The initial search identified 7639 studies from six databases. After excluding 1553 studies due to duplication, 122 studies were screened based on titles and abstracts. Ultimately, 18 articles met the inclusion criteria. The process of study selection is displayed in Fig. 1.

Fig. 1Characteristics of Included Studies

Eighteen studies [21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38] were included. The sample size, interventions, control, follow-up duration, and outcomes are summarized in Table 1. Interventions included manual acupuncture, dry acupuncture, warm needle moxibustion, and press needles. Seven studies combined acupuncture with manipulation, exercise, or other treatments [25, 26, 29, 32, 36,37,38]. Control group (CG) can be divided into three categories: sham acupuncture (minimal acupuncture at a non-acupoint or unrelated point, or blunt needle without actually piercing the skin), no treatment, and active treatment (such as TENS, traction treatment, self-exercise, and massage). Pain was assessed using the NRS, VAS, and MPQ; neck dysfunction was assessed using the NDI, NPQ, and NPDS; and quality of life was assessed using SF-36.

Table 1 Characteristics of included studiesRisk of Bias Assessment

In our meta-analysis of 18 studies, rigorous randomization resulted in a low risk in this domain. However, 15 studies [21,22,23,24,25,26, 28,29,30, 32,33,34, 36,37,38] lacked detailed descriptions of intervention deviation and adequate blinding, posing a latent risk of bias. Regarding missing outcome data, 11 studies [21, 22, 24, 27,28,29,30,31,32, 34, 36] meticulously reported participant withdrawals and final analysis inclusions, substantiating a low bias risk, while seven studies [23, 25, 26, 33, 35, 37] failed to provide a study flowchart, thereby obscuring the precise number of patients lost to follow-up and thus leading to a high risk of bias. Six studies [22,23,24, 36,37,38] showed unclear selective reporting bias owing to the lack of protocol registration, whereas the remaining 12 [21, 25,26,27,28,29,30,31,32,33,34,35] had a low bias risk in this domain. Overall and individual trial bias risks are both presented in Figs. 2 and 3.

Fig. 2Fig. 3

Risk of Bias of included studies

Acupuncture Versus Sham AcupuncturePain Intensity

Two studies evaluated the prolonged effects of acupuncture versus sham acupuncture on CNP by assessing the VAS scores at 3, 6, and 12 months post-treatment. Liang et al. [27] reported no statistically significant difference at 3 months (MD: − 0.12; 95% CI − 0.06 to 0.36; p = 0.63) (Supplementary Fig. 1). Similarly, Gattie et al. [31] observed no statistically significant difference at 6 months (MD: 0.01; 95% CI − 1.16 to 1.18; p = 0.99) (Supplementary Fig. 2), a trend that continued at 12 months (MD: − 0.42; 95% CI − 1.55 to 0.71; p = 0.47) (Supplementary Fig. 3).

Disability

Three studies compared acupuncture with sham acupuncture in functional improvements using the NDI and NPQ. At 6 months post-treatment, Gattie et al. [31] observed no statistically significant difference in NDI scores (MD: 2.40; 95% CI − 5.46 to 10.26; p = 0.55) (Supplementary Fig. 4). At 12 months, the difference in NDI scores remained insignificant (MD: − 0.11; 95% CI − 7.69 to 7.47; p = 0.98) (Supplementary Fig. 5). However, significant functional benefits were observed at 3 months post-treatment for NPQ. The synthesized data from Liang et al. [27] and Xu et al. [35] showed a notable improvement (MD: − 6.06; 95% CI − 8.20 to − 3.92; p < 0.01; I2 = 45%) (Fig. 4). Nevertheless, this improvement did not reach the minimal clinically important difference (MCID) criterion, defined as a 25% reduction in score from baseline [39].

Fig. 4

Forest plot of the mean difference in change of NPQ scores between acupuncture and sham acupuncture at the 3-month follow-up after intervention, compared to baseline, for CNP

Quality of Life

The study by Liang et al. [27] found that observations 3 months post-treatment revealed that acupuncture did not exhibit statistically significant improvements over sham acupuncture in both mental component summary (MCS) scores (MD: 5.36; 95% CI: − 1.53 to 12.25; p = 0.13) (Supplementary Fig. 6) and physical component summary (PCS) scores (MD: 1.02; 95% CI: − 6.20 to 8.24; p = 0.78) (Supplementary Fig. 7).

Acupuncture Versus No-Treatment Control

The study by Witt et al. [21] investigated the effect of acupuncture on the quality of life compared with no-treatment using the SF-36. At 3 months post-treatment, acupuncture provided a statistically significant improvement in MCS (MD: 3.20; 95% CI 1.30 to 5.10; p = 0.001) (Supplementary Fig. 8), meeting the MCID threshold of 2.5 [40]. However, this improvement diminished at 6 months post-treatment (MD: 0.90; 95% CI 0.36 to 1.44; p = 0.001) (Supplementary Fig. 9). The PCS scores initially improved significantly at 3 months post-treatment (MD: 4.60; 95% CI 1.86 to 7.34; p = 0.001) (Supplementary Fig. 10), exceeding the MCID of 2.6 [41], but the effect size decreased at 6 months post-treatment (MD: 0.60; 95% CI 0.24 to 0.96; p = 0.001) (Supplementary Fig. 11). The study by Witt et al. did not report on pain intensity or function.

Acupuncture Versus Active ControlPain Intensity

Five studies [22, 23, 25, 30, 34] compared the efficacy of acupuncture with active controls in the management of pain intensity, as measured by the VAS or NRS scoring systems. De et al., Irnich et al., and Valiente et al. compared acupuncture with active treatment in VAS or NRS scores three months after treatment (MD: − 0.17; 95% CI − 0.46 to 12; p = 0.24; I2 = 44%) (Fig. 5), while the studies by Franca et al. and Ilbuldu et al. made a similar comparison at six months (MD: − 1.27; 95% CI − 17.41 to 14.87; p = 0.88; I2 = 55%) (Fig. 6). No statistically significant differences were observed in either time frame.

Fig. 5

Forest plot comparing the standardized mean difference in pain score changes between acupuncture and active control at the 3-month follow-up after intervention, relative to baseline, for CNP

Fig. 6

Forest plot comparing the mean difference in pain score changes between acupuncture and active control at the 3-month follow-up after intervention, relative to baseline, for CNP

Disability

Six studies [24, 25, 28, 30, 33, 34] analyzed the long-term efficacy of acupuncture versus active control using the NDI and NPQ to measure disability and functional improvement. At 3 months post-treatment, NDI showed no significant difference (MD: − 0.29; 95% CI − 2.37 to 1.80; p = 0.79; I2 = 23%) (Fig. 7), but at 6 months, a significant difference was observed (MD: − 9.00; 95% CI − 14.06 to − 3.94; p = 0.0005) (Supplementary Fig. 12), meeting the MCID of 3 points [41]. NPQ results indicated significant improvement at 3 months post-treatment (MD: − 6.67; 95% CI − 9.42 to − 3.92; p < 0.01; I2 = 24%) (Fig. 8), persisting at 6 (MD: − 6.33; 95% CI − 9.22 to − 3.44; p < 0.0001) (Supplementary Fig. 13) and 12 months (MD: − 4.75; 95% CI − 7.86 to − 1.64; p = 0.003) (Supplementary Fig. 14).

Fig. 7

Forest plot of the mean difference in change of NDI scores between acupuncture and active control at the 3-month follow-up after intervention, compared to baseline, for CNP

Fig. 8

Forest plot of the mean difference in change of NPQ scores between acupuncture and active control at the 3-month follow-up after intervention, compared to baseline, for CNP

Acupuncture with Active Control Versus Active Control (add-on)Pain Intensity

Six studies rigorously evaluated the long-term effectiveness of combining acupuncture with active control interventions versus active control alone for treating CNP. These studies employed various pain assessment tools, including the VAS in five studies [25, 26, 29, 37, 38] to measure pain intensity, the NRS in one study [36], and the MPQ for a comprehensive evaluation of pain in another [38]. At the three-month post-treatment mark, a fixed-effects model analysis of VAS (0–10) and NRS scores (0–100) showed a standardized mean difference (SMD) favoring acupuncture combined with active control over active control alone (SMD: − 0.79; 95% CI − 1.13 to − 0.46; p < 0.01; I2 = 13%) (Fig. 9). At the six-month post-treatment mark, a random-effects model analysis of VAS scores (0–100) indicated that acupuncture with active control maintained a benefit over active control alone (MD: − 18.13; 95% CI − 30.18 to − 6.07; p < 0.01), although with a substantial heterogeneity (I2 = 92%) (Fig. 10). The MPQ reported a mean difference (MD: − 1.03; 95% CI − 2.38 to 0.32; p = 0.13) (Supplementary Fig. 15), meeting the MCID of 1 for MPQ, though it did not reach statistical significance.

Fig. 9

Forest plot comparing the standardized mean difference in pain score changes between acupuncture with active control and active control at the 3-month follow-up after intervention, relative to baseline, for CNP

Fig. 10

Forest plot comparing the mean difference in pain score changes between acupuncture with active control and active control at the 6-month follow-up after intervention, relative to baseline, for CNP

Disability

Five studies compared the efficacy of acupuncture combined with active treatment with active treatment alone. Two studies [32, 38 ] reported outcomes at 3 months post-treatment (MD: − 3.83; 95% CI − 9.22 to 1.57; p = 0.16; I2 = 74%) (Fig. 11), and four studies [25, 29, 32, 37] provided data at 6 months (MD: − 9.00; 95% CI − 19.22 to 1.22; p = 0.08; I2 = 98%) (Fig. 12), with no statistical significance.

Fig. 11

Forest plot of the mean difference in change of NDI scores between acupuncture with active control and active control at the 3-month follow-up after intervention, compared to baseline, for CNP

Fig. 12

Forest plot of the mean difference in change of NDI scores between acupuncture with active control and active control at the 6-month follow-up after intervention, compared to baseline, for CNP

Safety Assessment

Among the 18 included studies, eight did not report on adverse events, while 10 noted mild adverse events with no severe cases reported. The incidence rate of adverse events in the acupuncture group was 8.5% compared to 3% in the sham acupuncture group, 6.4% in the active treatment group, and 13.8% in the combined acupuncture and active treatment groups. The minor adverse effects associated with acupuncture reported in most studies [25, 29, 31, 32, 34] included minor local bleeding or hematoma after acupuncture and needling pain, with a likelihood of occurrence at 23.7%. Studies by Salter et al. [24] and Stieven et al. [32] indicate a 13.2% chance of transient dizziness, 5.9% of fatigue, and 20.6% of worsening symptoms associated with acupuncture. Liang et al. [27] reported that three patients in the acupuncture group fainted during treatment, and the symptoms were relieved entirely after lying down and drinking hot water.