Children-10-00523-V3 1
Children-10-00523-V3 1
Children-10-00523-V3 1
Article
Algorithm for Schroth-Curve-Type Classification of Adolescent
Idiopathic Scoliosis: An Intra- and Inter-Rater
Reliability Study †
Sanja Schreiber 1 , Eric C. Parent 1, * , Gregory N. Kawchuk 1 and Douglas M. Hedden 2
1 Department of Physical Therapy, University of Alberta, 8205 114 Street, 2-50 Corbett Hall,
Edmonton, AB T6G 2G4, Canada
2 Department of Surgery, University of Alberta, 8440 112 Street, 2D2.24 WMC, Edmonton, AB T6G 2R7, Canada
* Correspondence: [email protected]; Tel.: +1-780-492-8889
† This is a part of the Ph.D. Thesis of Sanja Schreiber.
Abstract: Schroth exercises for scoliosis are prescribed based on curve types. This study aimed to
determine the reliability of an algorithm for classifying Schroth curve types. Forty-four consecutive
volunteers with adolescent idiopathic scoliosis, 10 to 18 years old, with curves 10◦ to 50◦ , were
recruited from a scoliosis clinic. Their standing posture and Adam’s bending test were videotaped.
Ten consecutive Schroth therapist volunteers from an international registry independently classified
the curve types using the proposed classification algorithm. Videos were rated twice at least seven
days apart. Reliability was calculated using the Gwet’s AC1 agreement coefficient for all the raters
and for subgroups reporting full understanding (well-trained) and with prior algorithm experience.
The intra-rater and weighted agreement coefficients for all the raters were 0.64 (95% CI: 0.53–0.73)
and 0.75 (0.63–0.84), respectively. For the well-trained raters, they were 0.70 (0.60–0.78) and 0.82
(0.73–0.88), respectively; for the experienced raters, they were 0.81 (0.77–0.85) and 0.89 (0.80–0.94),
respectively. The inter-rater versus weighted agreement coefficients for all the raters were 0.43
(0.28–0.58) versus 0.48 (0.29–0.67). For the well-trained raters, they were 0.50 (0.38–0.61) versus 0.61
(0.49–0.72), and for the experienced raters, they were 0.67 (0.50–0.85) versus 0.79 (0.64–0.94). Full
Citation: Schreiber, S.; Parent, E.C.;
understanding and experience led to higher reliability. Use of the algorithm can help standardize
Kawchuk, G.N.; Hedden, D.M.
Schroth exercise treatment.
Algorithm for Schroth-Curve-Type
Classification of Adolescent
Idiopathic Scoliosis: An Intra- and
Keywords: scoliosis; algorithms; classification; posture; reliability; reproducibility of results; Schroth;
Inter-Rater Reliability Study. Children adolescent
2023, 10, 523. https://doi.org/
10.3390/children10030523
Schroth therapists use a classification system [9] to prescribe exercise treatment for
patients with scoliosis that is adapted to each patient’s curve type. During a clinical assess-
ment, the influence of scoliosis on the alignment and posture of each of the four body blocks
is visually appraised to determine the curve pattern. The four blocks are: (a) hip-pelvic
block, representing pelvis and lower extremities; (b) lumbar block, representing the part
of the spine where the lumbar curve appears; (c) thoracic block, where the thoracic curve
appears; and (d) shoulder block, including the shoulders and the neck (Figure 1). Scoliosis
presentation can be classified into four Schroth curve types (Figure 1), two thoracic (3c and
3cp) and two thoracolumbar/lumbar (4c and 4cp) patterns. The reliability of the Schroth
classification system is unknown.
Figure 1. Schroth classification illustrated based on the most common right-convex thoracic or left
lumbar curve direction.
From left to right, the first illustration shows a body without scoliosis; 3c is a major
thoracic curve, with a small or no lumbar curve and a balanced pelvis; 3cp is a single
long thoracic curve with the unbalanced pelvis deviated in the opposite direction of the
thoracic convexity; 4c is a major lumbar curve compensated by a thoracic curve of similar
importance (not necessarily in terms of Cobb angles) and a balanced pelvis; 4cp presents
a major thoraco-lumbar or lumbar curve, with or without a thoracic curve, and with
the unbalanced pelvis shifted in a direction opposite to the lumbar convexity. The plus
underneath each curve representation signifies the side on which the major curve convexity
appears [10].
Rigo et al. developed a classification to guide the design of the Cheneau rigid
braces [11]. Rigo’s classification is based on the Schroth classification, but it distinguishes
five curve types based on the location of external features and the position of the pelvis.
Radiological assessment is used to further subclassify the curves. The Rigo classification
has acceptable intra- and inter-rater reliability, with a kappa of 0.87 and ranging from 0.61
to 0.81, respectively [11]. Weiss also developed a classification to guide the Cheneau Light
bracing, which, similarly, relies on the Schroth classification [12]. Weiss further differenti-
ated between the curve patterns where the pelvis is imbalanced to create two additional
categories, but the reliability is unknown. Several other classifications to guide treatment
for scoliosis exist, including King’s [13], Lenke’s [14], and Peking Union Medical College’s
classifications [15]. All aim to help plan surgery, with King’s also used to guide bracing.
Although the Peking Union Medical College’s classification was found to be more reliable
(kappa = 0.90) [15] than King’s (0.64) and Lenke’s classifications (0.73) [16], Lenke’s has
been endorsed by the Scoliosis Research Society and is used most widely.
Although these reliable scoliosis classifications exist, they are used to inform bracing
or surgical treatments rather than exercise therapy, and they rely on radiographs. Con-
Children 2023, 10, 523 3 of 11
versely, the Schroth classification mainly relies on clinical assessment to guide exercise
treatment. This is consistent with most Schroth therapists not having direct access to
radiographs. A reliable curve classification strategy is needed to help standardize Schroth
exercise treatment delivery. Therefore, for the Multicenter Schroth exercise trial for scol-
iosis (NCT01610908) [6,7,17], we proposed a rule-based algorithm to assist therapists in
classifying patients. Algorithms have previously been used successfully to maximize the
reliability of the Cobb angle measurements and the King’s classification of patients with
scoliosis, for surgery purposes [18].
To demonstrate the generalizability of the novel classification algorithm, it was im-
portant to involve varied therapists, representative of the population of certified Schroth
therapists worldwide. Although Schroth therapists are spread internationally, the majority
were trained by the same instructor in Germany and could, thus, be contacted from the
institution’s registry. It was not feasible to have many raters test the same patients in person
due to travel requirements and patient fatigue. However, because the curve classification
assessment is simple, a reliability study was conducted by presenting videos of the same
patients. The benefits of video recordings are that they allow for (1) blinded ratings, (2) a
larger sample of widely distributed raters to be involved to maximize generalizability, and
(3) minimizing the patients’ burden due to repeated evaluations.
Given these considerations, the purposes of this study were to: (1) propose a Schroth
classification algorithm; and (2) determine the intra- and inter-rater reliability of Schroth thera-
pists in classifying adolescents with idiopathic scoliosis using the algorithm. We hypothesized
that therapists experienced with the algorithm and those reporting that they fully understood
the classification algorithm training material would achieve adequate reliability.
2. Methods
2.1. Research Design
In this intra- and inter-rater reliability study, therapists independently rated videos
twice, with at least one week in between assessments.
Figure 2. Snapshots from the video assessment of a 3cp curve type. The plumb line was aligned
with the umbilicus, left malleolus, right malleolus, and the gluteal cleft in the front, left, right, and
back views, respectively. The Adam’s bending test video clip displayed scoliometer readings for the
largest rotation to the right and left, measured over the trunk.
2.7. Analyses
2.7.1. Gwet’s Agreement Coefficient
Because of kappa’s well-documented limitations [20,21], we used Gwet’s agreement
coefficient (AC1) [20] to determine intra- and inter-rater reliability as well as weighted relia-
bility [20]. Percentage of agreement is also reported with 95% confidence intervals [20]. Reli-
ability estimates were obtained for: the entire sample of raters; a subgroup of two therapists
who conceptualized and had used the algorithm in an ongoing trial (labeled “experienced”);
and a subgroup of six therapists who self-reported full understanding of the algorithm
(labeled “well-trained”). Calculations were performed using AgreeStat 2013.1 for Excel
Windows (Advanced Analytics, Gaithersburg, Maryland, USA).
Children 2023, 10, 523 6 of 11
We used Gwet’s benchmarking method, which focuses only on the part of the esti-
mated agreement coefficient that confidently reflects genuine and not chance agreement [20].
This “significant part” was obtained by subtracting Gwet’s “critical value”, determined
based on the number of participants, raters, and rating categories, from the estimated
agreement coefficient [20]. With 44 participants, 10 raters, and 4 classification options,
the critical value was 0.08 [20]. Reliability estimates between 0.61 and 0.80 were, thus,
considered substantial. Adequate reliability to recommend clinical use was set a priori at
≥0.61. Mean reliability coefficients were calculated using Fisher’s transformation.
3c 3cp 4c 4cp
3c 1.00 0.75 0.50 0.00
3cp 0.75 1.00 0.00 0.00
4c 0.50 0.00 1.00 0.75
4cp 0.00 0.00 0.75 1.00
3. Results
3.1. Participants and Therapists
The participants’ characteristics are shown in Table 2. Based on the first experienced
rater’s ratings, there were nine 3c, twelve 3cp, six 4c, and seventeen 4cp Schroth curve
types. Sixty-five international therapists were invited to participate, sixteen consented,
and the first six to complete the ratings, in addition to the three research therapists and
the Schroth instructor, were included. The age of the six female and four male therapists
(mean ± standard deviation) was 46.7 ± 12.3 years old. At the time of rating, they had
worked for 6.6 ± 6.4 years as Schroth therapists (range: 3 to 23 years). Five of the therapists
had treated more than 75 patients in their career, two between 26 and 75, two between 11
and 25, and one between 4 and 10 patients. The therapists’ self-reported understanding
of the classification algorithm ranged from 50% to 100%. Six therapists reported 100%
understanding. All therapists found the algorithm useful and user-friendly, and 7 out of 10
stated they would continue using it clinically.
Children 2023, 10, 523 7 of 11
Standard
N Mean Minimum Maximum
Deviation
Age (years) 44 14.2 2.0 10.0 18.0
Upper thoracic Cobb angle (◦ ) 8 25.0 7.1 15.0 38.0
Major thoracic Cobb angle (◦ ) 36 26.7 10.3 10.0 49.0
Thoracolumbar/lumbar Cobb angle (◦ ) 37 25.8 10.0 10.0 48.0
Table 3. Intra-rater AC1 and weighted AC1 coefficients and percent agreement with 95% confidence intervals.
Weighted Percent
AC1 Percent Agreement Weighted AC1
Agreement
(95% CI) (95% CI) (95% CI)
(95% CI)
Experienced 1 0.79 (0.64–0.94) 84.1 (73.0–95.2) 0.92 (0.87–0.98) 96.0 (93.2–98.9)
Experienced 2 0.83 (0.69–0.96) 86.3 (75.9–96.8) 0.85 (0.71–0.99) 91.5 (83.7–99.2)
Rater 1 0.47 (0.27–0.67) 59.1 (44.1–74.0) 0.60 (0.41–0.79) 78.4 (68.5–88.3)
Rater 2 0.52 (0.32–0.71) 63.6 (49.0–78.3) 0.58 (0.36–0.79) 78.4 (67.5–89.4)
Rater 3 0.73 (0.57–0.89) 79.5 (67.3–91.8) 0.89 (0.82–0.96) 94.3 (90.7–97.9)
Rater 4 0.51 (0.30–0.71) 61.5 (45.8–77.2) 0.57 (0.35–0.80) 76.3 (64.2–88.4)
Rater 5 0.61 (0.43–0.80) 70.5 (56.6–84.3) 0.68 (0.50–0.87) 83.0 (73.1–92.8)
Rater 6 0.34 (0.14–0.54) 48.8 (33.5–64.2) 0.41 (0.18–0.64) 67.4 (54.8–80.1)
Rater 7 0.72 (0.55–0.88) 77.2 (64.5–90.0) 0.80 (0.66–0.95) 88.0 (79.9–96.2)
Rater 8 0.64 (0.46–0.82) 72.7 (59.2–86.3) 0.83 (0.72–0.94) 90.9 (85.3–96.5)
Mean (overall) 0.64 (0.53–0.73) 72.0 (63.9–78.6) 0.75 (0.63–0.84) 86.7 (79.9–91.3)
Mean (experienced) 0.81 (0.77–0.85) 85.0 (82.9–86.9) 0.89 (0.80–0.94 94.0 (86.8–97.3)
Mean (well-trained) 0.70 (0.60–0.78) 76.6 (69.4–82.3) 0.82 (0.73–0.88) 90.1 (84.8–93.6)
Table 4. Inter-rater AC1 and weighted AC1 coefficients and percent agreement with 95% confidence intervals.
4. Discussion
In this study, we proposed a standardized algorithm to achieve adequate rater’s relia-
bility in classifying the Schroth curve type. This algorithm was developed to assist with
standardizing exercise prescription as part of the Multicentre Schroth Exercise Trial for Sco-
liosis [6,7,17]. For the first time, we presented the intra- and inter-rater reliability of Schroth
Children 2023, 10, 523 8 of 11
5. Conclusions
We proposed a simple algorithm to maximize the therapists’ reliability in classifying
Schroth curve type and assist in the prevention of treatment errors due to misclassification.
The algorithm was well accepted, and most raters reported that they were planning to use
it clinically. Experienced and well-trained raters achieved adequate reliability. Our results
suggest that the algorithm can be used clinically with sufficient training [28].
Supplementary Materials: The following supporting information can be downloaded at: https://www.
mdpi.com/article/10.3390/children10030523/s1, Supplementary Materials S1: Figures S1–S3 Presenta-
tion of a representative case with 3c, 4c, and 4cp curve types, respectively. Supplementary Materials S2:
Operational definitions for Schroth classification algorithm. Supplementary Material S3: Data file.
Author Contributions: Conceptualization, S.S. and E.C.P.; methodology, S.S. and E.C.P.; formal
analysis, S.S.; investigation, S.S.; data curation, S.S.; writing—original draft preparation, S.S.; writing—
review and editing, E.C.P., D.M.H. and G.N.K.; visualization, S.S.; supervision, E.C.P., D.M.H. and
G.N.K.; project administration, E.C.P.; funding acquisition, E.C.P., S.S. and D.M.H. All authors have
read and agreed to the published version of the manuscript.
Funding: This research was funded by a Clinical Research Grant from the Glenrose Rehabilitation
Hospital Foundation. Sanja Schreiber was supported by the Interfaculty Graduate Studentship
awarded jointly by the Faculty of Rehabilitation Medicine and the Faculty of Medicine and Dentistry,
University of Alberta.
Institutional Review Board Statement: The study was conducted in accordance with the Declaration
of Helsinki and approved by the Health Research Ethics Board—Health Panel at University of Alberta
(Pro00019547 4 October 2011).
Informed Consent Statement: Informed consent was obtained from all subjects involved in the
study. Written informed consent was obtained from the patients appearing in the figures to publish
this paper.
Data Availability Statement: The data presented in this study are available in Supplementary Material S3:
Data file.
Acknowledgments: The authors wish to thank: Kathleen Shearer for research coordination; the
Rehabilitation Medicine Technology Group for website development and information technology
support; Axel Hennes, international Schroth instructor for reviewing the algorithm and operational
definitions prior to the start of the study; and the patients and therapists for participating in our study.
Conflicts of Interest: Sanja Schreiber is now owner of Curvy Spine, a clinic offering Schroth exercise
treatment in Edmonton, and an instructor for International Schroth Three-Dimensional Scoliosis
Therapy. She was not involved with these institutions while conducting the research. No potential
conflict of interest is reported by the other authors. The funders had no role in the design of the
study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the
decision to publish the results.
Children 2023, 10, 523 10 of 11
References
1. Weinstein, S.L.; Dolan, L.A.; Cheng, J.C.; Danielsson, A.; Morcuende, J.A. Adolescent Idiopathic Scoliosis. Lancet 2008, 371,
1527–1537. [CrossRef]
2. Asher, M.A.; Burton, D.C. Scoliosis Adolescent Idiopathic Scoliosis: Natural History and Long Term Treatment Effects. Scoliosis
Spinal Disord. 2006, 1, 2. [CrossRef]
3. LehnertSchroth, C. Three-Dimensional Treatment for Scoliosis; The Martindale Press: Palo Alto, CA, USA, 2007.
4. Kuru, T.; Yeldan, I.; Dereli, E.E.; Ozdincler, A.R.; Dikici, F.; Colak, I. The Efficacy of Three-Dimensional Schroth Exercises in
Adolescent Idiopathic Scoliosis: A Randomised Controlled Clinical Trial. Clin. Rehabil. 2016, 30, 181–190. [CrossRef]
5. Zhou, Z.; Liu, F.; Li, R.; Chen, X. The Effects of Exercise Therapy on Adolescent Idiopathic Scoliosis: An Overview of Systematic
Reviews and Meta-Analyses. Complement. Ther. Med. 2021, 58, 102697. [CrossRef]
6. Schreiber, S.; Parent, E.C.; Moez, E.K.; Hedden, D.M.; Hill, D.L.; Moreau, M.; Lou, E.; Watkins, E.M.; Southon, S.C. Schroth
Physiotherapeutic Scoliosis-Specific Exercises Added to the Standard of Care Lead to Better Cobb Angle Outcomes in Adolescents
with Idiopathic Scoliosis—An Assessor and Statistician Blinded Randomized Controlled Trial. PLoS ONE 2016, 11, e0168746.
[CrossRef]
7. Schreiber, S.; Parent, E.C.; Moez, E.K.; Hedden, D.M.; Hill, D.; Moreau, M.J.; Lou, E.; Watkins, E.M.; Southon, S.C. The Effect of Schroth
Exercises Added to the Standard of Care on the Quality of Life and Muscle Endurance in Adolescents with Idiopathic Scoliosis—An
Assessor and Statistician Blinded Randomized Controlled Trial: “SOSORT 2015 Award Winner. Scoliosis 2015, 10, 24. [CrossRef]
8. Marti, C.L.; Glassman, S.D.; Knott, P.T.; Carreon, L.Y.; Hresko, M.T. Scoliosis Research Society Members Attitudes towards
Physical Therapy and Physiotherapeutic Scoliosis Specific Exercises for Adolescent Idiopathic Scoliosis. Scoliosis 2015, 10, 16.
[CrossRef]
9. Hennes, A. Schroth—Method: Manual for the Training—Course 2008; Asklepios Katharina Schroth Klinik: Bad Sobernheim, Germany.
10. Hennes, A. International Schroth 3D Scoliosis Therapy According to Katharina Schroth—Training Manual Part I; ISST-Training Institute:
Bad Sobernheim, Germany, 2023.
11. Rigo, M.D.; Villagrasa, M.; Gallo, D. A Specific Scoliosis Classification Correlating with Brace Treatment: Description and
Reliability. Scoliosis 2010, 5, 1. [CrossRef]
12. Weiss, H.-R. “Brace Technology” Thematic Series—The ScoliOlogiC® Chêneau LightTM Brace in the Treatment of Scoliosis.
Scoliosis 2010, 5, 19. [CrossRef]
13. King, H.A.; Moe, J.H.; Bradford, D.S.; Winter, R.B. The Selection of Fusion Levels in Thoracic Idiopathic Scoliosis. J. Bone Jt. Surg.
Ser. A 1983, 65, 1302–1313. [CrossRef]
14. Lenke, L.G.; Betz, R.R.; Harms, J.; Bridwell, K.H.; Clements, D.H.; Lowe, T.G.; Blanke, K.M. Adolescent Idiopathic Scoliosis. A
New Classification to Determine Extent of Spinal Arthrodesis. J. Bone Jt. Surg. 2001, 83, 1169–1181. [CrossRef]
15. Qiu, G.; Li, Q.; Wang, Y.; Yu, B.; Qian, J.; Yu, K.; Lee, C.I.; Zhang, J.; Shen, J.; Zhao, Y.; et al. Comparison of Reliability between the
PUMC and Lenke Classification Systems for Classifying Adolescent Idiopathic Scoliosis. Spine 2008, 33, E836–E842. [CrossRef]
[PubMed]
16. Ogon, M.; Giesinger, K.; Behensky, H.; Wimmer, C.; Nogler, M.; Bach, C.M.; Krismer, M. Interobserver and Intraobserver
Reliability of Lenke’s New Scoliosis Classification System. Spine 2002, 27, 858–862. [CrossRef] [PubMed]
17. Schreiber, S.; Parent, E.C.; Hedden, D.M.; Moreau, M.; Hill, D.; Lou, E. Effect of Schroth Exercises on Curve Characteristics and
Clinical Outcomes in Adolescent Idiopathic Scoliosis: Protocol for a Multicentre Randomised Controlled Trial. J. Physiother 2014,
60, 234. [CrossRef] [PubMed]
18. Stokes, I.A.; Aronsson, D.D. Computer-Assisted Algorithms Improve Reliability of King Classification and Cobb Angle Measure-
ment of Scoliosis. Spine 2006, 31, 665–670. [CrossRef] [PubMed]
19. Coelho, D.M.; Bonagamba, G.H.; Oliveira, A.S. Scoliometer Measurements of Patients with Idiopathic Scoliosis. Braz. J. Phys.
2013, 17, 179–184. [CrossRef]
20. Gwet, K.L. Handbook of Inter-Rater Reliability: The Definitive Guide to Measuring the Extent of Agreement among Raters; Advanced
Analytics, LLC: Gaithersburg, MD, USA, 2012; p. 279.
21. Feinstein, A.R.; Cicchetti, D.V. High Agreement but Low Kappa: I. the Problems of Two Paradoxes. J. Clin. Epidemiol. 1990, 43,
543–549. [CrossRef]
22. Cummings, R.J.; Loveless, E.A.; Campbell, J.; Samelson, S.; Mazur, J.M. Interobserver Reliability and Intraobserver Reproducibility
of the System of King et al. for the Classification of Adolescent Idiopathic Scoliosis. J. Bone Jt. Surg. Am. 1998, 80, 1107–1111.
[CrossRef]
23. Phan, P.; Mezghani, N.; Nault, M.L.; Aubin, C.É.; Parent, S.; de Guise, J.; Labelle, H. A Decision Tree Can Increase Accuracy When
Assessing Curve Types According to Lenke Classification of Adolescent Idiopathic Scoliosis. Spine 2010, 35, 1054–1059. [CrossRef]
24. Lenke, L.G.; Betz, R.R.; Clements, D.; Merola, A.; Haher, T.; Lowe, T.; Newton, P.; Bridwell, K.H.; Blanke, K. Curve Prevalence of a
New Classification of Operative Adolescent Idiopathic Scoliosis: Does Classification Correlate with Treatment? Spine 2002, 27,
604–611. [CrossRef]
25. Ross, D.H.; McCluskey, S.; Fletcher-Cook, P.; Stephenson, J. The Reliability of the Leeds Movement Performance Index (LMPI):
A New Tool for Neurological Physiotherapy. Physiother Theory Pr. 2014, 30, 581–587. [CrossRef] [PubMed]
Children 2023, 10, 523 11 of 11
26. Minick, K.I.; Kiesel, K.B.; Burton, L.; Taylor, A.; Plisky, P.; Butler, R.J. Interrater Reliability of the Functional Movement Screen.
J. Strength Cond. Res. 2010, 24, 479–486. [CrossRef] [PubMed]
27. Munro, A.; Herrington, L.; Carolan, M. Reliability of 2-Dimensional Video Assessment of Frontal-Plane Dynamic Knee Valgus
during Common Athletic Screening Tasks. J. Sport. Rehabil. 2012, 21, 7–11. [CrossRef] [PubMed]
28. Grivas, T.; de Mauroy, J.C.; Négrini, S.; Kotwicki, T.; Zaina, F.; Wynne, J.; Stokes, I.; Knott, P.; Pizzetti, P.; Rigo, M.; et al.
Terminology—Glossary including acronyms and quotations in use for the conservative spinal deformities treatment: 8th
SO-SORT consensus paper. Scoliosis 2010, 5, 23. [CrossRef]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.