This systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA). A prior protocol for this review was registered with PROSPERO (http://www.crd.york.ac.uk/PROSPERO, no. CRD 42021246403).

An electronic search of the PubMed, Embase, and Cochrane databases for studies published between January 2001 and March 2021 was performed. The search terms “cervicogenic headache,” “secondary headache,” “interventions,” “nerve blocks,” “occipital nerve block,” “zygapophyseal joint injection,” “median branch block,” “pulsed radiofrequency,” and “radiofrequency neurotomy” were combined in different ways to search the databases. Two independent researchers searched the available literature and collected all the relevant articles. All the selected abstracts were reviewed by another researcher. A well-drafted PICOS framework was used to conduct the study (Table 1).

After the electronic databases were searched and the duplicates were removed, 6,484 articles were retrieved. Articles in languages other than English, animal studies, and abstract-only articles were not included. We also excluded literature reviews, systematic reviews, editorials, case reports, case series, non-scientific commentaries, reports, and news articles from this analysis. The full text of the article was obtained if the title or abstract discussed interventions for CeH management. If there were other pathologies, such as cranial masses, head injury, or any intracranial surgeries, the article was excluded. A total of 130 full-text articles were reviewed for eligibility. The references of the selected articles were also searched for additional studies matching the inclusion criteria. A total of 23 articles were included in the final analysis (Fig. 1).

The methodological quality of the included studies was assessed using the “risk of bias” of the Review Manager Software version 5.4 (The Cochrane 14 Collaboration, UK). Two authors independently assessed the quality of each study, and disagreements were resolved through discussion. Seven categories, which included random sequence generation and allocation concealment to detect selection bias, blinding of the participants for performance bias, blinding of the outcome assessor for detection bias, incomplete outcome data for attrition bias, selective reporting for reporting bias, and other bias, were rated as “high,” “low,” or “unclear” to assess the internal validity of each study (Figs. 2 and 3).

The 23 included articles were fully reviewed by two reviewers who independently extracted and summarized the data in a table under the following headings: 1) author name, 2) year of publication, 3) type of study, 4) population, 5) intervention(s), 6) results, and 7) conclusion. Due to the lack of homogenous data and high-quality randomized controlled trials, only a systematic review could be performed.

Due to the convergence of the upper cervical and trigeminal sensory pathways, the bidirectional referral of nociceptive sensations between the neck and trigeminal receptive fields of the head and face leads to the referral of CeH pain from a cervical source to the forehead, temple, or orbit [6]. This forms the background for managing CeH through blocking the GON. Anesthetic blocks of the LON and facial nerve have also been found to be effective [9]. Inan et al. [14] compared the effect of GON blocks to C2/C3 spinal rami blocks in 28 patients with CeH and concluded that both blocks are equally effective. No significant difference was observed between the two groups in terms of pain frequency or degree of pain, except for pain frequency in the first week following the first therapeutic block, which was significantly reduced in the C2/C3 group. Another study by Naja et al. [9] evaluated 50 patients with CeH who received GON and LON blocks with or without facial nerve blocks. The anesthetic block group, which received a mixture of lidocaine, bupivacaine, epinephrine, fentanyl, and clonidine, was compared with the placebo group (normal saline) and a statistically significant improvement in pain intensity, frequency, and duration as well as a decrease in analgesic use were observed at 2 weeks in the block group compared to the placebo group. Lauretti et al. [15] evaluated 30 patients with unilateral CeH who underwent GON blocks using the classic technique (1 cm below the level of the superior nuchal line, just medial to the pulsation of the occipital artery). The visual analog scale (VAS), which is a tool used to evaluate pain using a 10 cm line with no marking that ranges from no pain (0) to worst possible pain (10), was used. Those with a score > 3 were randomly allocated into 3 groups (n = 10) who underwent GON blocks with 5, 10, or 15 ml of volume using the suboccipital compartmental technique. A significant decrease in the pain score and rescue analgesic consumption and an improved quality of life were seen in all subcompartmental groups for 24 weeks compared to only 2 weeks with the classic technique. Pingree et al. [16] evaluated 14 patients who underwent ultrasound-guided GON blocks at the C2 level and reported a successful block in 86% of patients 30 min post-injection. A significant decrease in the mean numerical rating scale (NRS) score, which is an 11-point scale ranging from 0 “no pain” to 10 “worst pain,” was observed at 30 minutes, 2 weeks, and 4 weeks compared to baseline. Although the sample size was very small, a significant reduction in the pain score was observed. Ertem and Yılmaz [17] retrospectively evaluated 21 patients with CeH who underwent at least three GON blocks and attended at least three follow-up appointments. A significant reduction in pain scores was seen at 3 months post-treatment. Some other previous studies found an overall pain reduction of more than 50% [14-17] or nearly 50% [9] in the mean NRS or VAS score following occipital nerve blocks to treat CeH, with a short duration of pain relief usually lasting for a few weeks [9,14-16]. Repeat injections may be effective for sustained pain relief [17].

The beneficial effect of facet joint injections for the treatment of CeH has been reported in a few studies. Slipman et al. [18] reviewed 18 patients with unremitting headaches after flexion/extension injuries associated with tenderness over the upper cervical zygapophyseal joint who underwent a C2–C3 zygapophyseal joint injection. A second injection was administered after 2 weeks if pain relief was < 90%. Although the average decrease in the VAS score (from 8.2 pre-injection to 5.5 post-injection) was not significant, the headache frequency, response to analgesics, and employment status improved significantly. Another retrospective chart review of 31 patients with refractory CeH who underwent C1–C2 and C2–C3 facet joint injections and C2 and C3 spinal rami blocks was conducted by Zhou et al. [19]. In that study, 28 patients showed a > 50% reduction in pain for an average duration of 21.7 days. A significant decrease in the mean pain intensity was observed immediately after injection. The study outcomes suggested that C1–C2, C2–C3 facet joint dysfunction and subsequent irritation of the spinal rami at C2 or C3 may contribute to CeH development and that steroid injections reduce spinal nerve root irritation and thus improve CeH. Despite the small sample size in the above two studies, the suggested contribution of upper cervical arthropathy in the generation of CeH and the effectiveness of both cervical facet joint injections and C2–C3 spinal rami blocks for pain relief were notable. No treatment-related complications were observed.

Narouze and Provenzano [20] conducted a retrospective chart review of 32 patients with CeH suggestive of AA joint pain who underwent AA joint intraarticular injections. Complete pain relief was observed in 15 patients, and 23 patients experienced a ≥ 50% reduction in pain. The mean pain score decreased significantly from pre-procedure to immediate post-procedure and at 1 month and 3 months, but not at 6 months. Therefore, this study showed the short-term pain relief provided by intraarticular AA steroid injections. However, there was not sufficient data to determine its long-term effects.

A deep cervical plexus block can be useful for refractory cases of CeH, as pain often occurs over the C2 or C3 spinal nerve root distribution. Goldberg et al. [21] demonstrated a significant reduction in pain scores immediately after receiving a deep cervical plexus block at the C2/C3 level in 39 patients with CeH. While some patients experienced effective pain relief for 3 months, pain scores had returned to baseline levels by 6 months. The injection effectiveness was rated at 42% effective for all first injections and 40% effective for the last injection. Wan et al. [22] evaluated 56 patients who underwent either an ultrasound-guided or fluoroscopic-guided deep cervical plexus block along the C2 and/or C3 transverse process and reported a significant decrease in pain intensity (NRS) in both groups at 2, 12, and 24 weeks post-injection, with no significant differences observed between the groups. However, the small sample size and lack of double-blinding limited the strength of these findings and a clear understanding of the role of this treatment for CeH management [21,22].

He et al. [23] evaluated 37 patients with CeH treated with continuous cervical epidural block using lidocaine, dexamethasone, and saline (5 ml/h) for 3–4 weeks and triamcinolone 5 mg once a week for 3–4 weeks, and found it to be effective for at least 6 months. However, further research is needed to elucidate the mechanism and validate this outcome.

For CeH patients who fail the interventions mentioned above or for those with severe or refractory CeH, radiofrequency lesioning may be an option. The targeted pain generators are the facet joint and its nerve supply (medial branch of the spinal dorsal rami), the third occipital nerve (branch of the dorsal rami of the C3 spinal nerve, supplying the C2–C3 facet joint), the GON, lateral C1–C2 joint, and C2 DRG.

Stovner et al. [25] evaluated RFA of the medial branch of the C2–C6 facet joints ipsilateral to the pain in 12 patients and compared them to those receiving sham treatment. A slight improvement was noted at 3 months, but after this time and over a duration of 2 years, no statistically significant differences were observed. Haspeslagh et al. [26] evaluated 15 patients who received RFA of the C3–C6 facet joints and the DRG and compared them with a local anesthetic block of the GON. No statistically significant difference in pain scores was seen, suggesting that RFA of the cervical facet joint was no better at reducing pain than local infiltration of the GON. Therefore, both of the above studies showed that RFA provided no significant benefit.

Govind et al. [27] described the effect of RFA of the third occipital nerve for the treatment of referred pain from C2–C3 facet joints in 49 patients and reported successful outcomes in 88% of the patients with pain-free intervals lasting for approximately 297 days. Fourteen patients underwent repeated procedures, 86% of which (12 patients) experienced pain relief for the next 217 days. The study concluded that third occipital nerve RFA was effective for pain relief and repeat ablations can prolong its efficacy. Lee et al. [28] evaluated 30 patients with CeH who underwent RF neurotomy of the cervical facet joints after positive diagnostic blocks and found substantial pain relief over a 12-month follow-up period. Another study by Park et al. [29] evaluated 11 patients with CeH who underwent RFA of the medial branches of the lower cervical nerves (C4–C7) and reported a significant decrease in VAS scores at 6 months (from 8.1 ± 1.1 to 2.7 ± 1.3). The study also concluded that lower cervical disorders may play a role in the genesis of CeH.

PRF of the GON was evaluated by Gabrhelik et al. [30] and compared with the GON block (using a local anesthetic and steroid). A significant decrease in VAS scores and analgesic consumption were observed in both groups at 3 months, with long-term pain control (at 9 months) in the PRF group. Halim et al. [31] evaluated 86 patients with CeH who underwent lateral C1–C2 joint PRF. The percentage of patients with pain relief ≥ 50% at 2 months, 6 months, and 1 year was 50% (43/86), 50% (43/86), and 44.2% (38/86), respectively. Long-term pain relief at 6 months and 1 year was predicted by ≥ 50 % pain relief at 2 months. The study concluded that PRF of the lateral C1–C2 joint was effective for pain relief in refractory CeH; however, outcome validation is limited by its retrospective nature and short follow-up period.

Hamer and Purath [32] evaluated 40 patients who received a bilateral RFA of the C2 DRG and were followed up for 6 months to one year. Pain relief was 100% in 35% of patients and ≥ 80% in 70% of patients. The mean duration of pain relief was 22.35 weeks. A total of 92.5% of patients reported satisfaction with the procedure and were willing to undergo the procedure again if the symptoms returned. The complication rate was 12–13%. Another study by Hamer and Purath [33] evaluated 23 patients with CeH who needed a repeat RFA of the C2 DRG and reported that the repeat RFA was effective. Compared to the first intervention, the repeat intervention showed either similar (in 59% of patients) or better (in 32% patients) effectiveness. Li and Feng [34] retrospectively evaluated 87 patients who underwent PRF of the C2 DRG and epidural steroid injection (ESI) and compared them with 52 patients who underwent only ESI. A significant reduction in the median pain score was observed in both groups at the 2-year follow-up. A significantly lower VAS score, pain attack frequency, analgesic use, total pain score, and improved quality of life were observed in the PRF + ESI group than in the ESI group. Median pain relief lasted 8 months in the PRF + ESI group and 4 months in the ESI group, suggesting that the combination of PRF of the C2 DRG and ESI may be an effective and safe option for CeH. Lee et al. [35] evaluated 45 patients who underwent C2 DRG PRF after CeH recurrence 24 h after receiving a diagnostic C2 DRG block. A ≥ 50% reduction in pain was observed in 40% of patients (success group). Significantly more patients in the success group than in the failure group showed a positive diagnostic block. The study concluded that C2 DRG PRF is an effective treatment, especially for patients with definite pain reduction after the diagnostic C2 DRG block.

Among the upper three cervical spinal nerves, the C2 spinal nerve is more susceptible to injury [38]. The ventral rami of C2 innervates the AA joint, and also gives rise to LON. The GON arises from the medial aspect of the dorsal rami of the C2 spinal nerve. The C2 DRG, therefore, may be an effective target for PRF; however, evidence is limited due to the lack of randomized trials.

To achieve a long-lasting analgesic effect, freezing destruction of nerve conduction has been attempted for refractory cases of CeH. Kvarstein et al. [36] evaluated the clinical efficacy of occipital cryoneurolysis and compared it with local anesthetic and steroid injections. Despite a significant reduction in pain scores, pain intensity gradually increased after 6–7 weeks but had not returned to baseline by 18 weeks in both groups. No or minimal improvement was seen in health-related quality of life and psychological distress in both groups. After 18 weeks, majority of patients (74%) reported much or moderately improved global status, 55% of patients reported much or moderately improved headache intensity and 29% reported improved neck movement in cryoneurolysis group. These results indicate that the role of occipital cryoneurolysis in treating CeH may be questionable; however, further studies with larger sample sizes are required.

In this review, various interventions targeting different pain generators for the management of CeH have been described. Occipital nerve blocks (GON, LON) showed only limited evidence, as most of the studies were non-controlled and yielded only transient benefits. Facet joint intraarticular injections, anesthetic blocks of the upper cervical spinal nerves, AA joint injections, deep cervical plexus blocks, and cervical epidural blocks may be effective treatments, they have generally only been shown to provide short-term relief, with limited or no long-term benefits. Further studies are needed to consolidate the role of freezing destruction of pain-generating fibers using cryoneurolysis. Radiofrequency lesioning may be preferable over other interventions because of its long duration of effect, better efficacy, and fewer side effects. Conventional RFA is neurodestructive and is associated with high complication rates, such as neuritis or deafferentation pain, which is not seen with PRF [32,33]. PRF, therefore, could be considered the preferred interventional approach for CeH management, given its better safety profile.

This systematic review had several limitations. First, most of the included studies were not RCTs. Second, the structure, inclusion/exclusion criteria, and outcomes assessed among the included studies were heterogenous. Third, most of the included studies had a small sample size and short follow-up period. Additionally, there were flaws and inconsistencies in the design of both randomized and nonrandomized trials. Although a few studies showed promising outcomes of a particular intervention for the management of CeH, carefully designed, high-quality, large, prospective, randomized trials are needed to investigate the long-term benefits of various interventions for effectively managing CeH.

In conclusion, based on the available literature, occipital nerve (GON, LON) blocks, cervical facet intraarticular injections, AA joint injections, deep cervical plexus blocks, and cervical epidural steroid injections may be reasonable options for CeH treatment. Radiofrequency lesioning was found to be better with long-term positive outcomes, and pulsed therapy had better safety. However, our review revealed only limited evidence, and more RCTs are needed to provide more concrete evidence and to establish the relative efficacy of the various available interventions discussed for the management of CeH.