Objective
The purpose of the present study was to compare quality of life and curve correction associated with use of the Moss Miami and Universal Spine System instrumentation systems for spinal fusion in patients with adolescent idiopathic scoliosis. We evaluated the relationship between surgeons' satisfaction with the instrumentation systems and the outcomes of surgery, including patient satisfaction.
Design
This study was a double-blind, randomized clinical trial conducted at one center.
Participants
All adolescents with idiopathic scoliosis who were scheduled for posterior instrumentation and fusion with or without anterior release were screened for eligibility. We excluded patients with neurologic symptoms or signs; patients with spinal cord abnormalities, including tethered cord (defined as positioning of the conus medullaris caudad to the L1-L2 disc level), syringomyelia, or spinal cord tumor; patients with hydromyelia requiring decompression of an Arnold-Chiari malformation; patients with scoliosis secondary to bone abnormalities such as osteogenesis imperfecta; and patients with neuromuscular or congenital scoliosis. The present study received ethical approval from the institutional review board.
Randomization
Randomization was performed after patients had been deemed eligible for the study and informed consent had been obtained from patients and parents (as required). Patients were randomly allocated to treatment with either the Moss Miami system or the Universal Spine System in variable random-size (four to eight-patient) blocks, which were stratified according to the five participating surgeons (including four of the authors [J.G.W., A.H., B.A., and D.H.]) and according to whether the patient was to receive posterior instrumentation and fusion only or anterior release combined with posterior instrumentation and fusion. Randomization schedules were created with use of a computer, with the treatment assignments placed in sequentially numbered opaque envelopes. The trial manager (S.D.) enrolled patients and assigned participants to their groups. Surgeons were informed of the treatment assignment twenty-four hours before the procedure to allow for appropriate operating-room preparations.
Interventions
All surgeons were fellowship-trained in pediatric orthopaedics, all had experience in spinal surgery, all performed spinal surgery as part of their practice, and all had received training on both systems. Surgeons instrumented the spine in keeping with standard surgical practice2. Patients were given the option of autogenous predonation of blood. All patients received cefazolin (30 mg/kg) at the time of induction of anesthesia and then every eight hours for twenty-four hours postoperatively3. Cell savers were used. All patients had motor and sensory spinal cord monitoring. All patients had bone-grafting with use of either autogenous iliac crest bone or morselized rib bone. Facetectomy and decortication were performed in all patients.
The Foley catheter was removed within three to four days after surgery. Chest tubes, if present, were removed when drainage was <50 mL every twelve hours. Patients were managed with transfusion if the hemoglobin level dropped to <60 g/L or if the patient was symptomatic. Standard discharge criteria included normal bowel and bladder function, oral intake of fluids and solids, the ability to walk independently, and adequate pain control. All patients were given a prescription for Tylenol (acetaminophen) with codeine at the time of discharge. Patients were restricted in their physical activities for approximately six months postoperatively.
Outcomes
The primary outcome measure for this trial was Quality of Life Profile for Spinal Disorders (QLPSD)4. Quality of life was chosen for several reasons. First, many patients with scoliosis seek improvement in physical appearance, which is an important component of quality of life. Second, pain is a frequent complaint, before and after surgery, and it may dramatically influence quality of life. Finally, spinal fusion may cause physical limitations that interfere with quality of life. Secondary outcome measures included the Activities Scale for Kids (ASK)5,6, radiographic and clinical findings, surgical outcomes and complications, surgeon satisfaction with the instrumentation, and patient satisfaction with the results of surgery.
Baseline Information
We obtained patient demographic information (age, gender, menarche status, and number of months postmenarche) and details on previous nonoperative treatment (the type of brace worn and the duration of bracing). Patients completed the Quality of Life Profile for Spinal Disorders and the Activities Scale for Kids. The twenty-one-item Quality of Life Profile for Spinal Disorders, a disease-specific scale for adolescents with spine deformities4, has five subscales: psychosocial functioning, sleep disturbances, back pain, body image, and back flexibility. Climent et al. reported that the internal consistency (Cronbach alpha) and test-retest reliability of the Quality of Life Profile for Spinal Disorders scale were 0.88 and 0.91, respectively4. Those authors reported the Quality of Life Profile for Spinal Disorders scores for 174 patients, ninety-two of whom were diagnosed with adolescent idiopathic scoliosis4. Discriminant validity was detected between the operative and nonoperative treatment groups for the total score (p < 0.001), the psychosocial functioning subscore (p < 0.001), and the back flexibility subscore (p < 0.001).
The thirty-item Activities Scale for Kids focuses on the abilities and functional activities that children perform. Young et al. reported that the Activities Scale for Kids had a Cronbach alpha of 0.90 and test-retest reliability (intraclass correlation coefficient) of 0.977. The validity of the Activities Scale for Kids was demonstrated on the basis of a correlation of 0.81 (p < 0.0001) with parent-reported Child Health Assessment Questionnaire (CHAQ) scores and a correlation of 0.92 (p < 0.0001) with clinician observation6.
We made standing 3-ft (0.91-m) anteroposterior radiographs (with breast shields), lateral radiographs, supine lateral bending radiographs, and traction radiographs. The Cobb angle was measured, and the curves were classified with use of the King classification scheme8. Anteroposterior radiographs of the pelvis were used to determine the Risser9 sign and the presence of an open or closed triradiate cartilage. The clinical assessment included a standard neurologic examination, measurements of height and weight, and evaluations of spinal balance (measured in centimeters as the deviation from central [sacral] line) and rib prominence (measured in degrees with use of a scoliometer).
Outcomes at Two Years
The outcomes of surgery were determined on the basis of (1) the patient's satisfaction with the outcome, (2) the correction of the spinal deformity, (3) the perioperative outcomes and the frequency of adverse outcomes of surgery, and (4) the surgeon's assessment of the instrumentation system. We assessed outcomes two years after surgery.
Patient satisfaction with the outcome of surgery was addressed with the Quality of Life Profile for Spinal Disorders and Activities Scale for Kids. Patients also rated their satisfaction with the "look of the back" and the "results of surgery" with use of a 7-grade rating scale on which the response options ranged from 1 ("completely satisfied") to 7 ("completely dissatisfied.")
Correction of spinal deformity was evaluated on the basis of radiographic and clinical data. Radiographic measurements at two years after surgery were performed by independent evaluation. Clinical evaluations at two years were done by an individual who was blinded to the treatment group (S.D.). We measured the absolute improvement in the magnitude of the curve, the percentage of curve correction, spinal balance, and rib prominence.
Perioperative outcomes included operative time (measured in minutes), the need for blood transfusion (measured as the number of patients receiving a transfusion), the time until the patient was able to walk to the bathroom (measured as the number of days postoperatively), and the time to discharge (measured as the number of days postoperatively). We also evaluated infections at the surgical site, neurologic complications, failure or loss of fixation, nonunion (rod fracture with progression of curvature or confirmed at the time of reoperation), pneumonia, urinary tract infection, deep-vein thrombosis, superior mesenteric artery syndrome (use of a nasogastric tube more than ten days), and reoperation.
Surgeons rated their "satisfaction with the spinal instrumentation system" within twenty-four hours after the procedure with use of a 7-grade rating scale on which the response options ranged from 1 ("completely satisfied") to 7 ("completely dissatisfied"); the surgeons' ratings were based on their assessment of the ease of use and their satisfaction with intraoperative curve correction. Surgeons also rated their satisfaction with the "look of the back" and "results of surgery" at two years with use of the same 7-grade rating scale.
Blinding
Although blinded to the instrumentation system, patients and parents were asked at the time of the final assessment if they knew which instrumentation system they had received. The clinical testers assessing the patient outcomes were blinded, but this was not evaluated.
Role of Funding Source
The investigators designed and conducted the trial, collected, analyzed and interpreted data, and prepared the final manuscript. The financial sponsors did not play a role in any of these areas. Neither a medical writer nor an editor was used to produce this manuscript before it was submitted for publication. Industrial partners were invited to review the final manuscript and agreed with the results without objection. Financial support to perform the trial was provided by a Canadian Institutes Health Research Industrial Partnership with DePuy AcroMed-Johnson and Johnson Medical Products and Synthes, Canada.
Statistical Methods
Sample Size
On the basis of a standard deviation of 12.8, an alpha of 0.05, a beta of 0.2, and a delta of 5.5 (an effect size of 0.5 or 0.5 of a standard deviation), approximately sixty-three patients were required per group10. We chose 5.5 points as the clinically important difference for three reasons: (1) this value is a "median" effect size as defined by Cohen (effect size = clinically important difference)10; (2) a difference in score of 5 points is smaller than the difference in scores that Climent et al.4 reported between patients with and without back pain and between patients with small and big curves; and (3) a difference of <5 points on a 100-point scale would probably not be clinically important to patients or clinicians. An additional 3% of subjects were recruited to compensate for any potential loss to follow-up.
Primary Outcome and Secondary Outcomes
Statistical analysis was performed on an intention-to-treat basis. The primary outcome was analyzed with use of analysis of covariance. Secondary outcomes were evaluated with use of analysis of covariance for continuous data (radiographic and surgical outcomes) and chi-square tests for categorical data (complications, surgeon satisfaction, and patient satisfaction). The estimated effect size and 95% confidence intervals were calculated for primary and secondary outcomes.
Recruitment and Participant Flow
Of the 163 patients who were screened for eligibility, twenty-four did not meet inclusion criteria and four patients (or their parents) cancelled the operation; the remaining 135 families were approached to participate (Fig. 1). Six of these families refused to participate (resulting in a consent rate of 96% [129 of 135]), but they allowed us to follow the patient for determination of the outcome. The reasons for nonenrollment were that the parents did not trust hospitals (one patient); the child refused, with no reason given (one patient); the parents insisted that the surgeon choose the rod (one patient); the parents did not trust the Canadian Blood Services (one patient); and the parents refused, with no reason given (two patients). Of the sixty-three patients who were assigned to receive the Universal Spine System, three were not managed with instrumentation (two because of neurologic abnormality and one because of the family's decision not to proceed with instrumentation). Of the sixty-six patients who were assigned to receive the Moss Miami system, one was not managed with instrumentation because of neurologic abnormality. Recruitment took place between September 1997 and April 2002, and the two-year follow-up extended to April 2004. Of the 129 subjects who were enrolled in the trial, sixty (95%) of the sixty-three patients from the Universal Spine System group and sixty (91%) of the sixty-six patients from the Moss Miami group were included in the final analysis. Nine patients were excluded from the analysis: three families moved without a forwarding address, three patients refused to return for follow-up, two patients did not have complete baseline data, and one patient had been inappropriately included.
Baseline Data
As detailed in Table I, the two groups of patients receiving either the Moss Miami system or the Universal Spine System were comparable at the time of randomization.
Patient Satisfaction with Outcome
Patient satisfaction is summarized in Table II. The change in the total Quality of Life score two years after surgery did not differ between the two groups (difference, 1.07; 95% confidence interval, -3.67 to 5.82; p = 0.66). Of the five subscale scores, only the psychological/social subscale score was significantly higher in the Universal Spine System group (difference, 1.93; 95% confidence interval, 0.07 to 3.80; p = 0.04). The 95% confidence intervals did not include the clinically important difference of 5.5 points. The total Activities Scale for Kids scores at two years did not differ between the Moss Miami system group (98.97 ± 2.06) and the Universal Spine System group (97.94 ± 5.02) (difference, 1.02; 95% confidence interval, -0.38 to 2.42; p = 0.15). There was no difference between the systems in terms of the patients' satisfaction with the "look of the back" (p = 0.90) or the "result of surgery" (p = 0.68) at two years.
Correction of Spinal Deformity
Corrections of spinal deformity were not significantly different between the two systems (Table III). The percentage of Cobb angle correction was not significantly different for thoracic curves (55.1% ± 18.3% for the Moss Miami group, compared with 54.1% ± 18.7% for the Universal Spine System group) (difference, -1%; 95% confidence interval, -7% to 5%; p = 0.77) or lumbar curves (45.4% ± 24.6% for the Moss Miami group, compared with 41.9% ± 26.8% for the Universal Spine System group) (difference, -4%; 95% confidence interval, -16% to 11%; p = 0.57). The percentage changes in spinal balance (0.65% ± 1.38% for the Moss Miami group, compared with 0.5% ± 1.41% for the Universal Spine System group) (difference, -0.10%; 95% confidence interval, -0.7% to 0.40%; p = 0.59) and rib prominence (1.81% ± 7.66% for the Moss Miami group, compared with 3.75% ± 8.28% for the Universal Spine System group) (difference, 1.93%; 95% confidence interval, -1.10% to 4.96%; p = 0.21) also were not significantly different.
Perioperative Outcomes and Surgical Complications
The two groups did not differ in terms of perioperative outcomes (p = 0.20) or the complications of surgery (p = 0.11). Overall, twenty-three (17.8%) of 129 patients had complications (Table IV). However, the study was underpowered to detect differences in the rates of specific rare complications. The complications occurred across the board; all five surgeons had similar complication rates. Six patients had evidence of neurologic abnormality intraoperatively or postoperatively. Three of these patients recovered fully. In the case of the fourth patient, instrumentation was aborted because of repeated abnormal somatosensory evoked potentials during attempted instrumentation. This patient had no neurological deficit but, because of family concerns, the rod was not reinserted. The fifth patient was not managed with instrumentation because of left foot weakness during a wake-up test; the foot strength returned to normal, but the patient continued to experience altered sensation in the left foot. The sixth patient had permanent paraplegia two years postoperatively. This rate of permanent neurologic loss (0.78%; one of 129) is not far from the rate of 0.32% as reported by the Scoliosis Research Society on the basis of a retrospective, self-report survey11. The six cases of neurologic abnormality occurred following procedures performed by three different surgeons. Four of the six cases occurred following procedures performed by two of the most experienced surgeons. Five of the six cases occurred in association with the Universal Spine System (which has been used for a longer period at our institution), and one occurred in association with the Moss Miami system; this difference between the systems was not significant.
Surgeon Satisfaction
Twenty-four hours postoperatively, surgeons were more satisfied with the Universal Spine System than they were with the Moss Miami system (difference, 42%; 95% confidence interval, 29% to 55%; p < 0.0001). Satisfaction ratings for the two systems remained significantly different throughout the five years of the study (p < 0.0001). Surgeons' satisfaction ratings at twenty-four hours after surgery were not related to operating room time (p = 0.14), quality-of-life scores (p = 0.11), the need for blood transfusions (p = 0.17), the overall number of complications (p = 0.36), thoracic curve correction (p = 0.59), lumbar curve correction (p = 0.91), surgeon satisfaction with the "results of surgery" (p = 0.62), or the "look of the back" (p = 0.91) at two years.
Blinding
Of the 129 patients, 105 did not know which system they had received, six thought they had received the Universal Spine System (two were correct), and ten thought they had received the Moss Miami system (seven were correct); the responses for the remaining eight patients were missing. The parents of ninety-seven patients did not know which system had been used, the parents of nine patients thought that the Universal Spine System had been used (five responses were correct), and the parents of ten patients thought that the Moss Miami system had been used (five responses were correct); the remaining thirteen responses were missing.
Idiopathic scoliosis affects up to 2.2% (0.9% to 3.2%) of the population12-16. As many as 10% of adolescents with idiopathic scoliosis have progression of the curve to the range at which surgery is considered. The long-term consequences for patients with idiopathic scoliosis who do not undergo surgery are controversial. Many of the curves progress in severity. However, in a long-term (fifty-year) follow-up study from Iowa, patients with scoliosis were not severely disabled17. Patients with scoliosis had more frequent back pain in comparison with patients without scoliosis, but the severity of the pain was similar in both groups17. In a twenty-year follow-up study, Danielsson et al. found that patients with scoliosis had physical function and psychological well-being equivalent to that in the general population but had lower cosmetic well-being18. Cardiopulmonary compromise is an infrequent outcome of severe and progressive scoliosis. Thus, a prime rationale for surgery is the treatment of current or future deformity.
Spinal instrumentation and fusion for the treatment of adolescent idiopathic scoliosis is a common and expensive surgical procedure. The outcome of surgery can be evaluated from the perspective of the patient or the surgeon. The present study demonstrated that two spinal instrumentation systems that work on the basis of different biomechanical principles did not result in differences in patients' quality of life, cosmetic appearance, or physical function after surgery. Although both groups of patients undergoing surgery had a modest increase in quality of life in comparison with their preoperative status, this increase did not differ between the two instrumentation systems.
Because the aim of surgery is to correct deformity and to have the spine fuse in the corrected position, the ability to correct the curve is important to surgeons and patients. The instrumentation systems were comparable with regard to the correction of all aspects of spinal deformity. Despite being a more rigid rod, the Universal Spine System did not have an increased ability to correct the curve. Another important outcome of surgery, for both the surgeon and the patient, is the occurrence of complications. Although the present study was not powered to detect a significant difference, the results of the study did not demonstrate differences between the two systems in terms of infection rates, neurologic complications, or the need for reoperation. The overall complication rate in the present study (17.8%) was higher than that in retrospective studies, most likely because of the prospective evaluation of complications with explicit definitions by a third party. A well-accepted principle of clinical research is that data on outcomes and complications should not be collected by the treating surgeon. Although the complication rates may be partially attributed to the use of two rather than one system during the trial, Richards et al. recently reported a similar rate of reoperation for the treatment of complications (13%)19 and Vitale et al., in a study of 937 patients managed with spinal fusion for the treatment of scoliosis, reported a complication rate of 26%20. On the basis of the results of this and other studies, the complication rate and the need for additional surgery may be higher than previously appreciated.
Although the two systems showed no difference in either absolute correction or percentage curve correction, surgeons expressed a strong preference for the Universal Spine System as compared with the Moss Miami system. To our knowledge, no previous study has directly addressed surgeons' preferences for surgical instrumentation. We did not prospectively collect information on the specifics of surgeons' preferences for the instrumentation system. However, in view of the findings of the present study, evaluating the origins of surgeons' preferences should be the topic for future investigation. Furthermore, the present study was unique in that it investigated surgeons' preferences in the context of a randomized trial. A randomized design was essential for investigating surgeons' satisfaction with the system to ensure that differences in surgeons' preferences could not be attributed to differences among patients; that is, without randomization, surgeons could have argued that their satisfaction with the instrumentation was because it was best for the patients.
The four potential explanations for surgeons' preferences include patient outcomes, patient preferences, financial incentives, and ease of use. First, as discussed above, no meaningful differences in patient outcomes were detected. Second, patient preferences could not have affected surgeon preference because patients had no role in choosing the type of instrumentation and the type of instrumentation was not apparent to them. Third, financial interest seems unlikely because the surgeons involved in the trial had no commercial interest in either system. Furthermore, although partial funding for this trial originated from the two companies, the study was equally and jointly funded. Fourth, the preferences of at least a few of the surgeons reportedly were due to ease of use. One surgeon expressed dissatisfaction with cross-threading of the locking nut of the Moss Miami system, and another had the impression that the Universal Spine System was better for straightening the spine. One surgeon indicated that, with the Moss Miami system, it was more difficult to reduce the rod into the hooks and that the closure mechanism to secure the rod was difficult. Another surgeon said that the Universal Spine System did not have as much play and that the Moss Miami system had more flexibility in building the construct. The surgeons who were involved in the trial, however, were familiar with both instrumentation systems. Furthermore, if lack of familiarity with a system affected surgeons' preferences, the preferences should have changed with time, which was not the case. Finally, the operative time and the degree of surgical correction were similar in the two groups. Thus, surgeons' preferences either reflect some factor that was not evaluated in the present study or they reflect surgeon bias. This finding is important because surgeons' preferences are likely the driving force at the institutional and family level when choosing instrumentation systems in many clinical contexts. An additional factor not evaluated in this study is the cost of the instrumentation systems. If two systems are associated with similar patient outcomes, and assuming adequate technical and service support, cost must also be a consideration. In evaluating cost, however, more than just the cost of implants should be considered. For example, systems that require shorter operative time, allow shorter fusion, obviate the need for an anterior approach, or have lower complication rates may be less expensive despite higher implant costs.
The present study had several potential limitations. First, it compared only hook-and-rod systems and did not evaluate other techniques such as pedicle screw fixation. Pedicle screw fixation may result in better correction of the curve but may increase the initial expense21,22. Long-term outcomes will determine if reduced rates of reoperation and less anterior surgery will offset the increased implant costs. Whether pedicle screw fixation results in improved quality of life without increased risk of consequences requires additional study. Second, the present study evaluated only two of the many instrumentation systems. We deliberately chose these two systems because they work on the basis of different principles and because they differ in terms of their stiffness. Thus, we believe that other systems that are based on similar principles probably would provide similar results. Third, the finding that surgeons' preferences for the instrumentation systems were not related to patient outcomes cannot be extrapolated to other situations in which surgeons have strong preferences. However, the results of the present study suggest that future surgical trials should include an evaluation of surgeons' preferences and should relate those preferences to patient outcomes. Fourth, the scale that was used for surgeon satisfaction was a subjective global appraisal. Future research will be necessary to determine the specific elements of surgeon preferences. Fifth, because scoliosis presents with a wide range of curve magnitude and stiffness, the results of the present study cannot be extrapolated to all curve types. However, stratified randomization as used in this study is the best method of minimizing this potential bias on study outcome.
In conclusion, patients' quality of life and the objective outcomes of surgery did not differ between systems, but surgeon preferences did. It is unclear why surgeon preferences differ, but these data suggest that surgeon preference may be an unreliable means of selecting implants from the perspective of patient outcomes. To our knowledge, the present study represents not only the first randomized clinical trial to compare spinal implants in patients with adolescent idiopathic scoliosis but also the first trial to directly address surgeon preferences for instrumentation systems within the context of a randomized clinical trial. 