The prevalence of lumbar zygapophyseal joint (Z-joint) pain has been reported to be approximately 6% in a primary care setting (94). However, in a tertiary spine center, lumbar Z-joint pain has been reported to range from 15% in younger individuals (96) to 40% in older populations (98) with chronic low back pain. In individuals with chronic neck pain after a whiplash injury, Z-joint pain occurred in approximately 50% of patients (99,100). More recent literature examining 500 patients has reported a prevalence of 55% cervical facet joint pain, 42% thoracic facet joint pain, and 31% lumbar Z-joint pain in patients with chronic spine pain identified with a double block model as described later (95). Lumbar facet joint pain has been reported in 16% of patients with chronic postsurgical lumbar spine pain (97). Studies of a larger population using a dual anesthetic block paradigm may be helpful in further identifying the prevalence in the general population for both acute and chronic low back pain.


Z-joints are pairs of small synovial joints in the posterior aspect of the spine, formed when the inferior articular process of one vertebra articulates with the superior articular process of the subjacent vertebra. Each lumbar Z-joint has a 1 to 2 mL capacity (101). Cervical and thoracic Z-joints can hold volumes of less than 1 mL (103) and 0.5 to 0.6 mL (104,105), respectively.

The Z-joint at C2-3 is innervated by the third occipital nerve from the superficial medial branch of the C3 dorsal ramus. Below the C2-3 Z-joint, each cervical Z-joint is innervated by the medial branches from the level above and below. The medial branches wrap around the articular pillar transverse processes. The C7 medial branch is located higher, as it is pushed up by the base of the transverse process. The joints between C0-1 (atlanto-occipital) and C1-2 (atlantoaxial) joints are technically not Z-joints, due to their anterior location. They are innervated by the anterior rami of C1 and C2, respectively. Significant variability in the location of the medial branches has been reported, particularly with regard to the anatomy in the cervical spine (102).

The medial branches innervating the thoracic Z-joints have a different course in relation to the transverse processes (105). The thoracic medial branches instead wrap around the junction between the transverse process and the superior articular process. As in the lumbar spine, they often exit in the middle portion of the intertransverse space, and they typically cross the superolateral corners of the transverse processes and then pass medially and inferiorly across the posterior surfaces of the transverse processes. Furthermore, at mid-thoracic levels (T5-8), the inflection occurs at a point superior to the superolateral corner of the transverse process. Therefore, the superolateral corners of the transverse processes are generally the more accurate target points for diagnostic blockade or radiofrequency denervation of the thoracic medial branches.

In the lumbar spine, each Z-joint is innervated by two medial branches of dorsal rami; one from the same level and the other from the level above (106). For example, the Z-joint at L4-5 is innervated by the medial braches of the L4 and L3 dorsal rami. The medial branches travel along the junction of transverse process and superior articular process. After exiting the mamilloaccessory notch covered by the mamilloaccessory ligament, the medial branches send branches to innervate the same level of the Z-joint and the Z-joint below. However, the L5 dorsal ramus crosses the groove between the sacral ala and the superior articular process of the sacrum. The L5 medial branch to the L5-S1 Z-joint is very short because it does not branch out until it comes just under the L5-S1 Z-joint. It is important to remember that the lumbar medial branch rests on the subjacent level rather than the same level of the transverse process. For example, the L3 medial branch rests on the L4 transverse process rather than on the L3 transverse process (Figs. 68-12 and 68-13). Therefore, to denervate the L4-5 Z-joint, it is necessary to target the L3 and L4 medial branches, which rest on the junction of L4 and L5 transverse processes and superior articular processes, respectively. One exception is the blockade or denervation of the L5 medial branch. Due to the short branch below the Z-joint, denervation of the L5 medial branch can only be performed by destruction of the L5 dorsal ramus proper at the groove between the sacral ala and the superior articular process.

FIGURE 68-12. AP view of left L5-S1 Z-joint (facet) arthrogram. Notice the contrast pooling in the superior and inferior articular recesses. This 33-year-old man injured his low back lifting a 70-lb piece of metal overhead. Unable to sustain the weight, he extended his low back obliquely to the right side to drop the metal. He felt a “pop” in his left low back and sustained persistent significant left low back pain. Several minutes after this left L5-S1 intra-articular Z-joint injection with 0.5 mL of 1% lidocaine, his low back pain was completely abolished.

FIGURE 68-13. AP lumbar MBB at the right L2, L3 and L4 vertebral levels, demonstrating needle tips and contrast injected at the base of SAPs and the transverse processes.

Pathophysiology of Z-joint Pain

The Z-joint is a well-innervated structure. The Z-joint capsule contains both nociceptive and mechanosensitive receptors (101,106–108). Immunocytochemical studies have demonstrated that the Z-joint capsule or synovial folds contain substance P, calcitonin gene-related product, vasoactive intestinal polypeptide (VIP) and neuropeptide, and tyrosine hydroxylase (101,107,108). Ostensibly, the Z-joint capsule is a potential pain generator if it is injured. In addition, like any synovial joints, Z-joints can develop synovitis under certain circumstances.

Typical Z-joint pathology often derives from pathologic mechanical stress or inflammation. In the lumbar spine, the sagittal orientation of upper lumbar Z-joints and the relative lateral oblique orientation of lower lumbar Z-joints make them vulnerable to injury from extension and torsion forces. In the cervical spine, both human and animal studies of whiplash injuries have demonstrated evidence of multiple pathologic findings, including Z-joint capsule tearing, intraarticular hemorrhage, articular cartilage, muscle injury, and subchondral bone fracture (110). These pathologic changes can serve as a basis of pathologic nociception and neck pain. However, none of these pathologies can be detected with conventional imaging studies (110).

Patients with lumbar Z-joint pain often have more pain in the lateral aspect of the low back unilaterally or bilaterally but not centrally (111). Pain can often be made worse with oblique extension of the lumbar spine (111,112). Pain is often worse after overnight rest or inactivity. Local tenderness and “muscle spasm” over the involved Z-joint are frequently noted. Although Z-joint pain can present with referred pain, groin pain or thigh pain, neural tension signs are negative and there are generally no neurological deficits, though recently a study where a rat model was used to induce Z-joint inflammation has documented associated radiculopathy as a possible sequela (109). Patients with cervical Z-joint pain experience pain located in the axial cervical areas. However, several studies have failed to demonstrate any pathognomonic physical exam findings of Z-joint pain (111,112). Imaging studies are also unable to confirm or refute the diagnosis of Z-joint pain (110,113,114).


Given the lack of a clinical diagnostic gold standard, clinicians have used regional anesthesia to identify Z-joint pain. Abolishment of low back pain after anesthetic Z-joint injection or medial branch block confirms the Z-joint as the pain generator (Figs. 68-14 and 68-15). Z-joint injection is indicated in patients with acute back and neck pain of suspected Z-joint origin, with no evidence of neurologic deficits, and whose pain pattern resembles that evoked in normal volunteers upon stimulation of their Z-joints. However, since the majority of acute back and neck pain, including Z-joint pain, will resolve in several weeks, the injection is often reserved for individuals with severe pain that has failed to respond to 4 to 6 weeks of conservative therapy including oral analgesics, directed physical therapy, and relative rest. Fortunately, injection can be performed earlier if pain is inhibiting therapy progress.

FIGURE 68-14. Oblique view of left L4-5 Z-joint injection showing the needle tip inside the Z-joint space.

FIGURE 68-15. Oblique view demonstrating blocks of right L1, L2 and L3 medial branches at the right L2, L3 and L4 pedicles, viewed as the “eyes of the Scotties dogs”, at the base of the SAPs and transverse process where the medial branches are located.


As is true of all spinal injection procedures, Z-joint injections are contraindicated in individuals with the following conditions: infection, bleeding diathesis, pregnancy (for use of fluoroscopy), allergy to the medications to be injected (contrast medium, local anesthetics, corticosteroid), and unstable medical conditions such as unstable angina or poorly controlled hypertension or diabetes mellitus. Injections are elective procedures, so an effort should be made to properly select patients so as to ensure safety and optimize outcome.


For lumbar Z-joint injection, the patient is placed in a prone position with a pillow under the abdomen to distract the Z-joint. After sterile preparation and skin draping, intermittent fluoroscopic views are used to identify the level of the Z-joint. For the L5-S1 level, the fluoroscope is tilted in a caudad direction to accommodate the lumbar lordosis, and rotated ipsilaterally until the joint space first comes into view, which is the posterior opening of the Z-joint, that is, the needle entry point. Although further rotating the image intensifier will more clearly visualize the Z-joint space, the visualized joint space at this angle is actually the middle or anterior Z-joint opening, not the posterior opening which is the injectionist’s target. For upper lumbar Z-joint injections, it may require less oblique rotation to better visualize the Z-joint space. The needle entry site is marked with a metal instrument. The skin and the underlying tissues are infiltrated with 1% lidocaine. For diagnostic Z-joint injections, the underlying tissues should not be infiltrated with anesthetic in order to maximize injection specificity. A 22- or 25-gauge 3.5 in. spinal needle is then inserted at the anesthetized site and directed toward either the superior or inferior articular processes of the targeted facet joint using a “tunnel view” technique by which the entire needle shaft is paralleled to the fluoroscopy beam in such a way that the needle hub appears as a dot. Once the needle contacts the bone, the tip of the needle is then “walked” off into the Z-joint space. Occasionally, due to the osteoarthritic changes, the needle cannot gain entry. The needle can then be directed into the inferior articular recess just off the lower margin of the articular processes. Once the needle is felt to be in the articular space or to have penetrated the Z-joint capsule, 0.2 to 0.3 mL of the water soluble and nonionic contrast is injected to outline the Z-joint and to confirm that the needle tip is not located inside the vascular or epidural space. For therapeutic benefit, 1.0 mL of a mixed solution containing 20 mg of methylprednisolone acetate and 1% lidocaine is slowly injected into the Z-joint.

Medial Branch Block

For lumbar medial branch blocks, skin preparation and C-arm fluoroscope positioning are essentially the same as for Z-joint injections. The difference is the target, which is the “Scottie dog’s eye” rather than “ear,” because the former represents the anatomical site where the medial branch is situated whereas the latter represents the articular processes forming the Z-joint. After local skin is infiltrated with 1% lidocaine, a 22- or 25-gauge 3.5 in. needle is inserted and directed until the needle touches the middle portion of the “Scottie dog’s eye.” The fluoroscope is then turned to the cross table or lateral view; the needle should be located at the site posterior to the spinal lamina. At the anterioposterior (AP) view, the needle tip should be at or slightly medial to the lateral margin of the superior articular process. At this point, the needle bevel should be turned to face medially, and 0.2 to 0.3 mL of the contrast is injected under real-time fluoroscopy to ensure that a vascular pattern or neuroforamial spread upon the dye injection has occurred. To ensure block specificity, less than 0.5 mL of either 2% lidocaine or 0.5% bupivacaine is used to block each medial branch.

For block of the L5 dorsal ramus, the fluoroscope should be rotated ipsilaterally oblique about 10 to 15 degrees. The needle is then directed in “tunnel view” down to the junction at the superior articular process of the S1 vertebra and the ala of the sacrum. In the AP view, the needle tip should be at the lateral margin of the S1 superior articular process.


Diagnostic Z-joint Injection or Medial Branch Block

The literature has demonstrated that Z-joint injections and medial branch blocks can be used for the diagnosis of Z-joint pain with comparable sensitivity and specificity (Fig. 68-16) (115–118). A medial branch block may be relatively easier to perform with less trauma to the Z-joint. In the cervical spine, medial branch blockade has been shown to be a valid technique for the diagnosis of Z-joint mediated pain (119). Overall, medial branch blockade has a reported 89% specificity and 11% false-negative rate (118).

FIGURE 68-16. Lateral view of cervical Z-joint injection, with the arthrogram demonstrating the typical dumbbell shape.

Since the degree of pain relief is a patient’s subjective response, Z-joint injection or medial branch blockade is susceptible to a placebo effect. Other possible causes of falsepositive studies include inadvertent anesthetic spread to pain generators outside of the Z-joint. Research has shown falsepositive rates of 27% to 38% for lumbar blocks, 27% to 63% for cervical blocks, 55% for thoracic blocks, and a 32% placebo effect (95,112,120,121). To minimize false-positive rates, various clinicians have advocated evaluating the response to injecting anesthetics of varying anesthetic durations as well as possibly performing a control injection with saline placebo. A true-positive response is considered to be pain relief lasting for 1 to 2 hours with 2% lidocaine, and 3 to 4 hours with 0.5% bupivacaine, but no effect with saline. However, such a triple block scheme requires three separate procedures and is thus time consuming and costly. Ethical issues arise if a patient accepts the procedural risks and financial costs expecting a therapeutic injection, only to receive a placebo injection. A compromise is to perform a “double block paradigm” using two local anesthetics with different durations of action, one on each of two separate occasions. If each of the two injections relieves pain for the duration expected for the anesthetic used, Z-joint pain can be reliably diagnosed. Published studies have validated this dual blockade paradigm, using comparative local anesthetics for medial branch blocks to anesthetize Z-joints, which constitutes a viable alternative to normal saline controls (100,122).

Potential causes of false-negative responses include venous uptake of the injected anesthetic, aberrant innervation of the target Z-joint, and other technical issues. Since both intravascular injection and epidural spread of injectate can reduce the injection specificity, fluoroscopy plays an important role in maximizing outcomes.

If injection is being performed solely for diagnostic, not therapeutic, purposes, then only local anesthetic should be injected, without corticosteroid. Regarding the placebocontrolled blocks, a positive diagnosis was recorded only if the patient’s pain was completely and reproducibly relieved by each of the local anesthetic but not via the normal saline.

As stated above, double block with local anesthetics of different durations on two separate occasions can reduce (although not eliminate) the false-positive response and placebo effect (122). What constitutes a positive diagnostic response remains the physician’s subjective judgment. For a true concordant response with comparative blocks (100), the patient should obtain complete pain relief with a duration that is consistent with each particular anesthetic solution, though recent published placebo-controlled studies have used a criterion of at least 80% pain relief. This concordant response criterion for identifying Z-joint pain yields a good specificity of 88%, but only marginal sensitivity of 54%, thus suggesting significant false-negatives (100). Expanding the comparative blocks diagnostic criteria to include all patients with reproducible relief, irrespective of duration, increases sensitivity to 100% but lowers specificity to 65% (100). Whether the criterion should be structured to more aggressively prevent false-positive responses versus to prevent false-negative responses may depend on what subsequent treatment will be based on the determination. For additional treatments that can irreversibly alter the patient’s anatomy (e.g., via surgery or radiofrequency ablation [RFA]), preventing false-positive responses becomes increasingly important (100).

Using the dual block technique in patients with chronic low back pain, lumbar Z-joint pain was demonstrated in 15% of younger patients and 40% of older patients (96,98). In chronic neck pain after whiplash injury from motor vehicle injury, Z-joint pain occurred in 54% of patients (99). Another study of patients with chronic neck pain for more than 6 months demonstrated that the prevalence of Z-joint pain was 36% (123). The C2-3 and C5-6 Z-joints were found to be the most common symptomatic joints (99,123). The C2-3 joint was found to be a pain generator in 50% of patients with chronic cervicogenic headache after a whiplash injury (99). It is interesting that using the intra-articular and/or medial branch block, studies have demonstrated that the coexistence of Z-joint pain and discogenic pain in lumbar region is only 4%, while in the cervical spine it is 40% (124,125).

Caution should be exercised using provocation of pain as a sole diagnostic criterion for patient undergoing a diagnostic lumbar Z-joint injection. One study has demonstrated no significant correlation between pain provocation during Z-joint injection and the analgesic response (126).

Therapeutic Effect

The therapeutic benefit of Z-joint injection with corticosteroids remains controversial (127). A past study has suggested positive efficacy of C2-3 Z-joint corticosteroid injections for cervicogenic headache after a whiplash injury (128). Controlled studies have failed to demonstrate such efficacy although the studies had various design flaws (Table 68-5), including the use of saline as a placebo, patient selection bias without using a double block technique, and injection in isolation of other treatments. Selection of patients without using the double block paradigm may potentially include those patients with false-positive results. Finally, injection is not recommended to be used in isolation of other treatments. Rather, when tolerable, patients should be involved in a directed therapy program if they experience significant pain reduction (therapeutic window) after the injection.

A recently published systematic review of the literature from 2004 to 2006 has demonstrated strong evidence for diagnostic injections in the cervical and lumbar spine and moderate evidence for diagnostic injections in the thoracic spine (115). Another published review from 2007 has noted “limited” evidence for intra-articular cervical facet joint injections and “moderate” evidence for intra-articular lumbar Z-joint injections for pain relief (129). It is interesting to note, however, that this study did conclude that there actually was moderate evidence for short-and long-term pain relief from medial branch blocks.


Potential complications of fluoroscopic-guided, contrastenhanced lumbar Z-joint injections or medial branch blocks are rare. The most common post-procedural problem is transient pain at the injection site. However, there have been rare case reports of meningitis, inadvertent spinal anesthesia, and infection after Z-joint injections (137,138). Recurrent back and neck pain after radiofrequency neurotomy may be due to incomplete ablation or medial branch regeneration (139). Since the dorsal root ganglia are left intact, the ablated medial branches may regenerate. Because radio frequency neurotomy does not permanently denervate the Z-joints or cause inherent instability in the spine, the concern of Charcot facet joint development has little ground and no such cases have been published (140). Also, because the DRG remains intact, deafferentation pain should not occur. Side effects from radiofrequency neurotomy are rare when the procedure is performed correctly but include local pain and infection.


Source:  Physical Medicine and Rehabilitation - Principles and Practice

See also