Study participants were recruited if they had NCNP, defined as pain of mechanical origin located in the anatomical region of the neck, with or without radiation to the head, trunk or limbs, as described by Guzman et al[3
]. The inclusion criteria were: aged between 18 and 60 years, neck pain lasting 12 weeks or more, no physical therapy, not currently under chiropractic care or rehabilitation for the neck area, willingness to adhere to the treatment protocol, and signed informed consent. Participants with concurrent headaches, non-radicular pain in the upper extremities and lower back pain were not excluded if neck pain was the main symptom. The exclusion criteria included neck pain due to a motor vehicle accident, neck surgery, severe osteoarthritis and inflammatory arthritis, neurological, cardiovascular, infectious metabolic and endocrine diseases, pregnancy and any cardinal signs of potential vertebral artery dissection.
Participants were recruited through radio and printed advertisements aimed at the French-speaking community of Trois-Rivières, Québec (population 126,000). The study was carried out entirely at the chiropractic clinic and human research laboratory of the Department of Chiropractic at the Université du Québec à Trois-Rivières. All participants gave their informed written consent according to the research protocol approved by the local ethics committee.
Three research assistants undertook standardised phone screening regarding major inclusion and exclusion criteria for study participation. Three other research assistants, who were experienced chiropractors and were blinded to future treatment allocation, performed the preliminary standardised history-taking and physical examination of eligible participants. Radiographs of the cervical spine were ordered when indicated[21
]. All preliminary outcome measures were collected at this point.
Randomisation and blinding
The trial was divided into 2 phases. The first was the non-randomised, symptomatic phase during which all eligible participants received a short course of SMT. After completing the symptomatic phase, participants were randomly assigned to one of three parallel groups. The second phase, the preventive phase, lasted 10 months.
Participants had an equal probability of assignment to any 1 of the 3 groups. Each study participant was assigned a number by the principal investigator (PI) who put these numbers on individual cards in an opaque envelope. An assistant, not involved in the research project and blinded to the process, drew numbers from the original, opaque envelope and arranged them in 3 different opaque envelopes which were numbered 1 to 3. Randomisation was concealed until the beginning of the preventive treatment phase. No block or stratification strategy was used in the randomisation process.
Randomisation was carried out after acceptance of the appropriate number of participants into the study and before beginning the non-randomised, symptomatic phase. The PI opened the sealed, opaque envelopes after the symptomatic phase was over, and baseline (second set) outcome measures were collected. Each participant was assigned to his/her appropriate group before beginning the preventive phase of the trial. They remained on the same allocation throughout the entire period if they continued in the trial.
A credible placebo for SMT of patients previously administered this type of therapy does not exist, especially if its specific and non-specific effects are considered[22
]. Therefore, neither the participants, the treating chiropractors nor the assessors were blinded to treatment allocation for the future preventive phase of the trial. Only the data analyst was blinded to treatment allocation.
During the symptomatic phase of the trial, all eligible participants received a short course of SMT designed to relieve symptoms. A team of 3 chiropractors using high velocity low amplitude spinal manipulation[8
] and with at least 3 years of experience was responsible for treating the study participants. The interventions were standardised beforehand and lasted 10 to 15 minutes. Between 10 and 15 treatments were provided over a 5- to 6-week period. Each treatment consisted of a maximum of 4 spinal manipulations to the cervical and upper thoracic areas (down to T4). Myofascial soft tissue therapy (brief trigger point therapy) was permitted but was to be kept to a minimum. No advice or educative strategies were allowed. At this stage, the chiropractors and study participants were blinded to treatment allocation for the preventive phase.
The preventive phase of the trial lasted 10 months during which the participants attended the clinic regularly. A chiropractor with at least 3 years of experience was in charge of the interventions for each group. The interventions were standardised beforehand. Medication(s) and co-intervention such as other manual therapies, physiotherapy, massage therapy as well as any other common neck pain treatment were discouraged. Before adopting any such pain control strategy, the participants were instructed to first call their treating chiropractor to discuss the problem. Patients in all 3 groups received identical verbal and written instructions regarding co-interventions and were asked to record any co-interventions in a personal diary. Patients were given an ice pack that could be used whenever pain intensity reached a level of 5/10. Again all 3 groups received identical instructions concerning ice application.
At each visit, the chiropractor asked the participants for a standardised, short health history regarding symptoms during the last period and had them complete a visual analog scale (VAS) for current symptoms. The chiropractor also performed standardised passive palpation of the cervical and upper thoracic spine. No advice or educative strategy was allowed during this phase, nor was any type of soft tissue therapy. Diaries were distributed at these visits. Other interventions during each visit were specific for each of the 3 groups:
Spinal manipulation group
This group received a maximum of 4 spinal manipulations to the cervical and upper thoracic areas. They were given 1 treatment per month that lasted 10 to 15 minutes.
Spinal manipulation combined with a home exercise program group
This group received a maximum of 4 spinal manipulations to the cervical and upper thoracic areas (down to T4). They were dispensed with 1 treatment per month, and each of them lasted 10 to 15 minutes.
Participants were advised to perform a home exercise program at least 3 times a week. The program was designed by an experienced kinesiologist and required low technology equipment (elastic tubing and foam physioballs). It included general range of motion (ROM) exercises that served for warm-up and cool down purposes, followed by 4 stretching/mobilization and 4 strengthening exercises (concentric and isometric contractions) of the cervical and upper thoracic spine, principally flexion/extension, lateral flexion and rotation of the cervical spine. Three series of each exercise were performed during a training session, with a 30- to 60-second rest period between each series. A complete training session lasted between 20 to 30 minutes.
All participants were instructed to follow the same exercise routine. However, exercise volume was tailored to each participant's strength and flexibility as well as his/her ability to complete the routine with minimal neck pain. Each patient received a written copy of the program. At trial onset, a kinesiologist met each participant individually on 2 different occasions to instruct and correct them on execution of the exercises. Between these 2 meetings, follow-up was conducted through phone interviews to answer questions and confirm that the exercises were not triggering cervical pain. The kinesiologist met the participants individually every 2 months (during the regular treatment visit) to ensure full understanding, appropriate execution and compliance. Minor changes were made only when exercises produced symptoms or when a patient was unable to perform a specific exercise. During the 10 months of the preventive phase, exercise type and volume could be modified under the following conditions: exercise-related pain described by the patient or major difficulty in carrying out an exercise.
This group received no treatment (no SMT or exercise program) but each participant attended the clinic once every 2 months. To give all trial participants the same attention, each visit lasted twice as long as the other two groups treatment time, specifically 20 to 30 minutes. The same procedures as for the other 2 groups were performed at these meetings (standardised short health history, VAS, standardised passive palpation and the distribution of diaries) but with much slower flow. As with the 2 other groups, no advice or educative strategy was allowed.
Primary and secondary outcome measures
The primary outcome measure throughout the trial was pain level. Pain was scored with a 10-cm VAS. The psychometric properties of the VAS have been studied extensively[25
]. The number of patients that stayed below a level of clinically acceptable pain (2 point difference from the symptomatic phase baseline VAS score) during the preventive phase of the trial was also assessed. For example, a patient that went from a score of 5 to 3 on the VAS during the symptomatic phase and maintained or improved this score throughout the preventive phase was considered to show a clinically meaningful response. Function and disability were considered the secondary outcomes in the study. Cervical spine function was assessed with the cervical range of motion instrument (cROM©
). Many studies have demonstrated acceptable validity and reliability of this instrument[26
]. Active cervical bilateral rotation, bilateral lateral flexion, and flexion and extension were measured with the patient in the seated position. The patient performed each active cROM three consecutive times in sequence-specific order. The mean of the 3 consecutive readings was used in this study. Disability was measured with the Neck Pain Disability Index (NDI) and the Bournemouth Questionnaire (BQ). The psychometric properties of these 2 instruments, including their translation into the French language[27
] have been thoroughly assessed and are adequate for such a trial[28
Exploratory outcome measures
Exploratory outcome measures included health-related quality of life (HRQOL), fear and avoidance phenomena, exercise adherence and co-intervention. HRQOL and fear avoidance phenomena were scored with the SF-12 Questionnaire[33
] and the Fear-avoidance Behaviour Questionnaire (FABQ),[34
] respectively, whose psychometric properties are adequate for such a trial. The physical health composite score and the mental health score from the SF-12 were calculated.
Exercise adherence and co-intervention were measured through a diary that patients completed on a weekly basis. Use of co-intervention, registered by the patients, was calculated by adding up the number of co-intervention episodes from one visit to another. Also recorded were the frequency of ice application due to pain intensity greater than 5/10 and the use of analgesics or consultation with other professionals for pain management. Patients performing home exercises also recorded the dates of their 3 weekly exercise sessions.
Preliminary data were collected after participants signed an informed consent form, before the beginning of the non-randomised, symptomatic phase of the trial. Baseline data were gathered once the symptomatic phase was over, on specific appointment. During the 10 months of the preventive phase, all outcomes were noted every 2 months. For the groups receiving treatments, these measurements were recorded before the treatments.
For preliminary and baseline data, the assessors were blinded to group allocation. The assessors were not blinded to the data during the preventive phase but were trained before the trial to ensure standardisation of the outcome measures.
Thirty-five participants per group were required to have a 90% chance of detecting a significant difference between groups in the mean of VAS scores (i.e. a 2-point difference on the VAS at the 2-sided 5% level) with an assumed standard deviation of 2.5 and a loss to follow-up of 20%. Since no RCT on the efficacy of preventive spinal manipulation for chronic cervical pain has been performed in the past, the 20% loss to follow-up level was estimated from the results of a similar RCT on the efficacy of preventive spinal manipulation for chronic lumbar pain[11
]. The effect size was also estimated from this trial.
All data were analysed according to a pre-established experimental design using Version 6.1 of Statistica software. One-way ANOVA was performed for baseline values of continuous variables. T-tests for dependent samples were conducted on primary and secondary outcomes to analyse data from the symptomatic phase of the trial (pooled data).
The main analysis was undertaken on an intention-to-treat basis. Missing values were imputed on the basis of the last-observation-carried-forward technique and included all randomly-assigned participants who stayed in the study until the first visit of the preventive phase. All clinical variables were analyzed using ANCOVA, with treatment and time intervals representing the main factors. Gender, age and pain improvement in the symptomatic phase were used as covariates in the analyses. Data were adjusted for gender and pain improvement in the symptomatic phase because these 2 variables were deemed to play an important role in possible further improvement of the patient's condition. One-way ANOVA served to compare co-intervention data across the 3 groups.
Whenever factorial analyses revealed significant effects, post hoc analyses were performed with the least significant difference test. For all analyses, statistical significance was set at p < 0.05.