NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Bigos S, Bowyer O, Braen G. Acute Low Back Problems in Adults. Rockville (MD): Agency for Health Care Policy and Research (AHCPR); 1994 Dec. (AHCPR Clinical Practice Guidelines, No. 14.)
This publication is provided for historical reference only and the information may be out of date.
After the first month of symptoms, the vast majority of patients have recovered from activity limitations due to low back problems. However, if the patient is limited by back symptoms for more than 1 month, special diagnostic and treatment procedures are often considered to find a reason for the slow recovery (Attachment A3).
The special studies are of two kinds. First are tests to provide evidence of physiologic dysfunction such as neurologic dysfunction, infection, inflammation, malignancy, or other systemic illness. Second are tests to define a potential anatomic reason for the dysfunction such as a herniated lumbar disc, spinal stenosis, infection, tumor, or abdominal mass. As a result of diagnostic considerations, therapeutic interventions including surgery may be recommended.
Except when serious underlying pathology is suspected, special diagnostic tests are usually not needed during the first month because it is not possible to predict early on which patients will and will not improve during this period. But those who are not improving at 1 month may include some who could benefit from specific therapeutic interventions. Waiting longer to start the diagnostic workup may delay recovery for these individuals.
Special Studies: Tests for Evidence of Physiologic Dysfunction
Tests commonly used to identify focal neurologic physiologic dysfunction include electromyography (EMG), sensory evoked potentials (SEPs), and thermography. To detect physiologic dysfunction of nonneurologic diseases, general laboratory screening tests, such as erythrocyte sedimentation rate (ESR), complete blood count (CBC), and urinalysis (UA), and bone scan have been proposed.
Electrophysiologic Tests (EMG and SEP)
Panel findings and recommendations:
Electrophysiologic tests are sometimes used in patients with sciatica to evaluate physiologic functioning of the spinal cord, nerve roots, and peripheral nerves. Overall diagnostic objectives of these tests are to assess suspected myelopathy (dysfunction of the spinal cord), radiculopathy (dysfunction of a spinal nerve root), neuropathy (dysfunction of a peripheral nerve distal to the nerve root), and myopathy (muscle abnormalities). The tests and their specific diagnostic objectives for low back problems are as follows:
- Needle electromyography (EMG), used to assess acute and chronic nerve root dysfunction, myelopathy, and myopathy.
- H-reflex, a test measuring sensory conduction through nerve roots, used mostly to assess S- [1] radiculopathies.
- F-wave response, a test measuring motor conduction through nerve roots, used to assess proximal neuropathies.
- Surface EMG, used to assess acute and chronic recruitment patterns during static or dynamic tasks using surface electrodes instead of needle insertion.
- SEPs, used to assess sensory neurons in peripheral and spinal cord pathways.
- Nerve conduction studies, used to assess acute and chronic peripheral entrapment neuropathies that may mimic radiculopathies.
Literature Reviewed
Of 52 articles screened for this topic, 8 met review criteria for adequate evidence about efficacy. [21], [254] - [260] Also reviewed was one study that evaluated findings in asymptomatic subjects who had no history of low back problems. [261] Other studies contained information used by the panel. [49], [50], [147]
Four studies evaluated needle EMG. [21], [254], [257], [260] Three studies evaluated H-reflex or F-wave tests. [21], [254], [256] Two studies evaluated surface EMG. [258], [261] One study evaluated SEPs. [259] No studies were found evaluating nerve conduction for assessing low back problems. All articles meeting review criteria involved groups of patients either with chronic problems or with unreported symptom duration.
Evidence on Efficacy
Reference tests used to determine diagnostic accuracy (true positive and true negative rates) of needle EMG included surgical findings. [21], [254], [260] They also included clinical followup (pain rating and work status), with posttest at 1 year. [257] The amount and quality of data that could be used to calculate diagnostic accuracy of these tests varied significantly between studies. Many studies had major methodological flaws such as biased test interpretations, faulty cohort assembly, and poor clinical descriptions. Determining the accuracy of EMG is difficult as well because anatomic defects seen at surgery, which are used in many of these studies as the reference standard for EMG, may or may not be the cause of symptoms. This makes true positive and true negative rate determinations of questionable value.
For needle EMG, four articles evaluated the test's accuracy in diagnosing nerve root problems. Aiello, Serra, Migliore, et al. [21] evaluated 25 patients who had clinical evidence of L3 and/or L4 nerve root impairment. Needle EMG was abnormal in all 24 patients who had positive findings at surgery, but predicted the exact levels of nerve root pathology in only 9 percent of these 24 patients. One patient had a false-positive EMG with no disc herniation found at surgery.
Aiello, Serra, Tugnoli, et al. [254] evaluated the accuracy of EMG for detecting and localizing nerve root compromise in patients who had surgical findings of a single lumbar disc prolapse at the L3-L4 level (100-percent true positive rate, 88-percent true negative rate), with disc herniation at L4-L5 (96-percent true positive rate, 38-percent true negative rate) and with disc herniation L5-S1 (71-percent true positive rate, 79-percent true negative rate).
Khatri, Baruah, and McQuillen [257] evaluated outcomes at 1 year in patients with radicular leg pain who had needle EMGs and computerized tomography (CT) scans and then went on to have disc surgery or nonsurgical treatment (including conservative care and/or epidural steroids). There were 35 patients who had abnormal EMGs and CT scan findings of a herniated lumbar disc. Of the 16 patients treated with surgery, 81 percent were better at 1 year, while of the 19 who had nonsurgical treatments only 47 percent reported improvement at one year. There were 24 patients whose EMGs and CT scans were both normal; none of these patients had surgery, and at 1-year followup 67 percent were improved.
Young, Getty, Jackson, et al. [260] evaluated 100 patients with clinical evidence of L5 or S1 radiculopathy and found that needle EMG correctly predicted the level of nerve root pathology in 84 percent of the 95 patients with positive surgical findings. The wrong level was predicted in seven patients, and in nine patients only one abnormal root was detected when two were involved. EMGs were negative in all five patients who had no root pathology noted at surgery.
In cases of radiculopathy or neuropathy, EMG results may be unreliable in limb muscles until a patient has had significant leg symptoms for over 3 weeks. Also, abnormal EMGs tend to normalize over time. There is some evidence that greater accuracy can be obtained in diagnosing lumbar nerve root compromise when information from needle EMG is combined in a systematic fashion with information from imaging tests and clinical findings, rather than relying on the results of each test alone. [49], [50]
For H-reflex and F-wave tests, three articles reported on accuracy in diagnosing nerve root problems. Aiello, Serra, Migliore, et al. [21] evaluated 25 patients with clinical evidence of L3 and/or L4 nerve root impairment. H-reflex tests were abnormal on the affected side in 96 percent of the 24 patients with L3 and/or L4 nerve root compromise confirmed at surgery, but the test was also positive in the 1 patient with no nerve root pathology noted at surgery (96-percent true positive rate, 0-percent true negative rate).
Aiello, Serra, Tugnoli, et al. [254] evaluated H-reflex tests and EMGs for 50 patients in whom a single disc prolapse was found at surgery. H-reflex tests were positive in 71 percent of 7 patients with L3-L4 disc herniations, 58 percent of 26 patients with L4-L5 disc herniations, and 100 percent of 17 patients with L5-S1 disc herniations. In addition, by combining information from H-reflex tests and needle EMG, 30 percent of single disc prolapses could be determined accurately.
Braddom and Johnson [256] evaluated H-reflex tests in 25 patients with clinically suspected S1 radiculopathy. A normal range was established in 100 asymptomatic subjects. All 25 patients had H-reflex-test latencies greater than 2 standard deviations above the mean for the control group. Only three patients were reported as having surgery, and long-term followup results were not given.
For surface EMG, two articles evaluated efficacy in assessing patients with low back symptoms. Arena, Sherman, Bruno, et al. [255] evaluated surface EMG of lumbar paraspinal muscles as a function of pain state. Surface EMG results were not significantly able to differentiate back pain patients with high pain states from those in low pain states. Sihvonen, Partanen, Hanninen, et al. [258] evaluated both needle and surface EMG of lumbar paraspinal muscles in 87 patients with back pain (not radiculopathy) longer than 6 months, but who had not had surgery, and in 25 controls with no history of back problems. When compared to the asymptomatic controls, the patients with back pain had significantly increased surface EMG activity while standing and immediately after lumbar flexion.
For SEPs, the only article reviewed [259] evaluated accuracy in 18 patients with clinical findings of suspected spinal stenosis who went on to have surgery. At operation, all patients were found to have spinal stenosis. In all but 1 of the 18, SEPs had been positive (94-percent rate).
Potential Harms and Costs
Inserting small needle electrodes into muscle tissue is invasive, causes some patients discomfort, and may cause bruising. Patients who have severe pain, low pain thresholds, excessive anxiety, or conflicting emotional symptoms may not be able to participate fully in the examination. [147] EMG and other electrophysiologic tests are moderately expensive.
Summary of Findings
The evidence suggests that in patients with low back problems who have a confusing clinical picture of severe leg symptoms of more than 3 to 4 weeks' duration, EMG and H-reflex tests appear to be useful (1) to document presence or absence of radiculopathy or neuropathy as the cause of symptoms in the lower extremities, (2) to provide more information on specific nerve roots that may be compromised, and (3) to help differentiate between acute and chronic nerve root dysfunction. Optimal time for needle-EMG testing is after the patient has had lower limb symptoms at least 3 to 4 weeks. Test results are not reliable before this time.
SEPs appear to be useful in diagnosing spinal stenosis and spinal cord myelopathy. However, F-wave tests and surface EMGs are not considered effective methods of assessing acute low back problems.
Accuracy of electrophysiologic testing is highly dependent on the skill with which the examination is performed. Clinicians are urged to assess the qualifications of the diagnostician before referring a patient with suspected neurologic compromise.
Bone Scan
Panel findings and recommendations:
Bone scanning, a type of radionuclide imaging, involves intravenous injection of radioactive compounds known to adhere to metabolically active bone. Gamma detectors localize regions of uptake. Most of the bone-seeking compounds used contain the radionuclide technetium- [99]m. The diagnostic objective is to detect occult fractures, infections, and bony metastases of the spine and to differentiate them from common benign pathology such as degenerative changes.
Evidence on Efficacy
In the studies reviewed, bone scan was used to detect several different clinical conditions: stress fracture injuries of the pars interarticularis; [263] inflammatory sacroiliitis; [262], [264] spine infections; [266] metastatic cancer and other systemic disease; [265] and symptomatic spondylolysis. [263] All of these studies either evaluated mixed patient groups with both acute and chronic problems or did not report symptom duration. [264], [265] There were no studies focusing specifically on patients with acute low back pain.
The reference standard used depended on the clinical condition to be assessed. Included were: clinical diagnoses; [262]- [264] clinical followup or autopsy; [265] and biopsy with microbial cultures. [266]
Schütte and Park [265] evaluated bone scans for two groups. One group comprised 138 patients with a history of malignancy. In the other group, 38 patients with no previously recognized systemic disease had nonspecific low back pain and normal x-rays, but with strong clinical suspicion of an underlying serious condition causing the back pain. Of these 38 patients with "nonspecific" low back pain, 14 (37 percent) were later found to have a systemic disease. Of the 14 patients, 13 had elevated ESRs, but only 7 had positive bone scans (all 7 also with high ESRs). Of patients with a history of malignancy, 40 percent had positive bone scans determined on followup to be metastases and 14 percent had osteoporotic rib or vertebral fractures.
The authors concluded that bone scan has a high yield in patients with known malignancy. In patients where clinical suspicion of an underlying serious problem was high, but who had no known malignancy and normal x-rays, ESR detected 93 percent of those with systemic disease. Bone scan detected only 50 percent of these cases. The authors concluded that bone scan has a low yield in patients with longstanding low back problems and normal x-rays and lab tests.
Whalen, Brown, McLeod, et al. [266] evaluated patients with low back pain who had extensive diagnostic workups because of suspected spinal infections, and who then had open biopsy or percutaneous needle aspiration for culture and microbiologic diagnosis. Duration of symptoms before workup ranged from 2 weeks to 4 years (37 percent with symptoms for less than 3 months). One subgroup of 19 patients had all of these tests before spinal biopsy: technetium- [99]m (Tc- [99] m) bone scan, indium- [111] (In- [111] ) white blood cell (WBC) bone scan, plain lumbar x-rays, and ESR. Of the 19 patients, 16 had spinal infections confirmed on culture of biopsy material.
For the 16 patients, true positive rates were: 81 percent for Tc- [99]m bone scan, 19 percent for In- [111] WBC bone scan, 44 percent for plain x-rays, 82 percent for ESR > 20, and 56 percent for ESR > 50. False-negative rates for diagnosing spinal infection were: 3 of 3 (100 percent) for Tc- [99]m bone scan, 14 of 17 (82 percent) for In- [111] WBC bone scan, 8 of 11 (73 percent) for plain x-rays, 2 of 2 (100 percent) for ESR > 20, and 6 of 8 (75 percent) for ESR > 50. Half the infections were attributed to prior spinal surgery or instrumentation, with the rest attributed to hematogenous spread.
Miron, Khan, Wiesen, et al. [264] evaluated the accuracy of a quantitative bone scan technique, scintigraphy, for diagnosing sacroiliitis in various groups of patients. First, 90 subjects with no history of low back symptoms were evaluated with the test to establish age- and sex-specific normal values for a sacroiliac index. A positive test was defined as a sacroiliac index greater than two standard deviations above the mean for normals. In evaluating patients with low back pain, the test was positive in 50 percent of 18 patients who had signs and symptoms of active sacroiliitis. The test was positive in only 7 percent (1 patient) of 14 patients who had low back pain, but no clinical, x-ray, or CT-scan evidence of sacroiliitis. The authors concluded that this test has a low true positive rate (50 percent), but high true negative rate (93 percent) for diagnosing sacroiliitis.
Esdaile, Rosenthall, Terkeltaub, et al. [262] evaluated Tc- [99]m bone scan scintigraphy for diagnosing sacroiliitis. Normal values for sacroiliac joint-to-sacrum ratios on scintigraphy were established in 18 controls with no clinical evidence of inflammatory back pain. These ratios were found abnormally elevated in 66 percent of 12 patients considered to have "possible" ankylosing spondylitis and in 46 percent of 22 patients considered to have "definite" ankylosing spondylitis based on clinical and x-ray findings. At the time of the initial testing, all patients were taking very little or no anti-inflammatory medication. All were then treated with such medication. The S1 joint-to-sacrum ratios significantly decreased after treatment. The authors concluded that this technique is not a useful screening technique to detect early ankylosing spondylitis.
Potential Harms and Costs
The primary potential complications of bone scan involve exposure to ionizing radiation from the radionuclide injected. The total radiation dose to the patient is equivalent to a set of lumbar spine x-rays. This test is contraindicated during pregnancy. The radionuclide may be found in breast milk, and breast feeding must be discontinued for a brief interval after the test. Bone scans are moderately expensive.
Summary of Findings
The bone scan is a moderately sensitive test for detecting suspected tumor, infection, or occult fractures of the vertebrae in patients with low back pain, but not for specifying the diagnosis. A positive bone scan suggesting one of these conditions will usually need to be confirmed using other diagnostic tests or procedures. Bone scan has been shown to be moderately sensitive for detecting metastases to the spine in patients with a previously established diagnosis of cancer who present with acute low back problems. Bone scan also appears to be effective for detecting serious conditions, such as tumor, infection, or fracture, in patients where there is suspicion of these problems based on clinical findings. Bone scan appears to be more accurate than plain x-ray for detecting these conditions. No studies were found comparing the relative accuracy of bone scan versus CT scan or magnetic resonance imaging (MRI) scan for detecting the conditions.
Thermography
Panel findings and recommendations:
In patients with low back problems, thermography involves measuring small temperature differences between sides of the body and evaluating the patterns on infrared thermographic images of the back and lower extremities. Because thermography is noninvasive and involves no ionizing radiation, the test has been proposed as a physiologic test with clinical utility for documenting the presence or absence of radiculopathy (nerve root compression).
Literature Reviewed
Of 17 articles screened for this topic, only 1 met review criteria for adequate evidence about efficacy. [267] The panel also reviewed a meta-analysis on thermography. [268] Reviewed as well were four articles providing data on thermography in asymptomatic subjects without back problems. [269]- [272]
Evidence on Efficacy
The one study meeting review criteria [267] evaluated thermography in 107 patients who had sciatica of unspecified duration, 19 of whom went on to have low back surgery, and 28 asymptomatic subjects with no history of back problems and no evidence of disease affecting the lower extremities. All the asymptomatic subjects had temperature differences on thermography of less than 1.9 degreesC in the feet and less than 1.0 degreesC in other parts of the lower limbs. These values were used as the upper limits of normal for evaluating patients with sciatica. Of the 19 patients with sciatica who went on to have surgery, only 53 percent had results on preoperative thermography in agreement with surgical findings.
The most recent meta-analysis [268] reviewed 81 relevant citations and analyzed 28 studies of thermography used for diagnosing lumbar radiculopathy. Articles were graded excellent, good, fair, or poor based on the following criteria: technical quality of the reference test, uniform application of the reference test, independence of interpretations, clinical description, cohort assembly, and sample size. Only excellent and good studies were considered reliable sources of data on diagnostic accuracy. No study was graded excellent, 1 study was graded good, 3 were graded fair, and the remaining 24 were graded poor. Because of the methodological flaws in 27 studies, summary pooled statistics were not reported. The only study considered reliable [267] found no discriminant value for thermography in diagnosing lumbar radiculopathy. True positive and true negative rates were both 48 percent.
The four studies evaluating thermography in persons without back problems found abnormal thermography of the lower limbs in 7 to 81 percent of these asymptomatic subjects. Chafetz, Wexler, and Kaiser, [269] evaluating 15 asymptomatic subjects with no current back pain and no history of back surgery or disability from back pain, found that 40 percent had abnormal thermograms.
Harper, Low, Fealey, et al. [270] evaluated thermography in 37 asymptomatic subjects (carefully screened for no history of back pain, back surgery, or disease or injury affecting the lower extremities) as well as in 55 patients with clinically suspected radiculopathy. All thermograms were interpreted independently by five readers experienced in thermography who were blinded to all clinical data. The different readers interpreted thermograms as probably or definitely abnormal in 56 to 81 percent of the asymptomatic controls.
Perelman, Adler, and Humphreys [271] evaluated thermography in 16 asymptomatic subjects with no low back complaints and found 25 percent of these subjects to have abnormal thermograms. Tests were considered abnormal if the thermogram had an asymmetric visual pattern, but temperature differences were not measured. So, Aminoff, and Olney [272] evaluated thermography in 30 patients with symptoms of lumbar radiculopathy and in 27 asymptomatic controls. Thermogram readers were blinded to clinical data. Abnormal thermograms, defined as temperature differences between sides greater than three standard deviations from the mean for all asymptomatic subjects, were found in 7 percent of the asymptomatic subjects.
Potential Harms and Costs
No reports of significant risks from thermography were found. Thermography is considered moderately expensive.
Summary of Findings
The one study meeting review criteria found that thermography did not accurately predict either the presence or absence of lumbar nerve root compression found at surgery. In addition, several studies have shown thermography of the lower limbs as abnormal in a substantial proportion of asymptomatic subjects without back problems. Based on the available research evidence, thermography does not appear effective for diagnosing low back problems.
Special Studies: Tests to Provide Anatomic Definition
In addition to x-rays, the imaging studies most generally used to define a possible anatomic cause for evidence of physiologic abnormalities include plain myelography, MRI, CT, CT-myelography, discography, and CT-discography.
Abnormal findings on anatomic studies such as MRI, CT, myelography, and discography may be misleading, however, if they are not corroborated with evidence of physiologic abnormality from the medical history, physical examination, or physiologic tests. One problem with imaging studies is that in many patients, there is an inability to find any defects. Another problem is the lack of a "gold standard" in determining if an anatomic defect seen on imaging tests is actually the cause of symptoms. Anatomic abnormalities of the lumbar spine, such as degenerative changes and bulging or herniated discs, are found to increase with aging on x-rays and other imaging tests in subjects asymptomatic for low back problems. [273]- [275]
Several studies stress the importance of not relying too heavily on imaging studies alone for assessment when nerve root compromise is suspected. [30], [40], [49] , [50] The anatomic level of imaging study findings must correspond to the side and the level of concern physiologically detected through the history, physical examination, or other physiologic methods.
Plain X-Rays
Panel findings and recommendations:
X-ray, or radiography, is the oldest and most widely available modality for imaging the lumbar spine. The most commonly used x-ray views of the lumbar spine, the standard anteroposterior and lateral views, permit assessment of lumbar alignment, comparison of vertebral body and disc space size, assessment of bone density and architecture, and gross evaluation of soft tissue structures. Oblique views of the lumbar spine are used in the detection of unilateral or bilateral spondylolysis. Other special views include sacroiliac views to evaluate possible ankylosing spondylitis. The diagnostic objective of x-rays is to reveal the bony and structural abnormalities associated with back pain.
Literature Reviewed
Of the 128 articles screened for this topic, 20 articles reporting on 18 studies met criteria for review. [26], [276] - [294] Other articles contained information used by the panel, but did not meet article selection criteria. [295], [296]
Evidence on Efficacy
Two articles evaluated patients who had less than 3 months of symptoms. [26], [278] Two articles involved patients with chronic low back pain. [282], [292] Two articles had three separate groups, one asymptomatic, one with patients seen for acute low back problems, and a third comprising patients with chronic low back symptoms. [277], [281] The remaining 14 articles (reporting on 13 studies) involved a mix of patients with acute and chronic symptoms or did not report symptom duration. [276], [279], [280], [283] - [291], [293], [294]
Hansson, Bigos, Beecher, et al. [281] evaluated the degree of lumbar lordosis seen on x-rays in three groups of age-matched men engaged in heavy labor jobs. They included 200 asymptomatic subjects who had preemployment x-rays, 200 patients seen for acute low back problems, and 200 patients with low back disability longer than 6 months. Films were read by back specialists blinded to all other data. No differences were found between groups in the degree of lordosis noted on x-ray.
Deyo and Diehl [278] evaluated the use of a set of criteria based on medical history questions for selectively ordering x-rays in 621 patients presenting to a hospital emergency room with low back pain (patients with urinary tract symptoms excluded). The 11 criteria, any one of which would prompt an early x-ray, were: (1) age over 50, (2) significant trauma, (3) neuromotor deficits, (4) unexplained weight loss, (5) suspicion of ankylosing spondylitis, (6) drug or alcohol abuse, (7) history of cancer, (8) use of corticosteroids, (9) temperature greater than or equal to 100 F, (10) a return visit for the same problem within 1 month and not improved, and (11) patients seeking compensation for back pain.
X-ray findings were considered therapeutically important if they detected a malignancy or fracture. Of these patients who had x-rays, 227 met one or more of the criteria for x-ray, and 6.6 percent of these had therapeutically important findings. Another 84 patients had x-rays even though they did not meet any of the criteria. None of this group had therapeutically important x-ray findings. The highest diagnostic yield was in patients over age 50. Of the 119 patients in this group, 11 percent had therapeutically important findings (13 fractures and 2 malignancies). Strict use of the selective criteria would have resulted in x-ray studies in 390 patients.
Deyo and Diehl [26] evaluated 1,975 walk-in patients at a public hospital, their chief complaint back pain, in order to estimate the prevalence of cancer as an underlying cause of back pain. A search of an institutional tumor registry at least 6 months after the index visit identified 13 of these patients whose back pain was attributed to cancer. History findings significantly associated with the diagnosis of cancer were age over 50, prior cancer history, unexplained weight loss, pain lasting more than 1 month, and no improvement following conservative therapy. Laboratory test results significantly associated with cancer were an ESR over 20 and anemia. The authors presented an algorithm for ordering x-rays based on these history and lab test findings that would have resulted in 22 percent of the total group receiving x-rays including all patients who were found to have cancer.
Eleven articles evaluated x-ray findings in patients with low back problems, compared with asymptomatic subjects with no history of back problems. The findings of all of these studies were similar. In general, x-ray findings correlated poorly with low back problems. [276], [277], [279], [281], [284], [285], [287], [288], [291], [292], [294]
In all these studies, degenerative changes were noted in some persons with no history of low back problems whereas other persons with back problems showed no degenerative changes. When groups of subjects of similar age were compared, some studies show an increased prevalence of degenerative changes of the lumbar spine in those who have had back problems, compared to control groups with no history of back problems. Other studies found no significant difference in prevalence of degenerative changes between these groups.
These studies generally support the idea that degenerative changes of the spine, as seen on x-rays or other imaging studies, are not by themselves a cause of back pain. In fact, many authors suggest that degenerative changes of the spine are merely signs of aging of the back. Furthermore, degenerative changes were seen more commonly with increasing age for both those with and those without a history of low back problems.
A possible exception is spondylolisthesis, a forward slippage of a vertebra on the one below it. Two studies showed significantly higher prevalence of spondylolisthesis in patients with low back problems than in asymptomatic individuals. Torgerson and Dotter [294] found spondylolisthesis on x-rays in 4.9 percent of 387 symptomatic low back pain patients, but in only 1.5 percent of 217 asymptomatic subjects. Magora and Schwartz, [290] who evaluated adults currently working, found x-ray evidence of spondylolisthesis in 3.1 percent of 648 subjects with a history of low back pain patients, but in none of 376 subjects who had never had back pain.
Three other studies found no significant difference in prevalence of spondylolisthesis between groups with low back problems and asymptomatic controls. Biering-Sorensen, Hansen, Schroll, et al. [276] evaluated a group of men and women 60 years of age, including 308 with low back pain and 358 asymptomatic controls. This study found an overall prevalence of spondylolisthesis of 2.8 percent, with no significant difference between the groups of symptomatic patients and controls. Bigos, Hansson, Castillo, et al. [277] found no difference in incidence of spondylolisthesis or other abnormalities in the evaluation of age-matched males whether asymptomatic job applicants, back injury claimants, or disabled patients with greater than 6 months' duration of symptoms. Leboeuf, Kimber, and White [285] found spondylolisthesis on x-rays in 4.9 percent of patients with low back pain and in 5.5 percent of asymptomatic controls. This difference was not significant.
None of the studies reviewed on spondylolisthesis attempted to measure spinal instability, and most did not compare groups in terms of degree of spondylolisthesis. In addition, these studies did not investigate whether there was any therapeutic benefit from determining if a patient has spondylolisthesis (especially in patients with no findings that suggest nerve root compression).
Some authors maintain that adults with severe spondylolisthesis (degree of slip greater than 50 percent or a severe increase in affected joint motion) may require special treatment. [295] However, the degree to which spondylolisthesis and abnormal motion may benefit from special interventions is controversial and beyond the scope of this guideline.
Other reported findings on lumbar x-rays for adult patients with low back problems included: transitional vertebrae (lumbarization or sacralization), spina bifida occulta, increased or decreased lordosis, mild or moderate scoliosis and spondylolysis (an interarticular defect of the posterior vertebral arch), and degenerative joint disease. These findings were reported in similar frequency for 200 subjects without symptoms, for 200 subjects reporting back injury claim, and for 200 subjects disabled more than 6 months. This confirms multiple studies that have questioned both the use of lumbar radiographs for preemployment screening and the diagnostic significance of many radiologic findings. [277]
Spondylolysis without spondylolisthesis is equally common in persons with and without low back symptoms. [277] Spondylolysis has not been documented to be a cause or precursor of low back problems in adults, and no special treatment is required.
Potential Harms and Costs
Potential harms associated with lumbar spine x-rays are primarily related to the degree of ionizing radiation exposure. Small cumulative doses of ionizing radiation are believed to present minimal or no risks. But lumbar x-rays expose the male and female reproductive organs, especially with routine use of oblique views or repeated exposures. These practices are of questionable value and a particular concern for younger females. Oblique lumbar views approximately double total radiation exposure, compared to standard views, which alone are equivalent to female gonadal radiation of daily chest x-rays for 6, 16, or 96 years, depending upon the machine. [296] Lumbar spine x-rays are seldom indicated during pregnancy.
The cost of a lumbar spine x-ray is low compared to other imaging modalities such as bone scan, CT, or MRI, and x-rays are more readily available.
Summary of Findings
Plain lumbar x-rays have been demonstrated to be useful in helping detect or define spinal fractures, but alone do not rule in or out tumors or infections suspected by other findings (such as when red flags are present). Evidence suggests plain x-rays are rarely useful in evaluating or guiding treatment of adult acute low back pain in the absence of red flags. Plain x-rays are not effective for diagnosing lumbar nerve root impingement of herniated disc or spinal stenosis, or for ruling out cancer or infection.
The use of lumbar x-rays to screen for spinal degenerative changes, congenital anomalies, spondylolysis, spondylolisthesis, or scoliosis very rarely adds useful clinical information. Only 1 of 2,500 x-rays detects something not suspected on medical history and physical examination that has an impact on patient care. Even in the rare cases where a condition may be clinically significant, the history and physical examination findings (that is, evidence of neurologic dysfunction) should dictate more extensive diagnostic evaluation whether a routine x-ray screening is positive or negative. Oblique lumbar x-rays, usually done to screen for spondylolysis, rarely add useful clinical information in adults, and they double the x-ray dose to the patient. Patients with spondylolisthesis can be safely treated in the same fashion as those with other types of acute low back problems. Thus, x-rays done specifically to screen for the presence of spondylolisthesis are unnecessary in adults during the first 3 months of symptoms.
CT, MRI, Myelography, and CT-Myelography
Panel findings and recommendations:
Introductory Discussion
The four imaging tests commonly used in assessing the anatomy of the lumbar spinal canal and its contents are plain myelography, CT scan, MRI scan, and CT-myelography. These four tests are discussed as a group because they are used in similar clinical situations, provide similar types of information, and are often compared with each other in research studies. Evaluation is limited by lack of a gold standard for evaluating efficacy. Each test and the type of information it provides are described briefly as follows:
- CT scans use multiple x-ray beams projected at different angles and levels to produce computer-generated axial cross-sectional images of the body.
- MRI scans use magnetic fields to produce computer-generated axial and sagittal cross-sectional images of the body.
- Plain myelography uses plain x-rays, taken after a nonionic water-soluble contrast media is injected into the spinal canal via a lumbar puncture needle, to produce images of the borders and contents of the dural sac.
- CT-myelography uses a CT scan, done after a contrast media has been injected into the dural sac in the same manner as for plain myelography, to produce axial cross-sectional images of the spine that enhance distinction between the dural sac and its surrounding structures.
Significant technological advances have taken place in these imaging modalities over the past several years. Kent, Haynor, Larson, et al. [297] have suggested technical criteria for the performance of these imaging tests to assure a minimal level of quality. Studies of imaging tests were not considered by the panel if the technical protocols were inadequately described or if the protocols varied significantly among subjects within a study. The panel used the following technical criteria for including studies of individual imaging modalities:
- For myelography, contrast media should be water based, not oil based.
- For CT scans, the axial images (cuts) should be no wider than 0.5 cm and should be parallel to the vertebral endplates.
- For MRI scans, the image quality should be equivalent to or better than scans with magnetic field strength >0.5 T with an adequate scanning technique. Because of significant advances in technology, studies of MRI scans before 1985 were not considered.
The objective of using these imaging tests for patients with acute low back problems is to define medically or surgically remediable anatomic pathological conditions. Therefore, the tests are not done routinely, but are generally used only for patients who present with one of these three clinical situations:
- Back-related leg symptoms and clinically specific detectable nerve root compromise with symptoms severe enough to consider surgical intervention.
- A history of neurogenic claudication and other findings suggesting spinal stenosis with symptoms severe enough to consider surgical intervention.
- Clinical examination findings or other test results suggesting other serious conditions affecting the spine (such as cauda equina syndrome, spinal fracture, infection, tumor, or other mass lesions or defects).
This section assesses the usefulness of the four imaging tests for diagnosing nerve root compromise due to herniated discs or spinal stenosis. The use of imaging tests for evaluating other serious conditions affecting the spine is beyond the scope of this guideline and is not discussed in detail here.
Imaging Tests to Evaluate Suspected Nerve Root Compromise Due to Lumbar Disc Herniation
Literature Reviewed
Overall, 308 articles were screened for the subject of spinal imaging tests. They included 141 for CT scan, 97 for MRI, and 70 for myelography (including CT-myelography). Of the 308, there were 17 articles meeting criteria for review that evaluated lumbar disc herniation as an anatomic cause for suspected nerve root compromise. [20], [22], [30], [35], [298] - [310] Other information used by the panel came from Sackett, Haynes, Guyatt, et al. [311] All of these studies used surgical findings as a reference test.
Evidence on Efficacy
Four articles evaluated plain myelography alone. [20], [30], [35] , [301] Three of the four evaluated only patients with no history of prior back surgery. [20], [35] , [301] One included patients with prior back surgery. [30]
In the four studies, lumbar disc herniations were found at surgery in 68 to 96 percent of patients. Studies evaluating the true positive rate and true negative rate of a diagnostic test are generally considered more accurate when the target condition is present in 50 percent of the population studied. [311] Only two of these four studies had prevalence rates for lumbar disc herniation below 80 percent. Aejmelaeus, Hiltunen, Härkönen, et al. [20] studied a group with a 68-percent prevalence of disc herniation at surgery and calculated the true positive rate of 75 percent and true negative rate of 55 percent for myelography in diagnosing this condition. Herron and Turner, [30] found a disc herniation at surgery 74 percent of the time, and the true positive rate and true negative rate for myelography were 88 percent and 62 percent, respectively. The remaining two articles found a lumbar disc herniation at surgery in 84 to 96 percent of patients. In these studies, myelography was calculated to have a true positive rate of 83 to 94 percent and a true negative rate of 71 to 100 percent for diagnosing lumbar disc herniation. A total of eight studies evaluated CT scan compared with myelography. [22], [298] - [300], [304], [307] - [309] Only two of the studies looked at patient groups who had no prior back surgery. [22], [307] Two studies included some patients with prior surgery. [304], [309] The remaining four studies did not report on this variable.
For diagnosing lumbar disc herniation, all eight studies reported similar ranges of true positive and true negative rates for both tests, and these results were also similar to those from studies of myelography alone. The number of subjects ranged from 37 to 195, and prevalence of lumbar disc herniation found at surgery varied from 55 to 91 percent. The calculated true positive rate of the tests for diagnosing lumbar disc herniation varied from 60 to 91 percent for CT scan and from 65 to 98 percent for myelography. The true negative rate varied from 57 to 100 percent for CT scan and from 43 to 100 percent for myelography. One study allowed estimation of the added diagnostic value of combining information from CT and myelography for the same patient. [309] This study found similar accuracies for CT and myelography when evaluated separately for diagnosing lumbar disc herniation (sensitivities 77 percent and 78 percent, specificities 83 percent and 72 percent). However, when the results of both tests were evaluated together, the true positive and true negative rates were 91 percent and 56 percent for a positive result defined as either one of the two tests being positive. True positive and true negative rates were 64 percent and 100 percent if a positive result was defined as both tests being positive. This implies that the probability that a true lumbar disc herniation will be detected on imaging is 77 percent if one test is done and 91 percent if both tests are done. This also implies that the probability of an imaging test showing a false-positive lumbar disc herniation is 17 percent or 28 percent if one test is positive, but approaches zero if both tests are positive.
The five remaining articles evaluated multiple imaging tests done in the same patients. [302], [303], [305], [306], [310] MRI scan was compared to plain myelography, [310] to CT-myelography, [305] to both plain and CT-myelography, [306] and to CT, plain myelography, and CT-myelography. [303] Another study, also by Jackson, Cain, Jacobs, et al., [302] evaluated CT, myelography, and CT-myelography and compared them as well to discography and CT-discography. The five studies all found no significant differences between CT, MRI, and CT-myelography in terms of their true positive rates and true negative rates for diagnosing lumbar disc herniation although all these tests were better than plain myelography.
Jackson, Cain, Jacobs, et al. [302] found that for diagnosing lumbar disc herniation, CT-discography had a significantly greater true positive rate (92 percent) than CT-myelography (78 percent), plain CT (72 percent), plain myelography (70 percent), or plain discography (31 percent). However, in a subsequent study, Jackson, Cain, Jacobs, et al. [303] recommended MRI over other imaging tests as it is noninvasive and exposes the patient to no ionizing radiation. No differences were found between CT-discography, CT-myelography, CT, or myelography in the true negative rates although all had a significantly higher true negative rate than plain discography.
Imaging Tests to Evaluate Suspected Spinal Stenosis
Literature Reviewed
Lumbar spinal stenosis involves impingement of the cauda equina nerves in the spinal canal, lateral spinal canal, or neural foramina. This is usually a degenerative condition resulting from ligamentous infolding and bony hypertrophy commonly seen in persons over age 60. Spinal stenosis is occasionally developmental, due to congenitally short pedicles, and can affect younger individuals.
One meta-analysis on the accuracy of CT, MRI, and myelography for diagnosing lumbar spinal stenosis in adults was reviewed [297]. Of the 116 articles the authors screened for this meta-analysis, 14 studies met their inclusion criteria for review. [273], [275], [306], [309], [312] - [321]
Evidence on Efficacy
In their meta-analysis, Kent, Haynor, Larson, et al. [297] concluded that estimates of accuracy of the imaging tests for diagnosing spinal stenosis were imprecise and that all 14 studies had methodological flaws (all judged to be of fair or poor quality). However, based on data available, the meta-analysis found CT and MRI of similar accuracy for diagnosing spinal stenosis (true positive and true negative rates approximately 90 percent and 80 percent) and plain myelography with lower accuracy. The authors cautioned that because of bias, estimates of test accuracy obtained by the meta-analysis were likely to be too high.
Spinal Imaging Findings in Asymptomatic Subjects
Literature Reviewed
Six articles were found that evaluated lumbar spinal imaging findings in asymptomatic subjects with no low back symptoms. These studies included one evaluating CT scan, [275] four on [sup]MRI, [273], [291], [322], [323] and one for plain myelography. [274] Other articles contained information used by the panel, but did not meet article selection criteria. [30], [40], [49], [50]
Evidence on Efficacy
Wiesel, Tsourmas, Feffer, et al. [275] evaluated readings of CT scans for 52 subjects with no history of low back problems, with all scans read independently by three neuroradiologists blinded to other data. For those under age 40, herniated discs were diagnosed in an average of 19. 5 percent of subjects (range for the three readers, 13 to 24 percent). For subjects over age 40, those who had abnormal findings averaged 50 percent (with a range of 30 to 82 percent among the readers). Included were 27 percent of subjects diagnosed with herniated discs, 10 percent with facet joint hypertrophy, and 3 percent with spinal stenosis. Technical quality of this study was considered adequate.
Boden, Davis, Dina, et al. [273] evaluated readings of MRI scans in 67 subjects with no history of back problems. Scans were read by three neuroradiologists blinded to all other data after the scans were randomly mixed with scans of symptomatic patients who had proven back pathology. Of the asymptomatic subjects, 35 were age 20 to 39, 18 were age 40 to 59, and 14 were age 60 and over. The percentages for each of the above age groups were reported respectively by neuroradiologists as: a definite or probable abnormality in 22 percent, 22 percent, and 57 percent; bulging disc in 56 percent, 50 percent, and 79 percent; disc herniation in 21 percent, 22 percent, and 36 percent; degenerative disc in 34 percent, 59 percent, and 93 percent; spinal stenosis in 1 percent, 0 percent, and 21 percent. This study was considered of good technical quality.
Paajanen, Erkintalo, Dahlström, et al. [291] evaluated MRI scans in age- [20] male military conscripts including 75 subjects with low back pain and 34 controls with no history of low back problems. Five intervertebral discs, from L1-L2 to L5-S1, were evaluated for each subject. Disc degeneration, indicated by reduced MRI signal intensity for the disc, was found in 57 percent of patients with low back problems and 35 percent of controls. The authors stated that it was unclear if the positive findings in controls represented pathological processes or merely normal aging changes that did not predispose to future low back problems. The technical quality of the MRI scans in this study was considered suboptimal because low field strength MRI (0.02 T) was used.
Powell, Wilson, Szypryt, et al. [322] evaluated the MRI scans in 302 women who had no symptoms of low back problems, but who had the scans done for nonback obstetrical problems. Using reduced MRI signal intensity of the disc as an indication of disc degeneration, the authors found one or more degenerative discs in 34 percent of women age 21 to 30, in 60 percent of women age 31 to 40, and in 95 percent of women by age 70. Bulging discs were found on MRI in 11 percent of 82 pregnant women and 13 percent of 56 women who had never been pregnant (these data not presented by age group).
Weinreb, Wolbarsht, Cohen, et al. [323] evaluated MRI findings in 45 pregnant women age 20 to 39 and in 41 nonpregnant women age 19 to 40 with no current low back symptoms. Three intervertebral disc levels, L3-L4, L4-L5, and L5-S1, were evaluated for all subjects. No significant differences were found between pregnant and nonpregnant women in terms of the percentage with lumbar disc herniation (9 percent compared with 10 percent) or disc bulging without herniation (44 percent in each group). The technical quality of this study was considered suboptimal because testing protocols varied between subjects and low field strength MRI (0.35 T) was used.
Hitselberger and Witten [274] reviewed lumbar myelograms done in 300 patients with no prior history of radiculopathy. The technical quality of this study was considered poor because oil-based contrast media were used, which are less sensitive than more modern techniques. Still, 24 percent had myelographic evidence of lumbar disc herniation, 9.3 percent had spinal stenosis without a history of radicular symptoms (1 of 300 with a complete block of the dye). This study was of limited value because findings were not given by specific age groups, and the age range of subjects was 18 to 76 with a mean age of 51.
Degenerative discs, bulging discs, and even herniated discs are part of the aging process for the spine and may be irrelevant findings; they are seen on imaging tests of the lumbar spine in a significant percentage of subjects with no history of low back problems. Therefore, abnormal imaging findings seen in a patient with acute low back problems may or may not be related to that individual's symptoms. A herniated disc noted on an imaging test is more likely to be associated with a clinically significant nerve root compromise in patients when there are other findings (such as leg pain, positive straight leg raising, neurologic deficits, or a positive electrodiagnostic test) that suggest physiologic nerve root compromise. [30], [40], [49], [50]
Potential Harms and Costs
Factors that may influence the decision on which imaging test to use in diagnosing low back problems include the following: (1) tissue of greatest interest for imaging (CT better for bone, MRI better for imaging neural tissues and bone marrow and for diagnosing tumor or infection); (2) claustrophobia (more of a problem for MRI); (3) obesity (both CT and MRI scanners have a maximum table weight, so that extremely heavy patients may need to have myelography); (4) presence of internal metallic objects such as implanted medical devices, metallic surgical clips, or metallic objects or fragments in the eye, which can be a major problem for MRI (and sometimes for CT); (5) preference of the consultant and of the patient; (6) availability, cost, and potential side effects of the test.
CT-myelography and myelography have a higher risk of complications than CT or MRI (for example, post-spinal-tap headaches and adverse reactions to contrast media). For CT scan, potential harms include minimal x-ray exposure (including gonadal radiation exposure in females). Amount of exposure (less than two rads) is similar to that from two standard x-ray views of the lumbar spine. For MRI scan, there have been no clearly documented adverse health effects. Long-term effects of magnetic field exposure from MRI are unknown, but this test is generally believed to involve minimal risk. CT scan, MRI scan, CT-myelography, and plain myelography are all considered moderate to expensive in cost.
Summary of Findings
Given the benign natural history of acute low back problems, with 80 to 90 percent of patients expected to improve at 1 month even without treatment, routine spinal imaging tests are not generally necessary during the first month of symptoms except when a red flag (suggesting a medically emergent condition) is noted on medical history and physical examination. After 1 month of symptoms, the use of imaging tests may be appropriate when surgery is being considered for a specific detectable loss of neurologic function or to further evaluate possibly serious spinal pathology in the presence of red flags.
Discography
Panel findings and recommendations:
Discography involves the injection of a water-soluble imaging material directly into the nucleus pulposus of the disc. Information is then recorded about the amount of dye accepted, the pressure necessary to inject the material, the configuration of the opaque material, and the reproduction of the patient's pain. There are two diagnostic objectives: (1) to evaluate radiographically the extent of disc damage on discogram (sometimes with the addition of CT) and (2) to characterize the pain response (if any) on disc injection to see if it compares with the typical pain of the patient. A symptomatic degenerative disc is considered one that disperses injected contrast in an abnormal pattern, extending to the outer margins of the annulus and possibly into epidural space as well. For many investigators, a painful reaction provoked in the patient that reproduces the patient's usual pain is required to classify the disc as abnormal.
Literature Reviewed
Of the 42 articles evaluated for this topic, only 1 met review criteria for adequate evidence about efficacy. [324] Another article contained information used by the panel, although it did not meet article selection criteria. [325] There were also two articles reviewed that evaluated results of discography in asymptomatic subjects. [326], [327]
Evidence on Efficacy
No studies were found that evaluated discography for patients with acute low back problems. One of the major problems in evaluating the literature on discography is that few studies evaluate discography using an independent reference standard. Instead, many studies either compare discography results with other diagnostic test findings or evaluate discography using pain provocation on disc injection as evidence of discogenic pain, which essentially means using part of a test to validate itself.
In the only study to meet panel review criteria, Colhoun, McCall, Williams, et al. [324] evaluated the results of discography using surgical findings and long-term clinical outcomes as independent reference standards. This study evaluated discography in 195 patients with persistent low back pain (symptom duration not otherwise specified), but with no history of prior back surgery. All patients went on to have back surgery (spinal fusion for 82 percent of patients, laminectomy and/or discectomy for the remaining patients). All patients were followed for 2 to 10 years postoperatively (results not presented separately by type of operation). Patients who had spinal fusion were excluded from the study if the surgery was deemed to be technically unsuccessful (that is, if there was x-ray evidence of pseudoarthrosis at the site of attempted fusion). Treatment success on followup was noted in 89 percent of 137 patients who had positive pain response on discography and abnormal discograms, compared to 52 percent of 25 patients who had negative pain response but abnormal discograms and 50 percent of 12 patients who had a negative pain response and normal discography.
Holt [326] evaluated discograms done in 30 male prison inmate volunteers who had no prior history of low back problems, normal lumbar x-rays, and normal back examinations. Their age range was 21 to 49, average age 26. Discography was attempted at the L3-L4, L4-L5, and L5-S1 disc levels for each subject. Of the 71 successfully performed discograms, a positive pain response was noted in 37 percent of the discs injected. All of these also had abnormal discograms (showing either degenerative changes or disc rupture). The remaining 63 percent of discs injected had negative pain responses and normal discograms. Holt interpreted these results as showing a 37-percent false-positive rate for discograms. However, in a reevaluation of this study, Walsh, Weinstein, Spratt, et al. [327] noted inconsistencies in Holt's data and recalculated the false-positive rate as either 26 percent or 4 percent depending upon assumptions made.
Walsh, Weinstein, Spratt, et al. [327] evaluated CT-discography done in 7 patients with low back pain and in 10 "asymptomatic" subjects with no history of low back problems. Discography was attempted at three lumbar disc levels for each subject, with a water-soluble nonionic contrast media. Disc injection was videotaped, and the subjects' pain reactions and discograms were later read independently by two orthopedic surgeons and three radiologists blind to all other data. Discography in the patients was considered positive if the discogram was abnormal, if there were two or more videotaped pain behaviors, and if disc injection provoked the patient's typical pain. In asymptomatic subjects, discography was considered positive if the discogram was abnormal and there was significant pain noted on disc injection. Discography was positive in all 7 patients with low back problems (65 percent of the 13 discs successfully injected being abnormal). In the asymptomatic subjects, 50 percent had an abnormal discogram (17 percent of the 30 successful discograms for this group), but none had positive pain response on disc injection. Therefore, none of the asymptomatic subjects was considered to have positive discography (that is, no false- positive tests).
Potential Harms and Costs
Discography is an invasive procedure with risk of complications. Potential complications include disc and disc space infections, disc herniation following disc injection, and significant amounts of ionizing radiation exposure with CT-discography (estimated at 1.5 to 4.0 rads when studying 2-3 discs). [325] Discography is expensive.
Summary of Findings
Although discography seeks to identify internal changes in the disc based on evaluation of a discogram and pain response on disc injection, the disease process is not clear (the internal disc disruption), and how to use the information for acute low back problems is not clear. The main reason put forward for using discography appears to be to determine the levels at which spinal fusion will be successful in patients with persistent low back problems (due to discogenic disease). This assumes the controversial premise of a "painful disc syndrome" which has not yet been adequately documented.
There is no good evidence that discography is useful to promote better treatment outcomes in patients with acute low back problems. Colhoun, McCall, Williams, et al. [324] noted more successful outcomes after back surgery for those with a positive discography, but these were presumably patients with chronic problems (as they had persistent pain). The rationale for using discography is to select patients who would most benefit from spinal fusion. Yet, in the Colhoun article, the predictive value of positive discography in establishing which patients do well after spinal fusion cannot be determined accurately because 18 percent of patients did not have fusions and outcomes were not reported separately for fusions and discectomies. Moreover, the two articles on discography in asymptomatic subjects report a substantial percentage of subjects who had positive discograms although they had never had low back problems.
In summary, there is limited evidence that discography can help select patients who would benefit from spinal fusion and no evidence that it is helpful in patients with acute low back problems. Potential serious risks from discography, including disc infection, have been identified. The use of discography or CT-discography to diagnose herniated discs appears to offer no significant advantage over other imaging methods with less potential risk of harm.
Surgical Information
Surgery is commonly discussed for back symptoms that are unresolved after special studies (Attachment A4). Patients may benefit from general information about the risks and potential outcomes of surgical treatment for different diagnoses.
Surgery for Herniated Disc
Panel findings and recommendations:
Surgery for herniated discs is invasive and comprises all types of surgical and injection techniques to remove or reduce the size of herniated intervertebral discs that compress nerve roots. Included are standard discectomy, microscopic discectomy, percutaneous discectomy, and chemonucleolysis (chymopapain injection). The therapeutic objective is to relieve pressure on nerve roots and reduce pain and possibly weakness and/or numbness in the lower extremities.
Literature Reviewed
Of 345 articles screened for this topic, 13 reporting on 10 studies met criteria for review. [322], [328] - [339] In addition, a meta-analysis article was reviewed by the panel. [340] Other articles contained information used by the panel, but did not meet article selection criteria. [30], [40], [49], [50], [341] - [345]
Evidence on Efficacy
Three of the studies reviewed were randomized controlled trials (RCTs) that evaluated chymopapain as compared with standard discectomy for patients with symptoms and findings of lumbosacral radiculopathy. Crawshaw, Frazer, Merriam, et al. [328] found that at 1-year followup, 85 percent of those undergoing discectomy had good or excellent results, compared to 44 percent of those receiving chymopapain injections. Both groups had improved leg symptoms, but only the discectomy group had significant improvements in back pain.
Ejeskär, Nachemson, Herberts, et al. [329] found that at 6 months, 56 percent of patients initially receiving chymopapain injections had undergone surgery due to unrelieved symptoms (all of these patients with disc herniation confirmed at surgery). When patients were followed up during the first 6 months before any treatment crossovers occurred, those receiving discectomy reported significantly greater improvement in symptoms than those receiving chymopapain injections.
Van Alphen, Braakman, Bezemer, et al. [336] found that at 1-year followup, physicians rated significantly more of the patients in the discectomy group as having good results from their initial treatment compared with the chymopapain group (85 percent compared with 63 percent). Also, 25 percent of the chymopapain group required subsequent discectomy while only 3 percent of the discectomy group needed a second operation. Open discectomy following prior failed chymopapain injection was successful in 44 percent of the cases.
Another RCT (reported in Fraser; [330] Fraser; [331] and Gogan and Fraser [332]) evaluated chymopapain injections compared with intradiscal saline injections in patients who had sciatica and evidence of a herniated lumbar disc, but who had not improved after 6 months of conservative therapy. Treatment success rates were significantly better for the chymopapain group than for the saline group at 6-month followup (80 percent compared with 57 percent), at 5-year followup (73 percent compared with 47 percent), and at 10-year followup (80 percent compared with 34 percent). Also, significantly fewer patients required a laminectomy for unrelieved symptoms in the chymopapain group compared with the saline group when followed up at 2 years (20 percent compared with 40 percent) and at 10 years (20 percent compared with 47 percent).
A double-blind RCT by Javid, Nordby, Ford, et al. [333] evaluated chymopapain injection compared with intradiscal saline injection in 108 patients with sciatica who had not improved after at least 6 weeks of conservative treatment including 2 weeks of bed rest. All patients had positive straight leg raising, a neurologic defect on physical examination, and myelogram evidence of a single lumbar disc herniation. At 6 months postinjection, the treatment success rate was significantly better for the chymopapain group compared with the saline injection group (83 percent compared with 42 percent).
Revel, Payan, Vallee, et al. [334] evaluated chymopapain injections compared with percutaneous discectomy in 141 patients with sciatica who did not improve after 1 month of conservative treatment. Overall success rates were significantly better for the chymopapain group than for the percutaneous discectomy group at both 6-month followup (61 percent compared with 44 percent) and 1-year followup (66 percent compared with 37 percent). Also, fewer patients in the chymopapain group required open laminectomy within 1 year due to unrelieved symptoms as compared with the percutaneous discectomy group (7 percent compared with 33 percent).
Only one RCT (reported by Weber [338], [339] ) compared standard discectomy with conservative (nonsurgical) care. This study looked at 280 patients who consecutively presented to a hospital neurology department with severe sciatica and clinical findings of possible or definite L4-L5 or L5-S1 disc herniation. All patients were initially hospitalized for 2 weeks of conservative treatment including 1 week of strict bed rest. After 2 weeks, 87 patients with possible but indefinite disk hernia who had shown continuous improvement were assigned to continued conservative care, and discectomy was performed for 67 patients who were deemed by their surgeon to have emergent indications for immediate surgery (intolerable pain, suddenly occurring or progressive muscle weakness, or impending bladder or anal sphincter paresis). This left a group of 126 subjects who had continued sciatic symptoms provoked by mild exercise, sitting, or Vasalva maneuver and with myelograms that showed definite disc herniations consistent with their clinical findings of neurologic dysfunction. Patients in this latter group were randomly assigned to receive either standard discectomy or continued conservative treatment.
Of these 126 subjects with definite disc hernia who were randomized, Weber [338], [339] found good or fair results in 61 percent of the nonsurgery group and in 90 percent of the discectomy group at 1-year followup. This difference was no longer significant on followup at 4 years (86 percent compared with 89 percent) or 10 years (93 percent compared with 92 percent). At 4-year followup, considerable sciatic pain was still reported by 9 to 10 percent of each group and considerable low back pain by 11 to 12 percent of each group. At 10-year followup, no patients in either group reported these symptoms. During the first year, 26 percent of the nonsurgery group demanded discectomy because of unrelieved sciatic pain. On followup, good or fair results were reported for 82 percent of this group at 4 years and for 100 percent of the group at 10 years.
Tullberg, Isacson, and Weidenhielm [335] evaluated microdiscectomy compared with standard discectomy in a RCT of patients with sciatica who had not improved after 2 months of conservative treatment and had evidence on CT scan of a single-level disc herniation. At 1-year followup, no significant difference was noted between microdiscectomy and standard discectomy groups in terms of patients reporting excellent or good results (86 percent compared with 90 percent), mean postoperative time off work (10.4 weeks compared with 10.1 weeks), or mean improvement in visual analog pain scale ratings over the prior year.
A meta-analysis for herniated lumbar disc surgery by Hoffman, Wheeler, and Deyo [340] found only two RCTs. [336], [338], [339] This meta-analysis concluded that patients with severe leg symptoms and confirmed lumbar disc herniation experienced faster symptom relief and improved functioning if they underwent standard discectomy rather than conservative treatment. The meta-analysis stated that there was inadequate evidence available to determine the efficacy of microdiscectomy or percutaneous discectomy for treating low back problems. However, percutaneous discectomy was noted to have a higher reoperation rate than standard discectomy. Discectomies were noted to be relatively safe procedures, but reoperations were also reported to be fairly common.
Several studies have emphasized the role psychosocial factors play in influencing the outcome of surgery for herniated disc. [30], [40], [49], [50] In fact, Spengler, Ouellette, Battié, et al. [50] found that psychological factors, especially elevated hysteria or hypochondriasis scales on a Minnesota Multiphasic Personality Inventory (MMPI), were better predictors of surgical outcome than were findings on imaging studies.
Potential Harms and Costs
Reported complications of herniated disc surgery include operative mortality, wound infection, discitis, dural tears, nerve root injuries, thrombophlebitis and pulmonary emboli, meningitis, cauda equina syndrome, psoas hematoma, vascular injuries, and risks associated with transfusions. Hoffman, Wheeler, and Deyo [340] stated that, overall, complications appeared to occur infrequently, but that information on complications was often hard to interpret in the studies reviewed, making determination of rates difficult.
A community-based study in Michigan, which reviewed the hospital discharge records of over 28,000 patients who had a lumbar disc surgery in 1980, reported incidence rates of 0.06 percent for mortality, 0.3 percent for infections treated with intravenous antibiotics, and 0.3 percent for major neurologic complications. A similar study of hospital discharge records in Washington State found an overall mortality incidence of 0.07 percent for spinal surgery. [345]
The reported complications for chymopapain therapy include allergic reactions (some resulting in death), discitis, thrombophlebitis, pulmonary embolus, neurologic injury, vascular injury, and transverse myelitis. [341]- [344] Skin tests for sensitization before chymopapain injection are reported to significantly reduce the risk of allergic reaction. Transverse myelitis is uncommon, and the risk factors for this are unclear.
Complication rates for discectomy are generally low. Complication rates for chymopapain are also low, but the use of chymopapain has decreased drastically in the United States because of concerns about transverse myelitis and anaphylactic reactions. The use of tests for allergic sensitization to chymopapain before this treatment may reduce the risk of allergic reaction.
Surgery for herniated discs is considered an expensive treatment.
Summary of Findings
Lumbar discectomy may relieve symptoms faster than continued nonsurgical therapy in patients who have severe and disabling leg symptoms (associated with clinical examination findings of definite lumbar nerve root compromise) and who have not improved after 4 to 8 weeks of adequate nonsurgical treatment. However, the evidence also showed that in such nonemergent patients, there appears to be little difference in long-term outcomes at 4 and 10 years between discectomy and conservative care.
There are direct methods of nerve root decompression and indirect methods. The best results from herniated disc surgery were with direct methods of standard discectomy and microscopic discectomy with no significant difference in results noted between these procedures. Studies indicate that the direct method of standard discectomy produces better results than the indirect method of chymopapain. Chymopapain is better than placebo injections of saline or the indirect method of percutaneous discectomy. Rates of initial treatment failure requiring a subsequent surgery are higher for chymopapain than for discectomy.
No studies to date have shown that any of the newer indirect methods of disc surgery, such as indirect automated percutaneous and percutaneous laser discectomy, produce better results than standard discectomy or chymopapain in appropriate patients.
Patient preference does and should play a large role in the surgical decision, but only if adequate information is available about efficacy, risks, and expectations.
Surgery for Spinal Stenosis
Panel findings and recommendations:
Spinal stenosis includes any constriction or narrowing of the central spinal canal, the lateral recesses, or foramina resulting in compression of nerve roots and/or the cauda equina. Surgery for spinal stenosis may include various types of surgical techniques, usually including decompressive laminectomy (sometimes combined with discectomy and/or spinal fusion) done to alleviate the symptoms of neural compression.
Spinal stenosis is generally a degenerative condition affecting patients over age 60 and is usually related to a variety of age-related changes in spinal anatomy that combine to cause compression of the cauda equina in the lumbar spinal canal and neuroforamina. These age-related changes may include disc bulges and herniations, thickening of the infolded ligamentum flavum, degeneration of the facet joints and joint capsules, osteophytes, and occasionally degenerative spondylolisthesis. There are also some younger individuals with severe congenital narrowing of the spinal canal who have spinal stenosis symptoms, but this is uncommon.
The primary symptoms of severe spinal stenosis are neurogenic claudication (leg pain with walking or standing, relieved by sitting or spinal flexion) and occasionally weakness of the legs.
Literature Reviewed
Of the 40 articles screened for this topic, only one met criteria for review. [346] In addition, the panel reviewed an attempted meta-analysis on the topic. [66] Other articles contained information used by the panel, but did not meet article selection criteria. [347], [348]
Evidence on Efficacy
No RCTs were found evaluating surgery compared with nonsurgical treatment for spinal stenosis. However, a controlled trial by Herkowitz and Kurz [346] compared decompressive laminectomy done with and without spinal fusion in patients with chronic low back problems (mean symptom duration 3.0 years). None of the patients had improved after adequate conservative treatment, and all had findings consistent with degenerative spondylolisthesis and spinal stenosis on CT, MRI, or myelogram. Internal fixation using metal devices was not done as part of the fusion procedure. Immediately following surgery, the fusion group reported significantly more pain relief in the back and legs than did the nonfusion group. At long-term followup (mean followup time of 3.0 years), the percentage of patients reporting excellent or good outcomes was significantly greater for the fusion group than for the nonfusion group (96 percent compared with 44 percent).
An attempted meta-analysis reviewed 74 studies involving patients who had decompressive laminectomy for lumbar spinal stenosis. [66] The authors of the attempted meta-analysis found three prospective studies, but no randomized controlled trials comparing surgery with conservative treatment. Therefore, no conclusions could be drawn as to the relative benefits and risks of surgery compared with conservative treatment. One study reported that of 27 unoperated patients with spinal stenosis, 19 remained unchanged, 4 improved, and 4 deteriorated during a followup period of 10 to 103 months with a mean of 49 months. [347] A comparison of these outcomes by Turner, Ersek, Herron, et al. [66] suggested that patients may show more improvement with surgery, but conservative management may be a reasonable alternative.
Of the 74 studies reviewed by Turner, Ersek, Herron, et al., [66] 31 provided sufficient information to calculate the proportion of patients with good-to-excellent as compared with poor-to-fair outcomes. Criteria for classifying outcomes into categories of excellent, good, fair, and poor varied across studies, making it difficult to combine and compare studies. When results were classified using a standard method for each article, the percentage of patients whose outcomes were classified as good to excellent varied from 26 to 100 percent (mean of 64 percent) for the 31 articles.
Most patients who have surgery for lumbar spinal stenosis have had their symptoms for over a year. It is not a common surgical consideration within the first 3 months of symptoms. [66]
Reoperation rates after spinal stenosis surgery (which were reported in 25 of the articles) ranged from 0 to 21 percent (with a mean of 8 percent). However, the authors speculated that these rates probably underestimated the true reoperation rates because of some followup periods that were very short.
The attempted meta-analysis by Turner, Ersek, Herron, et al. [66] concluded that in patients with severe symptoms from lumbar spinal stenosis, decompressive laminectomy appears to be beneficial for many patients by reducing pain and increasing the ability to function and is probably associated with an acceptably low complication rate.
There are also some data showing that although short-term results from spinal stenosis surgery may be good for most patients, for many patients there may be a progressive deterioration over time. A case series by Katz, Lipson, Larson, et al., [348] which followed 88 patients over age 55 who had decompressive laminectomy for symptoms of degenerative spinal stenosis, found that 89 percent of patients reported good outcomes (defined as absence of severe pain and no repeat operations) at 1-year followup, but only 57 percent of patients reported good outcomes at 3- to 6-year followups.
Potential Harms and Costs
Complications were also reviewed in the Turner, Ersek, Herron, et al. [66] meta-analysis. Death as a direct result of the surgery was rare, with a mean in-hospital mortality of 0.32 percent. Other reported complications of spinal stenosis surgery included dural tears (mean 0.32 percent), deep infection (mean 5.91 percent), superficial infection (mean 2.3 percent), and deep vein thrombosis (mean 2.78 percent).
The complication rate for spinal stenosis surgery is 10-15 percent, with half the complications being serious or life-threatening problems. These complications can result from the general anesthetic, the spinal problem itself, or coexisting medical problems. The high rate of complications may be because of the older age of this group.
Surgery for spinal stenosis is considered expensive.
Summary of Findings
No RCTs were found comparing surgery with nonsurgical treatment for spinal stenosis. Symptoms of severe and persistent spinal stenosis may either remain the same, gradually worsen, or improve with time. The duration of symptoms for most patients who have surgery for lumbar spinal stenosis exceeds 1 year. This surgery is not commonly considered within the first 3 months of symptoms.
For patients with a history of severe and persistent symptoms of neurogenic claudication who undergo spinal stenosis decompressive laminectomy, the most likely outcomes are less leg pain and improved walking tolerance. However, there is some indication that these results tend to deteriorate over time.
The potential for serious complications from this procedure is considered to be acceptably low, although the rate is higher than for other procedures primarily because this operation is usually done in older individuals.
Spinal Fusion
Panel findings and recommendations:
Lumbar spinal fusion surgeries use bone grafts, and are sometimes combined with metal devices, to produce a rigid connection between two or more adjacent vertebrae. The therapeutic objective of spinal fusion surgery for patients with low back problems is to prevent any movement in the intervertebral spaces between the fused vertebrae, thereby reducing pain and any neurologic deficits. Various theoretical rationales are given for the use of fusion in patients with low back problems. One theory postulates that in cases of significant spinal instability (abnormally increased motion at an intervertebral level), fusion prevents painful compression of the neural structures. Another controversial theory holds that, in some cases, back symptoms arise from the disc itself and fusion relieves symptoms by greatly reducing forces compressing the disc.
Literature Reviewed
Of the 129 articles screened for this topic, 3 met criteria for review. [346], [349], [350] In addition, a meta-analysis article was reviewed. [351] Other articles contained information used by the panel, but did not meet article selection criteria. [352]- [356]
Evidence on Efficacy
One RCT evaluated the rate of fusion with and without Knodt rods in patients with a diagnosis of multiple level foraminal stenosis. [349] All patients were treated by wide decompressive laminectomy, foramenotomy and bilateral-lateral fusion from L3 to the sacrum. The rates of fusion, patient-reported functional status, and the lengths of postoperative hospital stays were not significantly different between the two groups.
A controlled trial by Herkowitz and Kurz [346] compared laminectomy with and without fusion in patients with chronic symptoms (mean symptom duration 3.0 years) who did not improve after nonoperative treatment and who had a positive imaging finding (myelogram and either CT or MRI) consistent with degenerative spondylolisthesis and spinal stenosis. Internal fixation using metal devices was not done as part of the fusion procedure. Following surgery, the fusion group reported significantly greater pain relief in the back and legs than did the nonfusion group. At mean followup of 3.0 years, excellent or good outcome rates were significantly greater for the fusion group than for the nonfusion group (96 percent compared with 44 percent).
The third study meeting review criteria was a nonrandomized trial comparing laminectomy with and without Knodt rod fusion in patients with chronic symptoms, leg pain or neurologic deficit, who did not improve after 3 months of conservative care and who had a positive imaging finding (EMG and myelogram or CT) for a herniated disc. [350] At long-term followup (mean followup time of 4.9 years for the fusion group and 3.7 years for the nonfusion group), the percentages of patients reporting satisfactory results (excellent or good) were not significantly different between the two groups. There are serious design problems with this study, such as the fusion group having a significantly longer mean duration of preoperative symptoms: 5.7 years compared with 1.3 years.
An attempted meta-analysis on this topic [351] found only four nonrandomized studies comparing surgery with and without fusion for herniated disc [350], [353], [354], [356] Three of the four trials reported no significant difference in results between the two groups. One trial did report significantly better results with fusion, but the treatment groups were not comparable. The fusion group had significantly more abnormal findings on x-ray, fewer positive straight leg raising tests, longer duration of pain, and fewer workers' compensation patients. [356]
Potential Harms and Costs
Turner, Ersek, Herron, et al. [351] indicated that complications are frequent with lumbar spinal fusions. Based on a review of a large case series, the mean rates for the most common reported complications were 7.3 percent for instrumentation failure and 10.8 percent for bone graft donor site pain. The mean rates for other complications were 0.2 percent for in-hospital mortality, 1.5 percent for deep infection, 1.6 percent for superficial infection, 3.7 percent for deep vein thrombosis/ thrombophlebitis, 2.2 percent for pulmonary embolus, 2.8 percent for neural injury, 2.0 percent for graft extrusion, and 8.7 percent for other complications. Spinal fusion is also considered an expensive procedure.
Summary of Findings
There appears to be no good evidence from controlled trials that spinal fusion alone is effective for treatment of any type of acute low back problems in the absence of spinal fractures or dislocation. In the opinion of the panel, there may be two conditions where spinal fusion could be effective. The first is in cases of combined degenerative spondylolisthesis, stenosis, and radiculopathy where patients have decompressive laminectomy for spinal stenosis symptoms. The second situation is in some young patients (generally under age 30) with significant spondylolisthesis and severe leg pain who may receive some benefit from stabilization procedures, although this has not been proven in controlled trials.
Although the usual reasons stated for doing spinal fusion for degenerative problems are instability of the spine and disc disease, there is lack of scientific agreement on how to define spinal instability. Spondylolisthesis is often implicated as a cause of instability, but it may or may not have any detectable abnormal motion and the extent to which this contributes to low back symptoms is controversial. Moreover, there is no good evidence that patients who undergo fusion will return to their prior functional level. It appears that fusion is not commonly considered for adults within the first 3 months of symptoms except for fracture or dislocation. [346], [350]
Assessment of Psychosocial Factors
Panel findings and recommendations:
Social, economic, and psychological factors have been reported to be more important than physical factors in affecting the symptoms, response to treatment, and long-term outcomes of patients with chronic low back problems. [357] There are indications that such nonphysical factors may affect clinical outcomes for patients with acute low back symptoms (Attachment A5). A heightened awareness among clinicians to the way such factors may affect a patient's response to symptoms and treatment is therefore warranted.
Literature Reviewed
None of the articles screened about psychosocial factors in the assessment and treatment of low back problems were controlled trials and therefore do not meet panel review criteria for adequate evidence about efficacy. Five articles, however, that included prospective cohort studies contained information considered useful to the panel. [23], [50], [358] - [360]
Evidence of Efficacy
One large prospective study [358] of asymptomatic individuals at a worksite found premorbid nonphysical factors (i.e., measures of low work satisfaction and poor work performance reports) to be the best predictors of individuals reporting back problems at work. In a second study, [23] psychological variables measured early in the course of an acute low back episode did not predict outcome, although other nonphysical factors, such as educational level and perception of both job characteristics and "fault" concerning the back problem, were strong predictors of outcome. A population-based study [359] also suggested that psychosocial issues affect how individuals with low back symptoms make decisions about working.
Several studies have detected a strong correlation between the outcome of lumbar spine surgery and the preoperative psychological status of the patient. [50], [360]
Summary of Findings
The panel found evidence that psychological, social, and economic (nonphysical) factors can alter the response to symptoms and to treatment (including surgery) among patients with acute low back problems. While such nonphysical factors have been shown to affect outcomes, specific and effective interventions to address these factors and alter patient outcomes have yet to be defined. No studies that directly evaluated interventions aimed at psychosocial factors among patients with acute low back problems were found.
Given such limited information, the panel was55555 unable to recommend specific assessment tools or interventions focusing on psychosocial factors potentially important for patients with acute low back problems. Recognizing the impact such nonphysical issues can have on outcomes, however, the panel recommended that clinicians be aware of these factors, especially in patients whose recovery of activity tolerance following an acute low back problem seems delayed. Further research is needed to define specific methods of detecting nonphysical factors as well as interventions that might improve outcomes for those patients slow to recover from acute low back problems.
- Special Studies and Diagnostic Considerations - Acute Low Back Problems in Adult...Special Studies and Diagnostic Considerations - Acute Low Back Problems in Adults
Your browsing activity is empty.
Activity recording is turned off.
See more...