JBJS | Article

The Journal of Bone & Joint Surgery, Volume 83, Issue 1

Articles | January 01, 2001

Measurement Accuracy in Congenital Scoliosis

Fernando A.M. Facanha-Filho, MD; Robert B. Winter, MD; John E. Lonstein, MD; Steven Koop, MD; Thomas Novacheck, MD; E. A. L’Heureux, JrMD; Cheryl A. Noren, ART

View Disclosures and Other Information

Investigation performed at Twin Cities Spine Center, Minneapolis, Minnesota
Fernando A.M. Facanha-Filho, MD Av. Da Universidade 1940, Fortaleza Ceara, Brazil 60020
Robert B. Winter, MD John E. Lonstein, MD Cheryl A. Noren, ART Twin Cities Spine Center, 913 East 26th Street, Suite 600, Minneapolis, MN 55404
Steven Koop, MD Thomas Novacheck, MD Gillette Children’s Specialty Healthcare, 200 East University Avenue, St. Paul, MN 55101
E.A. L’Heureux Jr, MD Spine Center at Renaissance, 1701 Renaissance Boulevard, Suite 101, Edmond, OK 73013-3022
No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study.

The Journal of Bone & Joint Surgery. 2001; 83:42-42

5 Recommendations (Recommend) | 3 Comments | Saved by 3 Users Save Case

text A A A

Abstract

Background:

The accuracy of measurement of curves in idiopathic scoliosis has been extensively studied; however, we know of only one article in the literature concerning the accuracy of measurement of curves in congenital scoliosis. That article stated that intraobserver variability was ± 9.6° and interobserver variability was ± 11.8°.

Methods:

Sixty-nine curves in fifty patients with congenital scoliosis were measured on two separate occasions by seven different observers with varying experience in curve measurement.

Results:

Mean intraobserver variance ranged from 1.9° to 5.0°, with an average of 2.8° (95% confidence limit, ± 3°) for the seven observers. The interobserver variance was 3.35° (95% confidence limit, 7.86°).

Conclusions:

It is possible to measure curves in congenital scoliosis with much greater accuracy than previously reported. In the clinical situation in which a skilled observer can measure two radiographs at the same time, an accuracy of ± 3° can be expected 95% of the time.

Figures in this Article

Introduction

Introduction | Materials and Methods | Results | Discussion | References

The Cobb method is the standard technique for measurement of deformities of the spine¹. Although the Cobb angle is recognized as being a measurement of the tilt of the end vertebrae instead of a method that measures all aspects of the deformity, it has been used to document progression of the curve, to select the type of treatment, and to evaluate the effectiveness of treatment. This method was originally described for measurement of curves in idiopathic scoliosis, and because of its simplicity it became the method for measurement of other types of scoliotic curves and other spinal deformities in the sagittal plane. Its accuracy in the measurement of noncongenital curves has been tested by several authors^2-8.

Measurement is often more difficult in patients with congenital scoliosis because of the absence of defined end plates and the difference or absence of the pedicles. In a review of the literature, we found only one study of the reliability of the Cobb measurement in patients with congenital scoliosis. In that study, Loder et al. found an intraobserver and interobserver variability of ±9.6° and ±11.8°, respectively⁹. According to those authors, to ensure with 95% confidence that the increase in the curve is not due to error of measurement, at least 23° of change is necessary when the measurement is performed by two observers at different times; they believed that this margin can be reduced to 19° if the same observer performs both measurements.

It appeared to us that the variability found in the study of Loder et al.⁹ was too high. If one follows the guidelines of Loder et al. in a hypothetical situation of a patient with a curve of 40°, the curve should reach at least 59° (if both radiographs are measured by the same person) to ensure documentation of progression of the deformity and thus to indicate the need for surgical treatment.

Therefore, the purpose of the present study was to evaluate the reliability of the Cobb measurement in patients with a diagnosis of congenital scoliosis and to compare our results with those of Loder et al.⁹.

View Large | Download Slide(.ppt)
Add to My JBJS

Fig. 1:: In general, the intraobserver variance increased by 0.2° for every increase of 10° in curve magnitude (p = 0.040).

View Large | Download Slide(.ppt)
Add to My JBJS

Fig. 2:: In general, the interobserver variance increased by 0.3° for every increase of 10° in curve magnitude (p = 0.016).

View Larger

TABLE I: Intraobserver Variance: Summation of Seven Observers

*The values are given in degrees. †The sum of all of the degrees of variance between the first and second measurements of all sixty-nine curves. ‡The total variance divided by sixty-nine. §The values were derived by an analysis of the variability for all measures (across all observers).

Observer	Total Variance*†	Mean Variance*‡	Standard Error of Mean*	Minimum Variance*	Maximum Variance*	Lower 95% Confidence Limit*	Upper 95% Confidence Limit*	Rank of Observer
1	163	2.4	0.315	0	7	1.948	3.183	4
2	347	5.0	0.671	0	27	3.757	6.388	7
3	158	2.3	0.346	0	16	1.597	2.954	3
4	185	2.7	0.448	0	11	1.904	3.661	5
5	248	3.6	0.500	0	20	2.556	4.517	6
6	130	1.9	0.161	0	7	1.582	2.215	1
7	131	1.9	0.200	0	6	1.492	2.276	2
Overall§	1049	2.8	0.162	0	27	2.542	3.177

Add to My JBJS

TABLE I: Intraobserver Variance: Summation of Seven Observers

Observer	Total Variance*†	Mean Variance*‡	Standard Error of Mean*	Minimum Variance*	Maximum Variance*	Lower 95% Confidence Limit*	Upper 95% Confidence Limit*	Rank of Observer
1	163	2.4	0.315	0	7	1.948	3.183	4
2	347	5.0	0.671	0	27	3.757	6.388	7
3	158	2.3	0.346	0	16	1.597	2.954	3
4	185	2.7	0.448	0	11	1.904	3.661	5
5	248	3.6	0.500	0	20	2.556	4.517	6
6	130	1.9	0.161	0	7	1.582	2.215	1
7	131	1.9	0.200	0	6	1.492	2.276	2
Overall§	1049	2.8	0.162	0	27	2.542	3.177

Materials and Methods

Introduction | Materials and Methods | Results | Discussion | References

Fifty anteroposterior radiographs of children with congenital scoliosis were selected. These radiographs were chosen because they were of good quality, they showed a variety of malformations and curve sizes, and they represented patients with a wide range of ages. One curve was identified on thirty-three radiographs; two curves, on fifteen radiographs; and three curves, on two radiographs. Thus, a total of sixty-nine curves were measured. The youngest patient in the series was six months old, and the oldest was eighteen years and five months old. The congenital deformities were classified with regard to the type of malformation of the spinal elements and the area of the spine involved.

The observers were blinded to the dates on which the radiographs were made, the patients’ identities, and any other previous measurement. A number was used to identify each radiograph. The end vertebrae of the curves were preselected by the senior author (R.B.W.) in order to eliminate this source of variability and thus to use the same methodology as Loder et al. used⁹. The same set of fifty radiographs was measured by seven different observers: two highly experienced spine surgeons, two pediatric orthopaedic surgeons with major experience with treatment of the spine, two spine surgery fellows, and one highly experienced scoliosis research technician. In contrast to Loder et al., we did not involve radiologists. The same type of protractor (United States Manufacturing, Pasadena, California) and the same type of pencil were used for all angular determinations. After each measurement, the numbers were recorded and all lines and marks were erased. Each observer measured the radiographs twice, with an interval of at least three weeks between the two measurements, again in accordance with the methodology of Loder et al.

Results

Introduction | Materials and Methods | Results | Discussion | References

Intraobserver Variance

To determine intraobserver variance, we examined the differences between the two measurements of the same curve by the same observer. Table I, which summarizes the findings of the seven observers, clearly shows their differing degrees of accuracy. Upper and lower 95% confidence limits for each observer are shown, as is the overall 95% confidence limit, which was ±3°. This was a much "tighter" measurement than the ±9.6° in the study by Loder et al.⁹. We noted a spread of accuracy among the seven observers, with the mean variance ranging from 1.9° to 5.0°. There was a direct relationship between the experience of the observer and the degree of accuracy. Observer 7 (a highly experienced scoliosis research technician who did not have an MD degree) and Observer 6 (a highly experienced spine fellow) were the most accurate. They each had a mean variance of only 1.9°. Observer 1 (a highly experienced spine surgeon) and Observer 3 (a pediatric orthopaedist with an interest in the spine) were second in accuracy, with 2.4° and 2.3° of mean variance, respectively. Observer 4 (a pediatric orthopaedist with an interest in the spine) and Observer 5 (a highly experienced spine surgeon) were third in accuracy, with 2.7° and 3.6° of mean variance, and Observer 2 (a "new" spine fellow and the least experienced observer) was last in rank, with a mean variance of 5.0°. Contrary to Loder et al., we found significant differences between some of the observers (p £ 0.05, paired t test).

Interobserver Variance

Because each observer measured each curve twice, there were two sets of data for the range of measurements of the sixty-nine curves. On the first assessments of one curve, which was small with easily determined end plates for measurement, the lowest measurement (20°) and the highest (22°) varied by only 2°. On the second assessments of this curve, the lowest measurement was 20° and the highest was 23°, again an extremely tight variability. Thus, the mean range of variation of this particular curve was 2.5°, or ± 1.25°.

On the first assessments of another curve, which was very large, there was 12° of variance between the lowest measurement (105°) and the highest (117°). On the second assessments, there was 31° of variance between the lowest (83°) and highest (114°) measurements. This considerable difference between the lowest and highest measurements on the second assessment was due, however, to a single outlying measurement by Observer 2, who recorded the curve at 110° the first time and 83° the second time. If this outlier measurement is discarded, the measurements by the other six observers were very close (range, 108° to 114°). For the purposes of this review, the outlier measurement was included.

The interobserver variance was estimated by first calculating the standard deviation for each of the 138 (two times sixty-nine) groups of seven measurements and then averaging these values. The interobserver variance was estimated to be 3.35° (95% confidence limit, ± 7.86°). Thus, one can expect the difference between two observers to be less than 8° 95% of the time.

Analysis of Measurement Accuracy versus Curve Magnitude

Two analyses were conducted related to curve magnitude and measurement accuracy. The intraobserver variance increased as the magnitude of the curve increased (p = 0.040). In general, the intraobserver variance increased by 0.2° for every increase of 10° in the curve (Fig. 1). Similarly, the variability of the interobserver variance increased as the curve magnitude increased. Variance increased by approximately 0.3° for every 10° increase in curve magnitude (p = 0.016) (Fig. 2).

Discussion

Introduction | Materials and Methods | Results | Discussion | References

The present study, in which sixty-nine curves in patients with congenital scoliosis were measured twice (at least three weeks apart) by seven different observers using methodology that was identical to that used by Loder et al.⁹, showed substantial differences from the results of Loder et al.

Our overall 95% confidence limit for intraobserver variance was ±3° compared with ±9.6° in the study by Loder et al.⁹. We also noted a significant difference between some observers, which they did not. The mean intraobserver variance ranged from ± 5.0° for our least experienced observer to ± 1.9° for the most experienced observer.

Our overall 95% confidence limit for interobserver variability was ±7.86° compared with ±11.8° in the study by Loder et al.⁹. Interobserver variability is of less clinical importance than intraobserver variability because, in the usual clinical situation, the same observer measures the two comparative radiographs.

We also noted a correlation between curve magnitude and measurement accuracy. Generally speaking, the larger the curve, the more variable the measurement accuracy.

This study could be criticized for the preselection of the end vertebrae to be used for measurement, as an important variable was thus removed. However, we had no choice but to perform this preselection if we were to be able to compare our study with that by Loder et al.⁹. If Loder et al. had not preselected the end vertebrae, we would not have done so either.

We are aware that if the end vertebrae were not preselected and if the two radiographs were measured on different occasions, then the intraobserver and interobserver variability would be higher. However, in the usual clinical situation, the surgeon has the opportunity to simultaneously examine and measure the current radiograph and the previous radiograph or radiographs. The situation thus allows the surgeon to select the same end vertebrae on both radiographs and thus determine with a 95% confidence limit of ± 3° whether progression has taken place.

We admit that it is possible that a surgeon might make a judgment without viewing the two radiographs simultaneously, but that would be foolish medicine and beyond the scope of this review. We also acknowledge that it is possible for a surgeon to have the two radiographs simultaneously on the view box but select different end vertebrae on the two radiographs. That also would be foolish medicine and beyond the scope of this review. We think that, in reality, it is the failure to take the time to carefully and precisely measure the two radiographs simultaneously that leads to errors.

In conclusion, in this study of fifty patients with sixty-nine congenital scoliosis curves, we recorded the intraobserver and interobserver variances in curve measurements as performed by seven individuals. The accuracy of curve measurement appears to depend on three variables: the clarity of the measurement points on a given radiograph, the magnitude of the curve, and the experience of the person doing the measurements. Contrary to the findings by Loder et al.⁹, our overall 95% confidence limits were 2.5° (lower) to 3.2° (upper) for the intraobserver variance (±3° overall). This means that, for a difference between two radiographs to indicate a true curve change and not be due to measurement variability, a change of 6° is required to meet the 95% confidence limits. Our overall 95% confidence limit was ±7.86° for the interobserver variance. One must remember that this is a statistical average, and for any given situation the number may be higher or lower according to the above three variables.

Note: The authors thank Ms. Jill Wroblewski, Research Director of Twin Cities Spine Center, for her statistical support.

References

Introduction | Materials and Methods | Results | Discussion | References

Cobb JR: Outline for the study of scoliosis. Instr Course Lect,1948.5: 261-75, 5261 1948

Carman DL; Browne RH; and Birch JG: Measurement of scoliosis and kyphosis radiographs. Intraobserver and interobserver variations. J Bone Joint Surg Am,1990.72: 328-33, 72328 1990 [PubMed]

Dawson EG; Smith RK; and McNiece GM: Radiographic evaluation of scoliosis. A reassessment and introduction of the scoliosis chariot. Clin Orthop,1978.131: 151-5, 131151 1978 [PubMed]

DeSmet AA; Goin JE; Asher MA; and Scheuch HG: A clinical study of the differences between the scoliotic angles measured on posteroanterior and anteroposterior radiographs. J Bone Joint Surg Am,1982.64: 489-93, 64489 1982 [PubMed]

Goldberg MS; Poitras B; Mayo NE; Labelle H; Bourassa R; and Cloutier R: Observer variation in assessing spinal curvature and skeletal development in adolescent idiopathic scoliosis. Spine,1988.13: 1371-7, 131371 1988 [PubMed]

Goldsmith G; Morrissy R; Hall CE; Kehl D; and Cowie H: Accuracy of the Cobb technique in scoliosis measurements. Orthop Trans,1987.11: 32, 1132 1987

Oda M; Rauh S; Gregory PB; Silverman FN; and Bleck EE: The significance of roentgenographic measurement in scoliosis. J Pediatr Orthop,1982.2: 378-82, 2378 1982 [PubMed]

Sevastikoglou JA, and Bergquist E: Evaluation of the reliability of radiological methods for registration of scoliosis. Acta Orthop Scand,1969.40: 608-13, 40608 1969 [PubMed]

Loder RT; Urquhart A; Steen H; Graziano G; Hensinger RN; Schlesinger A; Schork MA; and Shyr Y: Variability in Cobb angle measurements in children with congenital scoliosis. J Bone Joint Surg Br,1995.77: 768-70, 77768 1995 [PubMed]

View Large | Download Slide(.ppt)
Add to My JBJS

Fig. 1:: In general, the intraobserver variance increased by 0.2° for every increase of 10° in curve magnitude (p = 0.040).

View Large | Download Slide(.ppt)
Add to My JBJS

Fig. 2:: In general, the interobserver variance increased by 0.3° for every increase of 10° in curve magnitude (p = 0.016).

View Larger

TABLE I: Intraobserver Variance: Summation of Seven Observers

Observer	Total Variance*†	Mean Variance*‡	Standard Error of Mean*	Minimum Variance*	Maximum Variance*	Lower 95% Confidence Limit*	Upper 95% Confidence Limit*	Rank of Observer
1	163	2.4	0.315	0	7	1.948	3.183	4
2	347	5.0	0.671	0	27	3.757	6.388	7
3	158	2.3	0.346	0	16	1.597	2.954	3
4	185	2.7	0.448	0	11	1.904	3.661	5
5	248	3.6	0.500	0	20	2.556	4.517	6
6	130	1.9	0.161	0	7	1.582	2.215	1
7	131	1.9	0.200	0	6	1.492	2.276	2
Overall§	1049	2.8	0.162	0	27	2.542	3.177

Add to My JBJS

TABLE I: Intraobserver Variance: Summation of Seven Observers

Observer	Total Variance*†	Mean Variance*‡	Standard Error of Mean*	Minimum Variance*	Maximum Variance*	Lower 95% Confidence Limit*	Upper 95% Confidence Limit*	Rank of Observer
1	163	2.4	0.315	0	7	1.948	3.183	4
2	347	5.0	0.671	0	27	3.757	6.388	7
3	158	2.3	0.346	0	16	1.597	2.954	3
4	185	2.7	0.448	0	11	1.904	3.661	5
5	248	3.6	0.500	0	20	2.556	4.517	6
6	130	1.9	0.161	0	7	1.582	2.215	1
7	131	1.9	0.200	0	6	1.492	2.276	2
Overall§	1049	2.8	0.162	0	27	2.542	3.177

Cobb JR: Outline for the study of scoliosis. Instr Course Lect,1948.5: 261-75, 5261 1948

Carman DL; Browne RH; and Birch JG: Measurement of scoliosis and kyphosis radiographs. Intraobserver and interobserver variations. J Bone Joint Surg Am,1990.72: 328-33, 72328 1990 [PubMed]

Dawson EG; Smith RK; and McNiece GM: Radiographic evaluation of scoliosis. A reassessment and introduction of the scoliosis chariot. Clin Orthop,1978.131: 151-5, 131151 1978 [PubMed]

Goldsmith G; Morrissy R; Hall CE; Kehl D; and Cowie H: Accuracy of the Cobb technique in scoliosis measurements. Orthop Trans,1987.11: 32, 1132 1987

Oda M; Rauh S; Gregory PB; Silverman FN; and Bleck EE: The significance of roentgenographic measurement in scoliosis. J Pediatr Orthop,1982.2: 378-82, 2378 1982 [PubMed]

Sevastikoglou JA, and Bergquist E: Evaluation of the reliability of radiological methods for registration of scoliosis. Acta Orthop Scand,1969.40: 608-13, 40608 1969 [PubMed]

Accreditation Statement

These activities have been planned and implemented in accordance with the Essential Areas and policies of the Accreditation Council for Continuing Medical Education (ACCME) through the joint sponsorship of the American Academy of Orthopaedic Surgeons and The Journal of Bone and Joint Surgery, Inc. The American Academy of Orthopaedic Surgeons is accredited by the ACCME to provide continuing medical education for physicians.

CME Activities Associated with This Article

Submit a Comment

Please read the other comments before you post yours. Contributors must reveal any conflict of interest.
Comments are moderated and will appear on the site at the discretion of JBJS editorial staff.

* = Required Field

Comment Author(s) * (if multiple authors, separate names by comma)

Example: John Doe

Affiliation & Institution*

Comment Title*

Comment*

Cancel

Print
PDF
Email

Recipient(s) will receive an email with a link (good for 5 Days) to 'Measurement Accuracy in Congenital Scoliosis' and do not need to have The Journal of Bone & Joint Surgery account to access the content.
Your Name:*
Example: John Doe

Email Address:* CC Me:
Enter your valid email address. Example: jdoe@example.com

Recipient's Email Address:*

Separate multiple email address with semi-colons (up to 5).

Subject:* 's The Journal of Bone & Joint Surgery: 'Measurement Accuracy in Congenital Scoliosis'
Subject for your email.

Message:

(Optional, message will truncate at 1000 characters)

Processing your request... Please Wait...

Copyright © in the material you requested is held by The Journal of Bone & Joint Surgery, Inc. (unless otherwise noted). This email ability is provided as a courtesy, and by using it you agree that that you are requesting the material solely for personal, non-commercial use, and that it is subject to Journal of Bone & Joint Surgery, Inc.'s Terms of Use. The information provided in order to email this topic will not be used to send unsolicited email, nor will it be furnished to third parties. Please refer to The Journal of Bone & Joint Surgery Inc.'s Privacy Policy for further information.

Copyright © The Journal of Bone & Joint Surgery Inc. All rights reserved.
Share
Get Citation

Fernando A.M. Facanha-Filho, Robert B. Winter, John E. Lonstein, Steven Koop, Thomas Novacheck, E. A. L’Heureux, Cheryl A. Noren; Measurement Accuracy in Congenital Scoliosis. The Journal of Bone & Joint Surgery. 2001 Jan;83(1):42-42.

Download citation file:

RIS (Zotero)

EndNote

BibTex

Medlars

ProCite

RefWorks

Reference Manager

Copyright © The Journal of Bone and Joint Surgery, Inc.
Rights & Permissions
Sign Up for Alerts on this
Article

Processing your request... Please Wait.
Submit a Comment