The study design is a randomized controlled clinical trial with two independent and blinded observers. Two strata were studied: an unbraced group and a braced group. Each stratum had patients treated by standard medical care (observation or brace treatment), standard medical care plus chiropractic manipulation, or standard medical care plus sham manipulation. Patient blinding was attempted in the manipulation and sham manipulation interventions only.
The target population was children aged 10–16 years who had been diagnosed (via x-ray) with AIS curves ranging from 20 to 30 degrees, and those with curves varying from 30 to 40 degrees who were undergoing bracing treatment. Study participants were recruited from the pediatric orthopedic clinic at The Kalamazoo Center for Medical Studies. Potential patients were screened using patient clinic records for inclusion and exclusion criteria. Those that met the study's criteria were invited to participate.
• Children aged 10–16 years
• Diagnosis of Adolescent Idiopathic Scoliosis
• Spinal curvature on P/A radiographs of between 20 and 30 degrees measured by the Cobb method in non-braced individuals and 30 to 40 degrees in braced individuals
• Palpatory evidence of subluxation (manipulative lesion) on chiropractic screening examination
• Signed informed consent by parent/guardian
• Signed child assent form
• Availability for follow-up evaluation
• Age <10 or >16 years
• Diagnosis other than AIS following clinical, radiographic and advanced imaging assessment
• Contraindications to manipulation: inflammatory arthritides, osteomyelitis, neoplasm, metabolic disturbances affecting the integrity of bone structure, fracture/dislocation/spinal instability, blood clotting disorders and connective tissue disorders
• Congenital or acquired structural spinal abnormalities
• Leg length inequality >3/8 inch (measured via x-ray)
• Pain as a primary clinical feature
• Mental incapacitation
• Previous back surgery
• Significant recent trauma
• Obesity impairing ability to manipulate
The Institutional Review Board of Borgess Medical Center, Kalamazoo approved the study protocol. Patients and their parents/guardians were informed and signed consent forms (parents/guardians) and assent forms (study participants) prior to participation. Following consent, further screening procedures and baseline data were gathered, including health history, physical exam (vital signs), neurological exam (deep tendon reflex, sensory deficit testing, muscle strength testing, Babinski), chiropractic spinal exam (static palpation, motion palpation, postural assessment), biomechanical evaluation (Shobers' modified-modified technique, Adams test, thoracic range of motion), quality-of-life self-report (Scoliosis Quality of Life Index, general health) and plain film radiographs. Patient recruitment occurred between March and July of 2003.
"Costs per randomization" were calculated by dividing the recruitment costs (the personnel time necessary to screen and recruit potential patients) by the number of randomizations. This information allowed assessment of future full-scale trial recruitment costs.
Prior to the start of the study, a computer-generated randomization schedule was prepared. Eligible patients were randomly assigned to 1 of 3 treatment groups (standard medical care, standard medical care plus chiropractic manipulation, or standard medical care plus sham manipulation). The allocation ratio was 1:1:1. Neither the participants nor the investigators knew whether a particular participant had been assigned to a study group or to a control group (standard medical care or standard medical care plus sham manipulation) until after assignment. The clinical research assistant e-mailed a member of the randomization team (off site) regarding the need for an assignment for a "qualified" participant by providing the clinic numeric code and bracing status. A member of the randomization team then provided the random assignment by return e-mail to the clinical research assistant.
Study participants randomized to the simulated and real treatment groups were given their choice of participating chiropractic treatment clinics. Once a clinic had been selected for attendance and treatment, the subject remained with that clinic for the duration of the trial. The treating chiropractor was provided with radiographs and Cobb angle measures for each study participant.
Participating treating chiropractors were recruited from Kalamazoo and Battle Creek, Michigan. Five chiropractic treatment centers were credentialed for the study and all used Diversified Technique as their primary method of treatment. Diversified technique is a widely used chiropractic manipulative technique that entails a high velocity, low-amplitude thrust.
Treating chiropractors were trained, tested and certified in study protocols and treatment methods. Three training sessions (approximately 2 hours each) were held at the medical center with the treating chiropractors, orthopedic surgeons and research staff. Each treating chiropractor was provided with a packet of training materials prior to the sessions. The first session focused on the study's research design and protocols and presented background information on adolescent idiopathic scoliosis. The second training session included an explanation of a standardized approach for detecting dysfunctional vertebrae, adjusting procedures and sham adjusting procedures. The final training session included a review of the study research design and protocols, standardized approach for detecting dysfunctional vertebras, adjusting procedures and sham adjusting techniques. During the last hour of the session, participants were evaluated using the Objective Structured Clinical Examination [21
Upon entry to a participating chiropractic clinic, each subject underwent a standardized consultation, history and physical examination, including a chiropractic assessment. The patient and his or her parent/guardian were given a report of findings, a standardized explanation of chiropractic principles related to scoliosis, and an outline of the treatment schedule and procedures to be performed. On subsequent visits, pre-treatment analysis, treatment, post-treatment analysis and adverse reaction were recorded on standardized study treatment sheets. The treating chiropractors followed a scripted interview for each patient visit.
Usual and customary medical care for AIS patients with a Cobb of 20 to 25 degrees consists of careful observation (e.g., physical and radiographic examination twice a year). Patients with curves between 26 and 40 degrees are potential candidates for bracing with physical and radiographic examination twice a year, and patients beyond 40 degrees are potential candidates for surgery [23
]. All groups received usual and customary medical care, including follow-up examination and x-ray at 6 months.
Active chiropractic treatment for this study consisted of prone, side posture and supine adjustments in conjunction with manual soft tissue therapy to the overlying tissues. Treatment consisted of the full spine chiropractic manipulation technique known as The Diversified Technique [24
]. Because this was intended to be a pragmatic study within the domains of adjustive and soft tissue therapies, the specifics of the treatment (i.e. vertebral segments, direction of manipulation and use of soft tissue therapy) were left to the discretion of the treating practitioner.
Sham (pretend) chiropractic treatment consisted of a standardized approach which mimics regular chiropractic treatment, but which does not have the same mechanical effect. Subjects were placed prone on the treatment table and the spine was palpated lightly in a posterior-anterior direction. In this prone position, the head was rotated first to the right, then to the left, and held for a few moments while the chiropractor palpated the ankles and feet (a distraction maneuver). Following this distraction, the patient was positioned in a side-lying posture, and positioned for a low back adjustment (superior leg and hip flexed) without joint slack taken up. This position was held for a few moments, and the chiropractor contacted the soft tissues overlying the gluteal region and administered a light (non-therapeutic) impulse. This was repeated on the opposite side. The patient was then placed prone on the table, and the chiropractor administered a light impulse bilaterally to the muscles overlying the scapulae. The subject was then positioned supine, and the neck was palpated gently. Following this, the head was rotated to the side and held for a few moments, followed by a light impulse on the cranium over the external occipital protuberance.
Schedule of visits
The schedule of treatment represents the frequency and duration of treatment typically used by the chiropractic profession. To form an estimate for the frequency and duration of care, a pre-study survey was administered to a random sample of American chiropractors (90% response rate) assessing treatment dosage for chiropractic management of AIS [26
]. Generally, treatment consisted of three treatments per week for the first month, two treatments per week for the second month, one treatment per week for the third and fourth months, followed by a maintenance program of two treatments per month for the fifth and sixth months. Our treatment protocol was chosen based on the results of the survey of the American chiropractors [26
]. For this effectiveness study, treating practitioners were allowed to increase or decrease the frequency of care, depending on clinical presentation (i.e. pain, postural changes, and changes evident on assessment). Patient compliance was deemed adequate when a patient received between 80% and 120% of the above-described treatments.
Data collection and statistical analysis
All patients followed the same algorithmic protocols for initial evaluation and follow-up review. We collected data on radiographs, demographics, clinical history and quality-of-life domains at entry. Radiographs and quality-of-life measures were evaluated at baseline and at 6 months.
Cobb is the primary outcome measure for this study and is the gold standard for the measurement of curve magnitude in scoliosis [27
]. Endorsed by the Scoliosis Research Society, measurement of the Cobb angle on full spine serial radiographs is used to make clinical decisions regarding initiation, termination and success of treatment. The reliability of the Cobb angle measure has been evaluated in many studies. When strict measurement protocols are used, consisting of sharpening marking instruments, standardizing protractors, and standardizing end plate selection, examiner error can be minimized [30
]. In this study, strict measurement protocols were used and intra- and inter-examiner reliability was measured using two independent orthopedic surgeons blinded to treatment allocation.
Because there is a clinically important increase of curve severity (5 degrees) in moderate idiopathic scoliosis between morning and evening, the comparative x-rays were taken at approximately the same time of day (+one hour) as the entry x-ray [31
]. Additionally, braced patients were required to remove their brace 6 hours prior to radiographic examination [32
Quality of life is important in AIS because of the psychosocial stresses experienced by these patients. As a secondary outcome measure, we used Scoliosis Quality of Life Index (SQLI). This measure is a 22 item self-reporting health-related quality-of-life questionnaire for patients 10 to 18 years of age with idiopathic scoliosis [33
]. SQLI is reliable (test-retest ICC 2,1; 0.80), valid (construct validity with Quality of Life Profile for Spine Deformities, Spearman's rho; 0.79) and demonstrates satisfactory distribution of scores. SQLI has five domains: physical activity performance (the presence and extent of physical limitations); back pain (the intensity and frequency of back pain); self-esteem (social confidence, self-regard, self-appearance, overall life assessment); moods and feelings (anxiety, depression and positive affect); and satisfaction with management. All scales are scored from zero (most pain, worst function, etc.) to 100 (no pain, best function, etc.). A global scale for SQLI (scored 0 to 100) was calculated by averaging the scores of all the scales.
Patient expectations about the therapeutic benefit of the treatment were assessed before randomization [34
]. Allowing only naive patients is problematic, because it may be difficult to recruit this population. Furthermore, an expectation for or against chiropractic could influence patient scores on subjective measures. If patient preference is disproportionately distributed among the groups, there may be an inflated threat of obtaining a skewed outcome [35
]. Patients for this study were asked to describe their expectations for improvement of their spinal condition (without regard to treatment) using a 5-point scale with choices varying from "very much improved," to "very much worsened." The patients then rated how helpful they believed chiropractic would be for their current spinal problems using a 5-point scale with choices varying from "very helpful" to "very unhelpful".
To provide proof of blinding for patients in the real and sham manipulation groups, patients were questioned after all therapy had been administered about whether they received active or inactive therapy. A 7-point scale was used with choices varying from "definitely real therapy," to "definitely pretend therapy." We also asked patients about co-interventions and contamination during the study period.
Touch screen technology (TST) was used to administer SQLI and expectation, blinding, co-Intervention and contamination questions. The quality of the data collected with the touch-screen system has been reported as good, with no missed responses [36
]. This method eliminates possible entry errors and the need for double-entry checks. Additionally, it is well accepted by patients, the majority of whom find all aspects of the TST system easy to use [38
Quality assurance procedures were established and implemented for all aspects of the trial. We developed a manual of operational procedures which included operational definitions of recruitment, measurement procedures, etc. All forms were standardized (pre-coded, self explanatory, easy to read, coherent, pretested, and labeled on every page with an ID number).
Simple descriptive analyses were used to report the findings. Outcome measures were not aggregated, but were reported individually because of insufficient sample size and the high risk of committing sampling errors. To assess intra- and inter-reliability, intraclass correlation coefficients (ICC 2,1) were calculated for the Cobb measures [39
]. All data were entered into a spreadsheet and analyzed with Minitab 12 (State College, PA). All data were checked for accuracy.