgery at Dartmouth Medical School in Hanover, NH.

“The indications for spinal fusion are less clear when there is back pain alone in the absence of neurological symptoms or radiographic instability,” Dr. Abdu said. “The situation is even more complex because the term ‘instability’ is confusing to everyone. Instability is not one of those easily defined problems,” said Dr. Abdu. Instability can be associated with trauma, but then instability can also be applied to the degenerative process where it is less well defined.

Dr. Abdu said instability is better defined in the traumatic setting, but it is still unclear when the surgeon is attempting to predict hypermobility and biomechanical integrity under physiologic stresses and loads based on 2-dimensional black and white images from x-rays, MRIs, and CT scans taken when the patient is supine and effects of muscle tone and gravity are eliminated.

“Spinal fusion should be performed only when there is instability of the spine. But most people who undergo this procedure for low back pain do not have instability,” said Gary Franklin, MD, Medical Director of the Washington State Department of Labor and Industries in Seattle. Dr. Franklin, a neurologist who reviews disability claims data for the state of Washington, explained that the clear indications are an unstable spine due to spinal fracture or measurable grade 3 or 4 spondylolisthesis.

Michael Wang, MD, Associate Professor of Neurosurgery at University of Southern California-Los Angeles, had a different opinion: “We operate for low back pain on two major categories of patients: one group is patients with microscopic instability of the spine – spondylolisthesis or malalignment of the vertebrae. If the procedure is done correctly, patients with this structural problem almost universally derive benefit. The second group is more ambiguous and controversial. This group of patients has discogenic back pain without gross evidence of structural spinal instability. We believe this is a phenomenon called ‘micro-instability’ where the disk may be incompetent to handle body weight, as in sitting or standing.”


Dr. William Abdu: “The indications for spinal fusion are less clear when there is back pain alone in the absence of neurological symptoms or radiographic instability.”


In the absence of MRI-confirmed spondylolisthesis and accompanying neurological symptoms, discography may be used to assess the competence of the disk, and some spinal fusions are based on this test.

Discography is the most often used test to confirm the need for fusion in patients with low back pain, and it is controversial. This is a provocative test aimed at reproducing the symptoms by injecting the disk with radio-opaque dye. If the symptoms are reproduced, then the disk is assumed to cause the pain. Once the disk is injected with dye, it should contain fluid “like an inflatable raft,” said Dr. Wang. Pressure measurements can be obtained and, if the measurements are low, then the disk is presumed to be incompetent, which can be supportive of the need for fusion. CT can determine where the dye is distributed; if the dye extrudes beyond its normal boundaries, the disk is also presumed to be incompetent. Anesthetic can be injected into the disk prior to discography; “if the pain disappears after discography, then that tells you that the disk is the source of pain,” Dr. Wang said.

“Discography is an imperfect test and it is controversial because it is imperfect; some studies show that it works and others show that it doesn't work,” Dr. Abdu said.

Dr. Franklin believes that discography is unreliable because it is based on subjective response and the test does not predict outcome. “The procedure itself is painful, and if the patient reports concordant pain at a given level of the spine, then the surgeon decides if a fusion is needed. This is an unreliable test, because the results do not predict outcome,” Dr. Franklin said. “The spinal study by Carragee et al. has corroborated the uselessness of this test,” he added (Curr Rev Pain 2000;4:301–308).

“Spinal fusion at the lumbar level in a patient with no measurable disability, based on the theory that a disk or several disks are causing the pain and based on an unreliable subjective response to a test, is unproven use of spinal fusion,” Dr. Franklin said.

Dr. Wang, on the other hand, believes that discography is a useful test because it helps reduce the possibility that we are operating on the wrong patient. “If the patient seems to have normal disks and tests positive on discography – meaning that the disks are normal on discography – this may be a sign that there is something in their physiological makeup – that is, their perception of pain – that will prevent them from getting better, even if you fix the anatomical problem.”

Other unproven methods to warrant fusion include flexion and extension views of the suspected disc, attention to MRI appearance of vertebrae, and isotope bone scans that might show focal uptake of radionuclide.


Success rates for spinal fusion for low back pain – that is, greatly improved pain and ability to return to normal activities – in the published literature are about 60 to 65 percent. Dr. Wang said that in his hands, the success rate is 80 percent. He emphasized that even if 60 percent of patients are improved, that should be considered a good outcome, given that most of people who undergo spinal fusion are totally debilitated and have tried everything else first – including acupuncture, chiropractic, hypnotherapy, and drugs.

“These people are desperate. By the time we see them, they are on narcotics, disability, and are often unemployed. They are often not going to get better any other way,” Dr. Wang stated.


Dr. Gary Franklin: “Spinal fusion should be performed only when there is instability of the spine. But most people who undergo this procedure for low back pain do not have instability.”

Dr. Franklin said that his own outcomes study showed that two-thirds of workers' compensation patients who were operated on for low back pain were still totally disabled two years after surgery (Spine 1994;19:1897–1904). He noted that a 2003 well-designed, randomized trial showed no clear benefit of lumbar fusion for discogenic low back pain (Spine 2003;28:1913–1921).

“Surgeons don't like to hear this,” Dr. Franklin added. “But this is a great example of a long-held unproven theory – that discogenic back pain will improve from immobilizing the adjacent spinal segments – in search of an effective treatment.”

Not only are the term success rates suboptimal, but also the procedure itself is associated with substantial morbidity and re-operations. According to Dr. Deyo's editorial, complications of spinal fusion include instrument failure (7 percent), complications at the bone-donor site like infection and chronic pain (11 percent), neural injuries (3 percent), pulmonary embolus (2 percent), and infections (3 percent). Failure to achieve a solid fusion mass (pseudoarthrosis) occurs in 15 percent of cases. Some studies suggest that the rates of re-operation after spinal fusion may be higher than the rates of re-operation after laminectomy or diskectomy, he said. Rates of re-operation are even higher for fusion with hardware compared with bony fusion alone, Dr. Deyo added.


Three randomized, controlled, European trials suggest that physical rehabilitation may be a reasonable alternative to fusion effective in patients with low back pain. Drs. Deyo, Loeser, and Franklin believe that results of these trials can be extrapolated to the US, but Dr. Loeser said that physicians in our country tend to discount studies from Europe.

A trial in Norway demonstrated similar outcomes with lumbar instrumented fusion versus cognitive behavioral intervention plus exercise in 63 patients with chronic low back pain and disc degeneration (Spine 2003;28:1913–1921). The cognitive intervention consisted of educating the patient to understand that normal physical activity would not be harmful as well as a recommendation to encourage movement and use of the spine; that is, to move, twist, and flex as normally as possible during the course of daily activities. This was reinforced by three daily physical exercise sessions for three weeks.

The Oswestry Disability Index was used as an outcome measure in the Norwegian study. This index, a validated measure, includes 10 questions related to disability, each with six possible answers; the sum of the answers is calculated and given as a percentage, with 0 percent representing no pain and 100 percent the most severe pain and disability. At one year, this index was significantly reduced in both groups: from 41 to 26 in the fusion group and from 42 to 30 in the group treated with cognitive intervention and exercises. Both groups had similar improvements in back pain, use of analgesics, emotional distress, life satisfaction, and return to work. In the surgery group, early complication rate was 18 percent. An independent observer rated success as 70 percent after surgery and 76 percent after cognitive intervention and exercises.


Dr. Richard Deyo pointed out that the annual number of spinal fusion surgeries performed in the US increased by 77 percent between 1996 and 2001; by contrast, two common surgeries – hip replacement and knee arthroplasty – increased by about 13 percent during the same period.

A randomized, controlled, multicenter study in Sweden showed more favorable results with lumbar fusion versus nonsurgical treatment with different types of physical therapy (Spine 2001;26:2521–2532). That study included 294 patients with chronic low back pain – 222 patients in the surgery group and 72 in the physical therapy group. Baseline demographic and pain characteristics were similar in both groups. Patients had met the criteria for disability insurance for a mean of 3.2 years in the surgery group and 2.9 years in the physical therapy group.

At two-year follow-up, back pain was reduced by 33 percent in the surgery group (64 to 43 percent) compared with 7 percent in the no-surgery group. Pain improved in the first six months and then gradually deteriorated. Disability, according to the Oswestry Index, was reduced by 25 percent in the surgery group compared with 6 percent in the nonsurgery group. Patient self-ratings of “better” or “much better” were reported by 63 percent of the surgery group and 29 percent of those treated with physical therapy. In the surgical group, the early complication rate was 17 percent.

Although that study supported the use of lumbar fusion in selected patients with low back pain, it has been criticized for lack of a standardized rehabilitation therapy that included a cognitive component for the nonsurgical group. Some experts believe that the cognitive component is critical. Dr. Deyo pointed out that only 63 percent of the surgical patients considered themselves improved, and the magnitude of improvement with fusion was only about 30 percent for pain and functional improvement; only one in six patients became pain-free. Also, improvements in the surgical group waned over time, suggesting that benefits were temporary.

The third study, which compared spinal fusion to rehabilitation with a cognitive component, was presented at two European meetings in 2004. The results have not yet been published in a peer-reviewed journal but were summarized in The Back Letter (2004;19:73). Jeremy Fairbank, MD, and colleagues in Great Britain randomly allocated 349 candidates for spinal fusion to immediate surgery or an intensive three-week rehabilitation program based on cognitive principles. The rehabilitation program had a cognitive component – to teach patients to identify thoughts and beliefs that underpin maladaptive behavior and adversely affect mood, and to change unhelpful ways of functioning and avoiding feared activity – as well as five days a week of exercise therapy and spine stabilization techniques.

All patients experienced more than 12 months of disabling, chronic back pain. Eleven percent had spondylolisthesis, approximately 8 percent had continuing pain after laminectomy, and about 81 percent had chronic back pain.

Patients were assessed at 6, 12, and 24 months. The surgical group had a marginally better outcome on the Oswestry Disability Index. On all other measures, including the shuttle-walking test – a demanding measure of walking ability – the SF-36, and the EuroQuol quality of life instrument, there were no differences between groups.

Dr. Fairbank said that intensive rehabilitation with a cognitive component provides a cheaper and noninvasive alternative to spinal fusion. Fusion surgery was about twice as expensive as physical therapy over that time frame. He also suggested that that if nonoperative treatment is to be successful, it should include a cognitive component and support of intensive rehabilitation efforts by the treating surgeons.


“Patient selection is absolutely critical to the success of spinal fusion for low back pain,” Dr. Wang said. Even if patients have the anatomical defects that constitute a clear indication for spinal fusion for low back pain, clear contraindications exist. These include psychological problems (as shown on the Minnesota Multiphasic Personality Inventory), family history, medical history, and physician's index of suspicion, morbid obesity, concurrent metabolic or systemic illness (for example, connective tissue disease), smoking, and advanced age or very young age.

Experts agreed that spinal fusion is increasing rapidly and that the indications for spinal fusion for low back pain remain unclear, with the exception of spondylolisthesis and accompanying symptoms.

More research is needed on low back pain and spinal fusion, said Dr Abdu. “It is not at all clear where the pain generators are located, and we need to define which patients will benefit from fusion interventions,” he said.

It will take a multidisciplinary team with infrastructure and financing to conduct the necessary randomized controlled trials. “At Dartmouth, we tend to rely on outcomes data and shared decision-making for many conditions, but the problem is there are not a lot of data available from prospective, randomized clinical trials – the ‘gold standard’ in clinical study design – on which surgeons and patients can base clinical decisions and treatment choices,” Dr. Abdu noted.

Dr. Deyo called for a shift in emphasis of research from looking at how to perform fusion to examining who should undergo fusion. “Only with more and better clinical trials will the indications and optimal technique for spinal fusion become clear,” Dr. Deyo said.


Dr. Deyo: Patients with low back pain should be encouraged to participate in a structured rehabilitation with a cognitive component as a reasonable alternative.

Dr. Wang: Neurologists should be aware that it is not clear whether fusion should be done for degenerative disks and spinal stenosis.

Dr. Wang: Patients with spondylolisthesis accompanied by back pain should be referred to a surgeon. However, the surgeon should be experienced in these procedures and use strict criteria for performing the procedure.

Dr. Deyo and Dr. Franklin: Several studies have shown that exercise benefits patients with low back pain. Neurologists should be educated about the need to reactivate their patients with chronic low back pain; that is, to get them to go back to work and to exercise. “More than two days of rest is not advisable for low back pain. Rest is helpful for sciatica, but not for low back pain,” Dr. Franklin said.

Dr. Franklin: Neurologists should be cautious about prescribing increasing doses of opiates in patients with chronic back pain. Doses should be low enough so that patients can return to work.


• Wang M. Managing spinal disorders: role of interbody fusion technologies. Medscape CME activity. 2004.
    • Deyo R. Spinal fusion surgery – The case for restraint. N Engl J Med 2004;350:722–766.
      • Weiner BK, Levi BH. The profit motive and spine surgery. Spine 2004;29:2588–259.
        • No advantage for spinal fusion over aggressive rehabilitation in large randomized trial. The Back Letter 2004;19(17):73–80.
          • Brox I, et al. Randomized clinical trial of lumbar instrumented fusion and cognitive intervention and exercises in patients with chronic low back pain and disc degeneration. Spine 2003;28:1913–1921.
            • Mendenhall Associates, Inc. Spinal industry update. Orthop Network News 2002;13(4):7–8.
              • Fritzell P, et al. Treatment for chronic low back pain: A multicenter randomized controlled trial from the Swedish Lumbar Spine Study Group. Spine 2001;26:2521–2532.
                • Carragee E. Is lumbar discography a determinate of discogenic low back pain: provocative discography reconsidered. Curr Rev Pain 2000;4:301–308.
                  • Franklin GM, et al. Outcome of lumbar fusion in Washington State workers' compensation. Spine 1994;19:1897–1904.
                    • Deyo R, et al. Lumbar spinal fusion: a cohort study of complications, reoperations, and resource use in the Medicare population. Spine 1993;z18(11):1463–1470.