Introduction The treatment of adolescent idiopathic scoliosis (AIS) is based on the ability to reduce truncal deformity, limit curve progression, and improve quality of life. Because AIS does not lead to increased mortality , successful treatment outcomes are often dependent on improvements in healthrelated quality of life (HRQOL) or prevention of future impact on HRQOL. In light of this, the Scoliosis Research Society (SRS) instrument was designed by Haher et al.  to provide a simple, practical, and disease-specific assessment of patients’ perception of their condition. Studies of the SRS-22 instrument have demonstrated it to have good internal consistency, score distribution, reproducibility, responsiveness, and concurrent validity [3–7]. An HRQOL questionnaire must also have the ability to distinguish between differing severities of diseases, defined as the discriminative validity .
The health-related quality of life as measured by SRS-22 has been generally shown to worsen with increasing disease severity in AIS [9–13]. However, the discriminative validity of SRS-22 has still not been definitively demonstrated [6,14]. Moreover, the extent to which SRS-22 can be relied on to guide management decisions is not clear [15,16]. For this reason, management of AIS continues to be heavily based on objective radiographic measures, trunk rotation, pain, and patient preference . To better understand the utility of SRS-22 in AIS, it is important to clearly define the association between patient perceived disease severity and the magnitude of spinal deformity among nonoperative and preoperative patients with increasing curve severity. The purpose of this study was to analyze the discriminative validity of each individual domain of the SRS-22 instrument across five subgroups of AIS patients with increasing curve severity.
To our knowledge, no study has evaluated the discriminative validity of SRS-22 in AIS using three preoperative groups and two large nonoperative groups in a single age- and sex-matched comparison. Materials and methods Study design This is a retrospective review of prospectively administered SRS-22 questionnaires issued to two pretreatment patient populations: nonoperative and preoperative patients with AIS. The two populations were then separated into five groups based on the major Cobb angle magnitude and planned treatment modality: nonoperative with nonbraceable, minimal/scoliosis (0–19), nonoperative (20–40), and preoperative (41–50, 51–60, andO60). Curve magnitude increments were determined to maximize the number of patients within each group while separating nonoperative from operative curves. Preoperative patients were divided into smaller curve increments of 10 to reflect variations in operative management. Progressively larger curves require more extensive osteotomies, a combined approach, and longer fusion constructs—and therefore represent separate surgical categories. Subjects were asked to complete all questions of the SRS-22 instrument with the exception of two items (21 and 22), which pertain to posttreatment satisfaction.
The SRS-22 domain and overall scores for each group were calculated as per Asher et al. . Subjects The SRS-22 questionnaires were administered to 67 nonoperative and 219 preoperative patients evaluated at the senior author’s practice between February 2000 and May 2009. Patients aged 10 to 21 years with documented spine curvature and no other major medical comorbidities were asked to complete the questionnaire during their first office visit with the senior author. All patients were without a history of AIS treatment either by bracing or surgery. Of note, two patients had curve magnitudes of less than 10 (6 and 8). Although these two values are less than the conventionally used minimal curve magnitude to define AIS, both patients were carefully followed because of a strong family history and parental concerns for progression. Hence, the treatment of these two patients (with close outpatient monitoring) was identical to the remainder of the 0 to 19 group.
Considering that manual curve measurements often include error of up to 5 and that the inclusion of these two patients within the nonoperative group mimics clinical practice, we believe that their inclusion is highly unlikely to change the outcome of the article or the message for surgeons using SRS-22. Nonoperative patients were subsequently divided into two groups by curve magnitude (0–19 and 20–40) and matched by both age (within 1 year) and sex (for each group, n531; 23 females and 8 males). Five nonoperative patients were excluded because of the absence of a corresponding match. Because nonoperative patients were least in number, all subsequent groups formed from preoperative populations were limited to 31 randomly selected patients matched for age and sex. Consequently, 95 preoperative patients were excluded from the study. Statistical analysis Descriptive statistics were used to characterize the comparison groups, SRS-22 domain, and total scores. Differences between the five study groups’ domain scores (activity, pain, image, mental health) and total scores were analyzed by one-way analysis of variance with a Tukey post hoc analysis. Statistical significance was achieved at p!.05. Results Descriptive statistics of the five study groups are contained within Table 1.
Of the 155 patients included in the 2 J.L. Berliner et al. / The Spine Journal - (2012) - study, 115 (74.2%) were female. Patient age ranged from 11.0 to 17.2 years with a mean age of 13.8 years. Mean age was 13.5 years for the nonoperative group and 14.0 years for the preoperative group. Curve magnitude ranged from 6 to 93 with a mean curve magnitude of 43.1. Mean curve magnitude was 21.9 for the nonoperative group and 57.2 for the preoperative group. The nonoperative group had a mean angle of trunk rotation of 8.7 compared with 15.1 for the preoperative group. The mean angle of trunk rotation of both the 0 to 19 and 20 to 40 subgroups was found to be significantly different (p!.05) compared with the other four subgroups. The nonoperative group was made up of the following curve types: 56.5% thoracic, 38.7% thoracolumbar, and 4.8% lumbar. In comparison, the preoperative group was 65.2% thoracic, 34.8% thoracolumbar, and 0% lumbar. The SRS-22 questionnaire domains and total scores by the study group are shown in Figure and Table 2. A significant difference between study groups was found within two of the four domains (pain and image) and total score (p values shown in Table 3).
The magnitude of all significant differences between study groups was larger than 0.5 on the fivepoint domain scale, which is within the range of values recommended as the minimum clinically important difference for SRS-22 domains . A more recent study of postoperative patients with AIS recommends 0.20 and 0.98 as the minimum clinically important difference for the pain and image domains, respectively . Although all statistically significant mean differences between subgroups within the pain domain did exceed 0.20, none exceeded the recommended value of 0.98 within the image domain. Both nonoperative groups (0–19 and 20–40) demonstrated significantly higher pain scores (indicating less pain) than the preoperative group with curves measuring 41 to 50 (p!.05). This difference did not persist, however, when comparing the nonoperative groups with the most severe curves (51–60 and O60).
The two nonoperative groups did not differ in terms of pain domain scores. Similarly, comparing the three preoperative groups showed no statistical differences in the pain score. Both nonoperative groups (0–19 and 20–40) scored significantly higher in the image domain (indicating better self-image) than all preoperative groups (41–50, 51–60, and O60) (p!.05). Once again, the two nonoperative groups did not differ in terms of image domain score. In the same way, none of the three operative groups differed in terms of image score when compared. Both nonoperative groups’ SRS-22 total scores were significantly higher than the three preoperative groups total scores with one exception; total score from the nonoperative 20 to 40 group was not significantly higher than that from the preoperative O60 group. No significant differences were found within the function and mental health domains when comparing the five groups. Discussion The SRS-22 questionnaire has been validated and reevaluated in numerous studies to better define and improve the questionnaire’s psychometric attributes. However, previous studies establishing the discriminative validity of Table 1 Descriptions of study subjects and curve-severity subgroups* Groups Mean Cobb ( ) Age (y) Sex Mean ATR ( ) y Lenke type (Type:#) Curve type All subjects 43.1620.7 13.861.6 115 F, 40 M 12.565.0 1 (44), 2 (10), 3 (3), 4 (2), 5 (11), 6 (1) 95 T, 56 TL, 3 L 0–19 14.163.4 13.461.7 23 F, 8 M 6.461.8z N/A (non-op) 16 T, 12 TL, 3 L 20–40 29.864.8 13.761.5 23 F, 8 M 10.963.3z N/A (non-op) 19 T, 12 TL, 0 L 41–50 45.162.9 13.861.6 23 F, 8 M 13.964.2 1 (22), 2 (1), 5 (3), 6 (1) 24 T, 7 TL, 0 L 51–60 54.362.5 14.261.7 23 F, 8 M 14.563.2 1 (14), 2 (5), 5 (4), 6 (5) 18 T, 12 TL, 0 L O60 72.069.7 14.061.5 23 F, 8 M 16.964.4 1 (8), 2 (4), 3 (3), 4 (2), 5 (4), 6 (6) 18 T, 13 TL, 0 L ATR, apical trunk rotation; F, female; M, male; T, thoracic; TL, thoracolumbar; L, lumbar; N/A (non-op), nonoperative. * Mean6standard deviation. y ATR values not available for four study subjects. z ATR value significantly different from all other subgroups (p!.05). Figure.
Comparison of mean Scoliosis Research Society 22 scores among curve-severity subgroups controlled for age and sex. Domains without statistically significant differences (function and mental health) not depicted. Error bars indicate 6standard deviation. J.L. Berliner et al. / The Spine Journal - (2012) - 3 SRS-22 have lacked sufficiently matched study groups and have limited their comparison to three or fewer subgroups of disease severity [6,14]. As management decisions are commonly dependent on curve severity and patient reported outcomes, the association between these two parameters should be more clearly defined. In a study of 119 patients with AIS, Asher et al.  found significant differences in the pain and image domain scores when comparing a control group and a nonsurgical group with an unmatched surgical group. However, no significant differences were found between the control and nonsurgical groups.
In a more recent study from Japan, an age- and sex-matched study comparing 141 AIS patients with 72 nonscoliotic adolescents found that AIS patients scored significantly lower in the pain and image domains. However, no differences in the function and mental health domains existed between groups. Although the study did further classify scoliosis patients into mild, moderate, and severe curve groups, the subgroups were unmatched and not compared in terms of individual domains and total SRS-22 scores . Parent et al.  studied the discriminative validity of SRS-22 across varying curve severities in a group of 158 females stratified into three curve severity subgroups (!30, 30–50, andO50). Although none of the study subjects had undergone a corrective surgery, 35 had previously received bracing treatment. Overall, curvesO50 had worse SRS-22 scores in all domains compared with the 30 to 50 group. Scores were also significantly worse for curves O50 than for curves !30 in the domains of pain, image, and total score. Although the study was statistically adjusted for age, curve severity groups were not age matched.
In the adult population, Baldus et al.  established the discriminative validity of SRS-22 in 935 primary adult scoliosis patients (with Cobb anglesO30) and a control group of 1,222 volunteers. Six age- and sex-matched groups were compared: both male and female, 20 to 40, 41 to 60, and 61 to 80 years. All individual domain scores from groups with scoliosis were lower than the corresponding age- and sexmatched normative groups. There was one exception, however, in the mental health domain; males aged 61 to 80 years with scoliosis scored similar to their corresponding control group . In a similar study, Berven et al.  compared 146 adults with scoliosis with 34 unaffected adults matched for age, sex, and socioeconomic status and found significant differences within all four domains. Neither study, however, performed an analysis on the deformity group comparing varying degrees of curve severity with SRS-22 scores. Our results demonstrate that the questionnaire’s discriminative validity within both the pain and image domains was capable of differentiating smaller nonoperative curves from larger curves indicated for operative management. Asher et al. , in a previous study, reported similar findings.
However, within the pain domain, the questionnaire was unable to generate statistically different scores between nonoperative patients and those with the largest curve magnitudes (ranging from 51 to 93, mean 62.3). Similarly, Weinstein  found no relationship between curve Table 2 Comparison of SRS-22 scores among curve-severity subgroups (controlled for age and sex)* Groups Function Pain Image Mental health Total 0–19 4.560.47 4.660.58 4.160.44 4.460.56 4.460.36 20–40 4.460.37 4.660.51 4.060.54 4.260.64 4.360.38 41–50 4.160.69 4.160.72y,z 3.560.59y,z 4.160.58 3.960.48y,z 51–60 4.260.54 4.360.65 3.360.58y,z 4.060.58 3.960.44y,z O60 4.360.55 4.160.91 3.560.57y,z 4.060.80 4.060.53y SRS, Scoliosis Research Society. * Means6standard deviations. Higher scores indicate better wellbeing for the patient. y Significant difference with 0 to 19 subgroup. z Significant difference with 20 to 40 subgroup. Table 3 p Values from comparison of curve-severity subgroup SRS-22 scores* Groups 0–19 20–40 41–50 51–60 O60 Function 0–19 — 1.00 0.09 1.00 1.00 20–40 1.00 — 0.63 1.00 1.00 41–50 0.09 0.63 — 1.00 1.00 51–60 1.00 1.00 1.00 — 1.00 O60 1.00 1.00 1.00 1.00 — Pain 0–19 — 1.00 0.05* 0.55 0.17 20–40 1.00 — 0.04* 0.46 0.14 41–50 0.05* 0.04* — 0.99 1.00 51–60 0.55 0.46 0.99 — 1.00 O60 0.17 0.14 1.00 1.00 — Image 0–19 — 1.00 0.00* 0.00* 0.00* 20–40 1.00 — 0.01* 0.00* 0.00* 41–50 0.00* 0.01* — 1.00 1.00 51–60 0.00* 0.00* 1.00 — 1.00 O60 0.00* 0.00* 1.00 1.00 — Mental health 0–19 — 1.00 0.97 0.44 0.83 20–40 1.00 — 1.00 1.00 1.00 41–50 0.97 1.00 — 1.00 1.00 51–60 0.44 1.00 1.00 — 1.00 O60 0.83 1.00 1.00 1.00 — Total 0–19 — 0.99 0.00* 0.00* 0.01* 20–40 0.99 — 0.04* 0.04* 0.15 41–50 0.00* 0.04* — 1.00 1.00 51–60 0.00* 0.04* 1.00 — 1.00 O60 0.01* 0.15 1.00 1.00 — SRS, Scoliosis Research Society. Significant differences within pain domain: 0 to 19 versus 41 to 50; 20 to 40 versus 41 to 50. Significant differences within image domain: 0 to 19 versus 41 to 50, 51 to 60, andO60; 20 to 40 versus 41 to 50, 51 to 60, and O60. Significant differences within total domain: 0 to 19 versus 41 to 50, 51 to 60, andO60; 20 to 40 versus 41 to 50 and 51 to 60. * Bolded values: Significant p values (p!.05). 4 J.L. Berliner et al. / The Spine Journal - (2012) - severity and pain in 194 AIS patients. This finding may be secondary to outside factors affecting an individual’s subjective pain assessment.
Additionally, the pain domain has been shown to have a moderate (O21.6%) ceiling effect (the proportion of patients answering any given question with a maximum score of 5) among AIS patients younger than 18 years, which may explain our observed results . In contrast to Parent et al. , our analysis does not show significant differences between the domain scores of preoperative (41–50, 51–60, andO60) or nonoperative groups (0–19 and 20–40). This may suggest a limit to the questionnaire’s discriminative capacity. In addition, we matched all study groups within 1 year to avoid the potential confounding variable of age. This is of particular concern in AIS as previous studies have demonstrated increasing ceiling effects of SRS-22 among progressively younger patient populations [21,24]. Our study was also limited to pretreatment patients to eliminate the effects of treatment course and satisfaction on SRS-22 scores. Total SRS-22 scores were found to be significantly lower in the preoperative groups compared with those in the nonoperative groups. The one exception was between the largest curve measurements (O60) and the nonoperative 20 to 40 group.
Again, this may be related to ceiling effects within the pain domain. Overall, these results are consistent with previous studies demonstrating adequate discrimination within the image, pain, and total scores but poor differentiation between patients with minimal-to-moderate curves [6,12,14]. Our results suggest that SRS-22 may also not effectively differentiate patients with the largest deformities. Taken together, SRS-22 does not seem to reliably differentiate between relatively small increments of spinal curvature across the full spectrum of disease. This may be attributable to minimal differences in quality of life with small incremental changes of curvature. Previous outcome studies have demonstrated that it is difficult to measure differences in quality of life resulting from small changes in disease severity [25,26]. It may also support the existence of a nonlinear relationship between SRS-22 scores and increasing curve magnitude as previously suggested .
In this study, the function and mental health domain scores did not differ significantly between study groups. These domains may be particularly susceptible to confounding influences and psychosocial factors. In 450 nonscoliotic adolescents, Verma et al.  reported that race, sex, household income, and household status were all independent predictors of function, mental health, image, and total SRS-22 scores. Previous studies have also failed to report a significant decrease in function among adolescent and adult scoliosis patients compared with controls [22,28]. Patient confidence, self-esteem, mood, obesity, and overall self-image may predominate over physical complaints and therefore exert a significant effect on HRQOL measures [29–32]. Furthermore, because AIS is a disease that is largely treated prophylactically, only subtle differences in HRQOL may exist between patients with pathology. This poses a challenge for current and future questionnaires to more effectively target the effects of AIS on patients’ HRQOL. As with any cohort study, there is a possibility of bias in the selection of study groups.
In addition, correlations between SRS-22 outcomes and disease severity have been shown to be the strongest in subjects with single thoracic curves compared with those with double or thoracolumbar curves . Therefore, stronger correlations between disease severity and SRS-22 outcome may be observed in future studies matching for curve type. Future questionnaires may be designed to better address the impact of psychosocial factors on HRQOL . The development and refinement of questionnaires sensitive to these additional factors, such as the Body Image Disturbance Questionnaire-Scoliosis version, will be important to increase our understanding of the relationship between clinical deformity and patient-centered outcomes . Conclusion The SRS-22 questionnaire demonstrated good discriminative validity between small nonoperative curves and larger surgical curves within the pain, image, and total domains.
However, SRS-22 lacked the ability to generate significantly different scores between small intervals of curve magnitude, suggesting a limit to the questionnaire’s discriminative capacity. These findings may be impacted by additional variables affecting SRS-22 outcome, namely psychosocial factors including socioeconomic status, confidence, self-esteem, mood, and obesity. Effective measures of psychosocial factors should be incorporated into the development of future HRQOL questionnaires for AIS. Key points 1. The Scoliosis Research Society (SRS)-22 questionnaire demonstrated good discriminative validity between small nonoperative curves and larger surgical curves within the pain and image domains as well as total score. 2. The SRS-22 questionnaire demonstrated poor discriminative validity within the function and mental health domains. 3.
No significant differences were found between the domain scores of nonoperative groups (0– 19 and 20–40) or between the domain scores of preoperative groups (41–50, 51–60, and O60), suggesting a limit to the questionnaire’s discriminative capacity at small curve intervals. 4. Our findings may be explained by additional variables affecting SRS-22 outcomes, namely psychosocial factors. J.L. Berliner et al. / The Spine Journal - (2012) - 5
 Pehrsson K, Larsson S, Oden A, et al. Long-term follow-up of patients with untreated scoliosis. A study of mortality, causes of death, and symptoms. Spine 1992;17:1091–6.
 Haher TR, Gorup JM, Shin TM, et al. Results of the Scoliosis Research Society instrument for evaluation of surgical outcome in adolescent idiopathic scoliosis. A multicenter study of 244 patients. Spine 1999;24:1435–40.
 Asher MA, Min Lai S, Burton DC. Further development and validation of the Scoliosis Research Society (SRS) outcomes instrument. Spine 2000;25:2381–6.
 Asher M, Min Lai S, Burton D, et al. The reliability and concurrent validity of the scoliosis research society-22 patient questionnaire for idiopathic scoliosis. Spine 2003;28:63–9.
 Asher M, Min Lai S, Burton D, et al. Scoliosis research society-22 patient questionnaire: responsiveness to change associated with surgical treatment. Spine 2003;28:70–3.
 Asher M, Min Lai S, Burton D, et al. Discrimination validity of the scoliosis research society-22 patient questionnaire: relationship to idiopathic scoliosis curve pattern and curve size. Spine 2003;28:74–8.
 Asher MA, Lai SM, Glattes RC, et al. Refinement of the SRS-22 Health-Related Quality of Life questionnaire Function domain. Spine 2006;31:593–7.
 Burton DC, Glattes RC. Measuring outcomes in spinal deformity. Neurosurg Clin N Am 2007;18:403–5.
 Parent EC, Hill D, Mahood J, et al. Associations between quality-oflife and internal or external spinal deformity measurements in adolescent with idiopathic scoliosis (AIS). Stud Health Technol Inform 2006;123:357–63.
 Parent EC, Wong D, Hill D, et al. The association between Scoliosis Research Society-22 scores and scoliosis severity changes at a clinically relevant threshold. Spine 2010;35:315–22.
 Wilson PL, Newton PO, Wenger DR, et al. A multicenter study analyzing the relationship of a standardized radiographic scoring system of adolescent idiopathic scoliosis and the Scoliosis Research Society outcomes instrument. Spine 2002;27:2036–40.
 Watanabe K, Hasegawa K, Hirano T, et al. Use of the scoliosis research society outcomes instrument to evaluate patient outcome in untreated idiopathic scoliosis patients in Japan: part I: comparison with nonscoliosis group: preliminary/limited review in a Japanese population. Spine 2005;30:1197–201.
 Asher M, Lai S, Burton D, et al. Spine deformity correlates better than trunk deformity with idiopathic scoliosis patients’ quality of life questionnaire responses. Stud Health Technol Inform 2002;91:462.
 Parent EC, Hill D, Mahood J, et al. Discriminative and predictive validity of the scoliosis research society-22 questionnaire in management and curve-severity subgroups of adolescents with idiopathic scoliosis. Spine 2009;34:2450–7.
 D’Andrea LP, Betz RR, Lenke LG, et al. Do radiographic parameters correlate with clinical outcomes in adolescent idiopathic scoliosis? Spine 2000;25:1795–802.
 Merola AA, Haher TR, Brkaric M, et al. A multicenter study of the outcomes of the surgical treatment of adolescent idiopathic scoliosis using the Scoliosis Research Society (SRS) outcome instrument. Spine 2002;27:2046–51.
 Cobb J. Outline for the study of scoliosis, instructional course lectures. J Am Acad Orthop Surg 1948;5:261–75.
 Berven SH, Deviren V, Polly DW. Minimal clinically important difference in spinal deformity: defining a threshold of change that matters. Paper presented at: Scoliosis Research Society 40th Annual Meeting, Poster 4. 2005; Miami, FL.
 Carreon LY, Sanders JO, Diab M, et al. The minimum clinically important difference in Scoliosis Research Society-22 appearance, activity, and pain domains after surgical correction of adolescent idiopathic scoliosis. Spine 2010;35:2079–83.
 Baldus C, Bridwell KH, Harrast J, et al. Age-gender matched comparison of SRS instrument scores between adult deformity and normal adults: are all SRS domains disease specific? Spine 2008;33: 2214–8.
 Berven S, Deviren V, Demir-Deviren S, et al. Studies in the modified Scoliosis Research Society Outcomes Instrument in adults: validation, reliability, and discriminatory capacity. Spine 2003;28:2164–9; discussion 9.
 Weinstein S. Long-term follow-up of pediatric orthopaedic conditions: natural history and outcomes of treatment. J Bone Joint Surg Am 2000;82:980.
 Parent EC, Dang R, Hill D, et al. Score distribution of the scoliosis research society-22 questionnaire in subgroups of patients of all ages with idiopathic scoliosis. Spine 2010;35:568–77.
 Parent E, Moreau M, Mahood J. Score distribution and discriminative validity of quality of life scores in patients with idiopathic scoliosis. J Can Chir 2006;49.
 Fischer D, Stewart AL, Bloch DA, et al. Capturing the patient’s view of change as a clinical outcome measure. J Am Med Assoc 1999;282: 1157–62.
 Taylor SJ, Taylor AE, Foy MA, et al. Responsiveness of common outcome measures for patients with low back pain. Spine 1999;24: 1805–12.
 Verma K, Lonner B, Hoashi JS, et al. Demographic factors affect Scoliosis Research Society-22 performance in healthy adolescents: a comparative baseline for adolescents with idiopathic scoliosis. Spine 2010;35:2134–9.
 Horal J. The clinical appearance of low back disorders in the city of Gothenburg, Sweden. Comparisons of incapacitated probands with matched controls. Acta Orthop Scand Suppl 1969;118:1–109.
 Johnson L. Outcomes analysis in spinal research. How clinical research differs from outcomes analysis. Orthop Clin North Am 1994;25:205–13.
 Pencharz J, Young NL, Owen JL, et al. Comparison of three outcomes instruments in children. J Pediatr Orthop 2001;21:425–32.
 Gaynes BN, Burns BJ, Tweed DL, et al. Depression and healthrelated quality of life. J Nerv Ment Dis 2002;190:799–806.
 de Zwaan M, Mitchell J, Howell L, et al. Two measures of healthrelated quality of life in morbid obesity. Obes Res 2002;10:1143–51.
 Asher M, Lai SM, Burton D, et al. The influence of spine and trunk deformity on preoperative idiopathic scoliosis patients’ health-related quality of life questionnaire responses. Spine 2004;29:861–8.
 Auerbach J, Lonner B, Crerand C, et al. Body image in patients with adolescent idiopathic scoliosis: validation of the body image disturbance questionnaire-scoliosis version. Scoliosis Research Society Annual Meeting, Kyoto, Japan, 2010, Podium Presentation.