This study was a double-blind, parallel randomized sham-controlled trial. Research methods and reporting were in accordance with the STRICTA (extension of CONSORT for acupuncture studies) guidelines.26 This study was prospectively registered (NCT03838224).

Forty participants with either traumatic or idiopathic neck pain were recruited from the metropolitan area of Valencia (Spain) through poster advertisement between March and June 2019. Volunteers were screened for potential eligibility via phone prior to the baseline assessment. Participants were included if they were 18–65 years old and had neck pain as defined by the International Association for the Study of Pain ≥3 months in the last year with a current pain intensity ≥30/100 mm on a visual analogue scale (VAS) and neck disability ≥10/50 on the Neck Disability Index (NDI).27,28 Participants also had to exhibit impaired cervical joint position error (JPE) in at least one direction of neck rotation determined by a cut-off value of 4.5°.29 Participants were excluded if they had a history of head trauma, cervical fracture, stenosis, surgery, or neurological signs/deficits suggesting nerve root compression as well as known or suspected vestibular pathology and vertigo or dizziness from ear or brain disorders or sensory nerve pathways (e.g. BPPV). Additionnaly, pregnancy, bleeding disorders, use of anticoagulant medication, or needle phobia as well as previous experience with DN treatment in the upper cervical region were reasons for exclusion.

The study was approved by the Ethical Committee at the University of Valencia, Spain (reference number H1542206264486)) and all procedures were performed in accordance with the Declaration of Helsinki. All participants provided written informed consent prior to participation. Data collection and treatment took place at the department of physical therapy, University of Valencia

Both interventions were performed by the same physical therapist with 3 years of DN experience. A 40 × 0.32 mm sterile needle with guided tube (AGU-A1041P, Agu-punt S.L., Spain) was used for DN whereas the Park sham device (DongBang, AcuPrime, UK), a blunted needle which appearance is similar to a DN needle, was used for sham needling.19,30 This sham needle has been previously validated and used as control in previous DN trials in patients with neck pain.31,32 For the DN intervention, the needle was inserted into the OCI muscle following a previously described and validated approach.19 The needle was inserted perpendicular to the skin at a mid-point between the spinous process of C2 and the transverse process of C1 (supplemental online material). Then, the needle formed an angle of 45° with the spinous process of C2 and the transverse process of C1 and was directed into a postero-anterior direction towards the anatomical location of the OCI with a slight inferior angle of 10° until reaching the C2 vertebra lamina. Once the first local twitch response was obtained, the needle was rapidly moved up and down within the muscle using the “fast-in and fast-out” technique described by Hong33 for a total of 12 insertions. As previous DN studies evaluating changes in muscle function, DN was performed bilaterally.23,34 A similar procedure to DN was adopted for the sham intervention to replicate an authentic clinical experience and maintain credibility and participants' blinding.35 Contextual clues associated with DN such as skin's cleaning, needle insertion, and manipulation (simulation in sham needling), and haemostatic compression after procedure were therefore identical in both interventions.

Cervical JPE was the primary outcome. Secondary sensorimotor related outcomes were cervical movement sense, the smooth pursuit neck torsion test (SPNT), and standing balance. Global cervical rotation range of motion (ROM), the flexion-rotation test (FRT) and pain intensity were also evaluated as secondary outcome measures. Measurements were taken at baseline, immediately post-intervention, and at one-week follow-up. Outcome assessment was blinded to treatment allocation.

Cervical JPE has been suggested to evaluate cervical propioception or kinesthesis.36 This was measured using a laser-pointer mounted onto a lightweight headband with the participants sitting blindfolded 90 cm away from a wall.29,36 Participants were asked to slowly perform full head rotation to limit vestibular input and return to the neutral position as accurately as possible and verbally indicate when they felt they were back in the neutral position.37 The examiner manually repositioned the participant's head after each trial to realign the laser-pointer with the starting position. The difference between start and end positions was measured in centimeters and converted into degrees.29,36 Six trials to each side were performed and the mean was calculated to reduce the vulnerability to outliers.38,39 The impaired side (i.e. right or left) was taken for the analysis when JPE ≥ 4.5° was only found in one direction and the mean was calculated when both sides were ≥4.5°. A moderate-good (0.71) between-day reliability has been reported for this procedure and the minimal detectable change (MDC) is −0.51°.40,41

Cervical movement sense was evaluated using the zigzag pattern (see Werner, Ernst, Treleaven, Crawford42), which is proposed to examine cervical proprioceptive or kinaesthetic afferent input as well as visuomotor function.43 Participants were sitting with the laser-pointer attached to their forehead 100 cm away from the pattern. They were asked to trace the main bold band of the pattern “as accurately as possible” in a clockwise direction to start and end in the center of the pattern.42 One familiarization trial was allowed. The test was filmed and the number and magnitude of errors and time were analyzed using SMIPlayer.42 This procedure has shown excellent intra-rater reliability (>0.90).42,43

The clinical assessment of the SPNT proposed by Daly, Giffard, Thomas, Treleaven44 was used to evaluate oculomotor control. Patients were sitting on a swivel chair. The examiner, who was at 1 m distance, moved a pen horizontally across the patient's visual field in a range of 40° at a speed of approximately 20°/s. Participants performed first the test in a neutral position (trunk and neck forward); followed by 45° trunk torsion to each side with the head fixed by the examiner. The examiner carefully observed the pursuit of the patient's eyes in each position and rated according to the score described by Della Casa and colleagues45 and adapted by Daly et al.44 The test was positive when more saccadic eye movements (excluding the outer limits and directional changes) were detected in either left and or right torsion when compared to neutral.

Standing balance evaluation consisted of one 30 s trial for each of four test conditions; firm and soft surface (high-density 9 cm thick foam rubber) with eyes open and eyes closed. These test conditions were selected because they are suggested to examine static balance disturbances due to cervical proprioceptive dysfunction rather than vestibular function and for their ability to demonstrate altered stability in people with neck pain.46,47 A force platform (Dinascan/IBV, Biomechanics Institute of Valencia, Spain) with a plate (600 × 370 × 100 mm) comprised of four force transducers was used. Participants were requested to take a comfortable standing position (feet shoulder-width apart in an angle of 20°) and look at an eye level reference point located 2 m in front. Signals were recorded with 40Hz-frequency by an amplified analogue-to-digital converter. The center of pressure displacement data were obtained in antero-posterior (AP) direction using NedSVE/IBV analysis software (Biomechanics Institute of Valencia, Spain).

A CROM device (Performance Attainment Associates, USA) was used to evaluate both tests of cervical mobility. To measure global cervical rotation ROM, patients were sitting on a chair, with the back supported on the backrest and the shoulders relaxed with the arms resting on their thighs. Then, they were requested to perform an active complete pain-free cervical rotation.48

The FRT was used to evaluate the rotation mobility of C1-C2 with the participants lying supine on a plinth.49 The examiner first passively pre-positioned participants’ neck in maximal full flexion and then rotated the head to each side. The end of the movement was determined either by a firm resistance felt by the examiner or participant's pain. Each of the cervical mobility test was repeated twice to each side and the mean of both repetitions and sides was calculated for the analysis.48 An excellent between-day reliability has been reported for the global cervical rotation ROM and the FRT (>0.9) and the MDC is 7.6° and 7.0° respectively.48,50,51

Current neck pain intensity was scored at baseline and at one-week follow-up using a VAS/100 mm with “no pain” on the left side and ‘‘maximum pain ever experienced’’ on the right side. Neck pain intensity was not assessed at post-intervention since post-needling soreness may have biased patients’ perceptions on pain. The VAS has shown an excellent between-day reliability and the minimal clinically meaningful change for patients with neck pain is 24 mm.52,53

Sample size was determined using G*Power 3.1.9.2. and calculated based on a significance level of 0.05 and a power of 80% to detect a difference of 2° in cervical JPE based on previous data.6 Following these criteria, at least 16 participants were required per group; and so, 40 participants in total were included, accounting for a drop-out rate of 20%.

Following the baseline assessment, patients were randomly assigned to each group. Randomization was achieved using a computer-generated sequence of numbers, created prior to data collection by an independent researcher, which were concealed in sealed and opaque envelopes. The therapist, who was blinded to baseline assessment results, opened the envelope and proceeded according to the group allocation. Outcome assessment and data analysis were conducted by a researcher who was blinded to participant's treatment allocation. Participants were blinded to group allocation and were instructed to not reveal any treatment experience to the outcome assessment examiner. Participants blinding was evaluated at one-week follow-up with a written form and the Bang's blinding index (BI) with a 2(DN and sham) x 3(DN, sham, and do not know) format was calculated.54

Statistical analysis was performed using SPSS statistics V25.0 and conducted according to an intention-to-treat approach. Inferential analysis including parametric and chi square tests were used to examine baseline between-group differences in patient's characteristics. Linear mixed-models with repeated-measures analysis, random effect models, and restricted maximum likelihood were used to model the intervention effect over time for primary and secondary outcome measures. We modeled the random effects of individuals and fixed effects of group (DN and Sham), time (baseline, post-intervention, and one-week follow-up), and group x time. All randomized participants were included in the analysis because the linear mixed-model estimates values for missing data.55 Pairwise comparisons with Bonferroni adjustment were used when interaction effect group x time or time was significant and change scores (compared with baseline) for post-intervention and one-week follow-up were calculated to examine if MDC was exceeded. Percentages of positive SNPT per group per timepoint were reported.

Our study has several strengths. It was prospectively registered and followed an appropriate reporting guideline.26 Similarly, it used a validated DN procedure,19 concealed allocation and blinded outcome evaluation, which improves the internal validity of the study. Another strength of this study was the sham needling protocol, which followed the most recent procedure recommendations published.35 The successful blinding measured with the BI also provides more robustness to the results of this study because poor blinding is associated with outcome bias in favor of DN.79 A further strength of the present research is the clinical rather than laboratory nature of most of the measures; which provides a more relevant message for clinical practice. On the other hand, some limitations should also be acknowledged. The low number of positive SNPT did not allow for any meaningful interpretation of the effects of the intervention on eye movement disturbances. The inclusion of a large proportion of people with idiopathic neck pain in our sample, who exhibit less marked SPNT impairments,14 may be a possible explanation.