Evidence

Discography was originally described by Lindblom to diagnose a herniated lumbar intervertebral disc and has been used clinically since the 1940s. In early discography studies, patients were noted to commonly report reproduction of their typical pain with stimulation, and abnormal discs were rarely painful in asymptomatic controls. In 1968, Holt found a high false-positive rate, but there were 16 ruptured discs, and there is conjecture that there could have been irritation of structures other than the disc.

The validity of discography has been more recently evaluated in several studies, but remains controversial. Discography has been performed in symptomatic and asymptomatic subjects to ascertain its reliability and validity. To determine the false-positive rate, the rate of positive lumbar discography in individuals asymptomatic of low back pain has been studied. In 1990, Walsh and colleagues reported a false-positive rate of 0% in asymptomatic controls and a true-positive rate of 89% in symptomatic subjects. The study used pressures 58 to 72 psi above opening pressure. In 2000, Carragee and colleagues performed lumbar discography on 10 volunteers who were asymptomatic of low back pain and without confounding factors, citing a false-positive rate of 10% per patient. However, pressure manometry was up to 100 psi above opening pressure, and the sample size was small. In 2005, a study by Derby and colleagues of lumbar discography on asymptomatic controls used more stringent criteria for a positive discogram including pain greater than or equal to 6/10 on pain scale, grade III annular tear, and pressure less than or equal to 50 psi above opening pressure. The investigators found a false-positive rate of 0% in 13 asymptomatic controls (including 6 with occasional low back pain, but <3 episodes per year). In asymptomatic patients without confounding factors and using standardized diagnostic criteria, lumbar discography has negligible false-positive rates.

Carragee and colleagues also studied asymptomatic patients, but who had previously undergone discectomy. The investigators found a false-positive rate of 40% per patient and noted that 7 of 8 of the positive results were at the site of previous discectomy. Wolfer and colleagues reviewed this study and noted that if ISIS/IASP standards were used, the false-positive rate decreased to 15% per patient and 9.1% per disc. Previous surgical intervention at the disc increases the risk of a false-positive response for lumbar discography, especially if strict diagnostic criteria are not used.

In patients without low back pain but with chronic pain of other origin, higher false-positive rates are noted. Carragee and colleagues studied the results of lumbar discography in patients without low back pain but with chronic pain at the iliac crest after bone graft harvest (for grafting at a site remote from the lumbar spine). False-positive discography results were found in 4 of 8 patients, with disc stimulation reproducing their usual iliac crest pain. However, pressures up to 100 psi above opening pressure were used, and on secondary analysis using ISIS/IASP standardized criteria, the false-positive rate dropped to 12.5% per patient and 7.1% (1/14) per disc. The reproduction of pain at their iliac crest harvest site with provocation of normal discs was more pronounced at high pressures. These false-positive rates correlate to the overlapping nociceptive fields in the lumbosacral spine in which differentiation between pain generators is difficult (eg, sacroiliac joint vs zygapophyseal joint vs disc) and segmental sensitization can occur. In this study, it was not stated whether discography was done ipsilateral or contralateral to the side of the iliac crest pain. If injection was performed ipsilateral to the usual chronic pain, within the sensitized nociceptive field, it may have caused an increased false-positive rate. A lower false-positive rate would then have been possible if injection had been done on the side contralateral to the usual chronic pain, outside the sensitized nociceptive field. The finding of increased false-positive rates in patients with pain from other structures within the lumbosacral region highlights the importance of considering other possible pain generators prior to discography.

In patients with chronic severe pain of the cervical spine, who had the poorest outcomes after cervical spine surgery but no significant low back pain, Carragee and colleagues noted the false-positive rate for lumbar discography to be 40% per patient. Again, if the results are calculated using ISIS/IASP standardized diagnostic criteria, the false-positive rate drops to 0%. This pool of patients had diagnoses of cervical degenerative disc changes and neck pain severe enough that they had undergone cervical surgery for their pain. Because cervical and lumbar disc disease coincide, it is possible that the lumbar discs of the patients also had some amount of disc disease. These mildly abnormal, albeit asymptomatic, discs could then test positive with less stringent diagnostic criteria.

In an extensive review by Wolfer and colleagues, the results of lumbar discography in patients with chronic pain of nonlumbar origin were pooled. Including those with chronic cervical pain and chronic iliac crest pain, the original investigators cited a false-positive rate of 44%. By changing the diagnostic criteria as pain at less than 50 psi above opening pressure, the false-positive rate dropped to 33%, and by including all ISIS/IASP diagnostic standards, the false-positive rate was only 5.6% per patient and 3.85% per disc. Even in patients with chronic pain, lumbar discography has a low false-positive rate.

Carragee and colleagues also studied lumbar discography false-positive rates in 4 patients with somatization disorder and no low back pain. The false-positive rate was high at 75% per patient with the investigators’ analysis and 50% with ISIS/IASP standards. However, this study had a small sample size and large confidence intervals (CIs) (0%–100%). In contrast, another study of lumbar discography in 81 patients, 27 with abnormal psychometrics but without somatization disorder, showed no correlation between psychometric scores and positive discography. It is unclear as to what degree psychological factors affect discography results. Nonetheless, the North American Spine Society position statement includes significant psychological overlay as a relative contraindication to discography. Especially in patients with somatization disorder, it should be considered that there is higher risk for iatrogenic illness and recurrent pain after any interventional or surgical procedure.

In another study, lumbar discography was performed on patients with chronic mild low back pain, and positive discography was cited as being false positive. A false-positive rate of 36% per patient was reported. Secondary analysis applying ISIS/IASP standards found a false-positive rate of 16% per patient. This increased false-positive rate is likely a result of some of these results actually being true-positives. Patients in this study with “low back pain every or almost every day” and a visual analogue scale (VAS) of 2.2 to 4.1 but not receiving medical care may still have discogenic pain for which a positive discography would be a true-positive rather than a false-positive result. Discography is not designed to determine the severity of the discogenic pain but rather to determine whether provocation of the disc elicits the patient’s typical pain.

Carragee reviewed the results of several of his own studies and evaluated for a change in the outcomes by changing from a threshold of 100 psi above opening pressure to the low-pressure standard of 22 psi. There was no statistical difference in false-positive rates, and therefore, it was concluded that there was no added value of using low-pressure criteria. On further review, however, the somatization group and the chronic low back pain group were included in this analysis. The somatization group had a high incidence of false-positives at low pressures and the low back pain group most probably included true-positives. In an analysis of all studies, excluding these two populations, the false-positive rate (per patient) was found to be 16% using 22 psi above opening pressure, 10.7% using 15 psi, and 9.3% using ISIS/IASP standards. The standardization of diagnostic criteria, including a low-pressure technique, does improve results.

In an extensive post hoc analysis of lumbar discography studies in asymptomatic patients without significant confounding factors, a total of 33 patients and 48 discs were included. Using ISIS/IASP standards across all studies, the false-positive rate is 3% per patient and 2.1% per disc. In all subjects without low back pain, but excluding those who underwent prior discectomy and 2 subjects (because of incomplete data sets) with somatization disorder, the false-positive rate is 6% per disc and overall specificity is 94% (95% CI, 0.88–0.98). This specificity is significantly improved over other studies because standardized ISIS/IASP diagnostic criteria were used. With these findings, lumbar discography remains a valid test to confirm or refute the hypothesis that a specific disc is the etiology of a patient’s low back pain.

In 2006, Carragee and colleagues measured the success of lumbar surgical fusion in subjects selected by positive lumbar discography in an effort to determine the positive predictive value of discography. The study followed 32 patients with single-level, low-pressure positive discography 2 years post-fusion. They were matched with 34 patients who underwent fusion for unstable spondylolisthesis (considered the best-case scenario for surgical fusion results and therefore chosen as the gold standard). Only 27% of the patients with positive discography had highly effective surgical success versus 72% of the spondylolisthesis cohort. The investigators reported a best-case positive predictive value for discography of 50% to 60%. The strengths of the study included stringent exclusion criteria for the discography group, including worker’s compensation, abnormal psychometrics, other chronic pain, greater than 12 months of disability, and greater than 25% relief with zygapophysial or sacroiliac joint injections, thereby creating an uncomplicated population for discography with little in the way of confounding factors.

However, a major weakness of the study was the difference in inclusion criteria between the 2 groups. Although the discography group included chronic low back pain without sciatica, the spondylolisthesis group allowed patients with or without sciatica. Within the spondylolisthesis group, 6 patients were noted to have positive electromyography (EMG), 4 of whom had denervation. It is not clear whether all patients underwent EMG, and it is possible that additional subjects could have had a positive EMG. With this discrepancy, it is not truly a matched cohort. The difference in surgical outcomes for those with radicular pain versus those without is widely accepted, and similar outcomes would not be expected.

Nonetheless, the finding that only 27% of patients with positive discography had highly effective surgical success is noteworthy. Discography was done with adequate diagnostic criteria: pain greater than or equal to 6/10, threshold less than or equal to 20 psi above opening pressure, annular fissure extending to the posterior annulus, and a control disc with less than 2/10 pain with pressurization up to 100 psi above opening pressure. The criteria for highly effective surgical success, however, were stringent and included low back pain less than or equal to 2/10, Oswestry disability index (ODI) less than or equal to 15, full return to occupational duties, no narcotics, and no daily analgesics. These are not the typical criteria used to judge surgical success. A more favorable outcome of 43.2% success for patients who underwent discography was noted with the less stringent minimal acceptable outcome as judged by patients before interventions. These are not exemplary outcomes but are favorable compared with surgical success rates for low back pain in other studies. Fritzell and colleagues found that only 16% of patients reported excellent results, and Slosar and colleagues concluded that only 10% of patients believed that surgery met their expectations. Discography may improve surgical results, especially in patients without confounding factors. Incomplete surgical success with discography may be caused by multiple factors aside from a low positive predictive value of the test, including nonmechanical pain emanating from the disc and surgical limitations for the treatment of axial low back pain.