Three hundred eighty-one patients (353 girls and 28 boys) with Boston brace treatment for idiopathic scoliosis were included and followed prospectively [7]. The reason for bracing was adolescent idiopathic scoliosis (AIS) discovered at 10 years of age or older [8] in 351 patients, and late-onset juvenile idiopathic scoliosis at 7 to 9 years in 30 patients. The mean primary major curve measured 33.1 (range 20–57)°. The indication for bracing was a major scoliotic curve ≥ 20° with an observed progression >5° after 4 months and Risser sign <3. Patients had follow-ups with clinical and radiological examination at 4-6 months intervals throughout the brace treatment period. Wearing of the brace was assessed by one of the authors (orthopaedic scoliosis surgeon JEL) and reported as: ordinary as prescribed, irregular, or aborted. Patients were recommended to use the brace for 23 h daily. Wearing of the brace <20 h daily was described as irregular. Initially bracing was planned until maturity (Risser 5 in boys or Risser 4 or two years after menarche in girls). Because some of the patients started bracing at a young age with a long time treatment perspective, the original plan was changed and the idea of planned early weaning to shorten the years of brace wear was concepted and realized 35 years ago by the senior author (JEL). He started early weaning in patients before reaching puberty after 2–3 years or more of brace treatment. These patients demonstrated a good effect of the brace with a stable reduction of the primary curve to 25° or less at repeated measurements. Later, influenced by the positive results associated with this new regimen, he also included a few patients with less than 2 years of brace wear. All had skeletal age <14 and Risser 3 or lower at planned early weaning. The weaning procedure was performed during a period of 2–3 months by 2 h less bracing time every week and the curve was controlled after 4 months. Patients with early weaning were followed at regular intervals of 4–6 months until skeletal maturity. After weaning patients with ordinary brace treatment time had follow-ups at 6, 12, and 24 months. After several years a long time follow-up was accomplished in all the patients.
Radiology
Prior to bracing standing radiographs were taken in the front and lateral projections. In addition non-weight bearing (flexibility) frontal radiographs were exposed in the prone position, and after brace fit standing (redressment) radiographs with the brace on were taken. Radiological measurements were performed by an orthopedic surgeon (JEL) and controlled by an experienced radiologist. Both used the Cobb method manually. Digital measurements were used at long-term follow-up. The intra-observer error for the Cobb angle was about 3° in a relatively recent study using manual and digital measurements, and <5° in a previous study [9, 10]. In the present study the measurement error was within these limits as evaluated by the reproducibility of radiographic readings of repeated measurements of all radiographs from 10 patients at regular intervals. In patients with double thoracolumbar curves the largest curve prior to bracing was defined as the major curve.
Surgery
Surgery was recommended in patients with curve progression to >45° during bracing and at weaning. At later follow-up patients with major curves <50° were not recommended surgery. Information about surgery was recorded in a standardised form and obtained from a long-term questionnaire and checked in the medical journals.
Questionnaires
At long-term follow-up, a standardised questionnaire was filled in by the patients. The questionnaire comprised validated measures of pain, disability, quality of life and work, comorbidity, surgery, and questions about demographics.
A validated Norwegian version of the Scoliosis Research Society 22 questionnaire (SRS-22) was used for evaluation of health related quality of life [11]. The SRS-22 covers four domains (function/activity, pain, self-perceived image, mental health) each with 5 questions, and one domain (satisfaction with treatment) with 2 questions. Each item has 5 verbal response alternatives ranging from 1 (worst) to 5 (best). Results are expressed as the mean (total sum of the domain divided by the number of items answered) for each domain.
Patients rated their overall function by the Global Back Disability Question [12]. This is a single question designed to measure the patients’ overall rating of their back disability today. There were five response alternatives: “excellent, none or unimportant complaints", “good, occasionally bothered by back pain", “fair, some back pain and limited function", “poor, unchanged, considerable complaints and severe disability", and “miserable, worse, not self-reliant in activities of daily living”.
A Norwegian version of the original Oswestry Disability Index (version 1.0) was used to evaluate back-specific disability [12]. The sum of 10 questions is calculated and presented as a percentage, wherein 0 % represents no pain and disability, and 100 % represents the worst pain and disability.
The General Function Score was used to measure disability in activities of daily living [13]. Patients answered nine questions using one of three alternatives: “can perform”, “can perform with difficulty due to back complaints”, and “cannot perform due to back complaints”. The score was presented as a percentage wherein 100 % represents maximum disability.
EuroQol is a generic (non-disease specific) questionnaire for measurement of health related quality of life and includes five items regarding quality of daily life, covering the domains of mobility, self-care, usual activities, pain and discomfort, and anxiety and depression (EQ-5D) and a visual analogue score (EQ-VAS) for assessment of overall current health [14, 15].
Statistical analysis
Results are presented as means and range or standard deviation, or as percentages. The normal distribution of baseline, follow-up data, and differences were checked by histograms and by the Shapiro-Wilk test. In sample cases of non-normal distribution non-parametric methods were used. A General Linear Model One-way analysis of variance was applied to test differences between the 3 subgroups in continuous variables at baseline, weaning, and follow-up: 1) patients with unplanned early weaning, 2) planned early weaning, and 3) ordinary brace wear until maturity. In a previous study we classified irregular wear and unplanned early weaning as poor compliance [7]. In the present study we classified patients with irregular wear as ordinary bracing unless bracing was aborted. With the assumption of unequal variances in unequally sized groups, Dunnett’s T3 was used for post hoc multiple comparisons. Chi-square analyses were applied for testing of categorical variables and Kendall’s Tau-b measure of association for multiple nominal variables. Statistical evaluation was performed by use of the Statistical Analysis System (SAS version 9.2; Cary, NC) and by SPSS software, version 21.0 (SPSS Inc., Chicago).
Consent
The committee for medical research ethics in the health Region South-East in Norway and the institutional review board (hospital’s patient ombudsman) approved the study (REK 2010-3677). Written informed consent was obtained from the participating patients.