SPINE Volume 39, Number 1, pp E26-E34 ©2013, Lippincott Williams & Wilkins
HEALTH SERVICES RESEARCH
Reliability and Validity of the French-Canadian Version of the Scoliosis Research Society 22 Questionnaire in France Guillaume Lonjon, MD,* Brice Ilharreborde, MD, PhD,* Thierry Odent, MD, PhD,† Sébastien Moreau, MD,* Christophe Glorion, MD,† and Keyvan Mazda, MD*
Study Design. Outcome study to determine the internal consistency, reproducibility, and concurrent validity of the FrenchCanadian version of the Scoliosis Research Society 22 (SRS-22 fcv) patient questionnaire in France. Objective. To determine whether the SRS-22 fcv can be used in a population from France. Summary of Background Data. The SRS-22 has been translated and validated in multiple countries, notably in the French-Canadian language in Quebec, Canada. Use of SRS-22 fcv seems appropriate for evaluating adolescent idiopathic scoliosis in France. However, French-Canadian French is noticeably different from the French spoken in France, and no study has investigated the use of a FrenchCanadian version of a health-quality questionnaire in another French population. Methods. The methods used for validating the SRS-22 fcv in Quebec were adopted for use with a group of 200 adolescents with idiopathic scoliosis and 60 healthy adolescents in France. Reliability and reproducibility were measured by the Cronbach α and intraclass correlation coefficient (ICC), construct validity by factorial analysis, concurrent validity by the Short-Form of the survey, and discriminant validity by analysis of variance and multivariate linear regression. Results. In France, the SRS-22 fcv showed good global internal consistency (Cronbach α = 0.87, intraclass correlation coefficient = 0.92), a coherent factorial structure, and high correlation coefficients between the SRS-22 fcv and Short-Form of the survey (P < 0.001). From the *Peditaric Orthopaedic Department, Robert Debré Hospital, Assistance Publique-Hôpitaux de Paris, Paris Diderot University, Paris, France; and †Peditaric Orthopaedic Department, Necker Hospital, Assistance Publique-Hôpitaux de Paris, Paris Descartes University, Paris, France. Acknowledgment date: March 15, 2013. First revision date: August 10, 2013. Acceptance date: August 23, 2013. The manuscript submitted does not contain information about medical device(s)/drug(s). No funds were received in support of this work. Relevant financial activities outside the submitted work: consultancy and royalties. Address correspondence and reprint requests to Guillaume Lonjon, MD, Pediatric Orthopaedic Department, Robert Debré Hospital, AP-HP, ParisDiderot University, 48 bd Serurier, 75019 Paris, France; E-mail: dr.guillaume [email protected] DOI: 10.1097/BRS.0000000000000080
E26
www.spinejournal.com
However, reliability and validity were slightly less than that for the instrument’s original validation and the validation of the SRS-22 fcv in Quebec. These differences could be explained by language and cultural differences. Conclusion. The SRS-22 fcv is relevant for use in France, but further development and validation of a specific French questionnaire remain necessary to improve the assessment of functional outcomes of adolescents with scoliosis in France. Key words: idiopathic scoliosis, health-related quality of life, SRS-22, questionnaire, validity, French, French-Canadian. Level of Evidence: N/A Spine 2014;39:E26–E34
A
dolescent idiopathic scoliosis (AIS) is the most common spinal disease in the pediatric population, negatively affecting patient quality of life.1–7 Haher et al8 and the Scoliosis Research Society (SRS) developed and validated a simple, practical questionnaire to facilitate assessment of the outcome of idiopathic scoliosis. The latest version of this questionnaire (SRS-22) contains 22 questions covering 5 domains: pain, self-image, function, mental health, and satisfaction with management. The SRS-22 has been shown to have high reliability and concurrent validity in patients with scoliosis.9–13 After the initial English version of the SRS-22 was validated, several authors published reliability and validity results for translated and adapted versions in other languages.14–23 In 2009, Beauséjour et al24 developed a French-Canadian version that was validated in Quebec. That version was used by several centers for clinical evaluation. However, Beauséjour et al24 did not validate this version of SRS-22 for Frenchspeaking countries in Europe (France, Switzerland, Belgium) and the literature reports no French-Canadian version of other clinical assessment tools in France. Actually, for native French speakers in France, particularly adolescents, Canadian French is noticeably different from their French. Moreover, besides language differences, cultural differences exist between Quebec and France in terms of health care resources, systems, and insurance coverage.25–28 January 2014
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE130351.indd E26
06/12/13 1:40 PM
HEALTH SERVICES RESEARCH In this background, the aim of this study was to determine whether the French-Canadian version of the SRS-22 (SRS-22 fcv) could be used for adolescents in France.
MATERIALS AND METHODS The methods described by Beauséjour et al24 in validating the SRS-22 fcv in Quebec were used, except for translation and cross-cultural adaptation step. The validation of the SRS-22 has been well described in several foreign validation studies and was based on study of reliability and validity.14–23,29 Reliability concerns the extent to which an experiment, test, or any measuring procedure yields the same results on repeated trials. Validity defines the degree to which an experiment, text or any measuring procedures measures what it is supposed to measure (Figure 1).30,31
Study Population All consecutive adolescents (11–18 yr) seen for follow-up of AIS, in 2 outpatient orthopedic clinics in Paris between March and September 2011 were included. Patients with previous scoliosis surgery and non-native French speakers were not included. Patients were seen in outpatient clinics for follow-up of conservative treatment (observation or brace) or to schedule surgery. All eligible patients completed the SRS-22 fcv in the waiting room of the clinic (see Supplemental Digital Content Appendix 1 available at http://links.lww.com/BRS/A840). According to the protocol put forth by Beauséjour et al,24 a subgroup of patients completed the Short-Form of the survey (SF-12) quality-of-life questionnaire.32,33 To test temporal reliability, another subgroup of patients received a second identical questionnaire (SRS-22 fcv) with a stamped return
Figure 1. Schema of validity and reliability explanations. Spine
SRS-22 fcv in France • Lonjon et al
envelope for return within 2 weeks. To limit the number of questionnaires for each patient and increase concentration, an unequal random selection was made to designate which patients would complete the SF-12 and which patients would complete a second SRS-22 fcv. Patients received an explanation of the origin of the questionnaire and the study purpose but no help in answering the questionnaire. Patients were distributed into 2 groups: (1) adolescents with “nonclinically significant scoliosis” (NCSS) (curve angle ≤10° according to the Cobb angle on anteroposterior radiographs), and (2) adolescents with AIS according to the SRS criteria34 (lateral deviation of the spine more than 10° according to the Cobb angle on anteroposterior radiographs). A third group of similar-aged healthy adolescents, recruited from a public high school in Paris and without back problems, completed the SRS-22 fcv in the classroom without assistance.
Analysis First, each item of the SRS-22 fcv was scored from 1 (worst) to 5 (excellent). For each domain, for the subtotal score (the subtotal score is calculated with items 1 to 20, without satisfaction management domains) and for the total score, the average value and floor and ceiling effects (percentage of participants showing the minimum and maximum possible scores, respectively) were calculated. Second, the reliability of the SRS-22 fcv was assessed (internal consistency and temporal reliability). Internal consistency was assessed by calculating the Cronbach α for each domain and for the total score for the 3 groups. The internal consistency was considered good with Cronbach α value more than 0.6.35 Then in the test-retest method, temporal reliability was measured by calculating the intraclass correlation coefficient (ICC) with an estimate of measurement error (Standard Error of measurement: SEm). An ICC value less than 0.2 indicates poor agreement, 0.2 to 0.3 fair agreement, 0.4 to 0.5 moderate agreement, 0.6 to 0.8 strong agreement, and more than 0.8 almost perfect agreement. Third for validity, construct validity was assessed. Convergent and discriminant validity are the 2 subtypes of validity that make up construct validity. Convergent validity refers to the degree to which 2 measures of constructs that theoretically should be related, are in fact related. Discriminative validity, refer to the ability of a tool to distinguish among different categories of subjects.36 For convergent validity, Pearson correlation coefficients were calculated for the corresponding domain scores of the SRS-22 fcv and SF-12. According to previous studies, the domains “function” and “pain” of the SRS-22 correspond to the domains “physical function,” “role-physical,” and “pain” of the SF-12, the domain “self-image” of the SRS22 corresponds to the domain “general health perceptions” of the SF-12 and the domain “mental health” of SRS-22 to domains “role emotional” and “mental health” of SF-12.15,20,24 Correlation values of 0.20 or more were considered at least fair. Then, a factor analysis was performed. Factor analysis investigating the correlation between items involved principal component analysis and varimax rotation factor to determine whether the SRS-22 fcv revealed a structure similar to the www.spinejournal.com
E27
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE130351.indd E27
06/12/13 1:40 PM
HEALTH SERVICES RESEARCH
SRS-22 fcv in France • Lonjon et al
original version of the instrument (items selection for loading >0.45). Here, we used data for the AIS group for 4 factors only and the items representing each factor and the explained variance. For discriminative validity, a univariate analysis (student test, analysis of variance) of the relationship between clinical variables and questionnaire scores was conducted under the assumption that unfavorable characteristics are associated with poor scores. The most relevant general characteristics for all participants were used for this univariate analysis: sex, age, age at diagnosis, body mass index, and radiographical indices (Cobb angle of the main spine curve, type of scoliosis, Risser
score for skeletal maturity [0–3, immature; 4 and 5, mature] and type of treatment [patients under observation, patients treated by a brace, and patients scheduled for surgery]). A multiple regression analysis of independent variables (P < 0.2) (stepwise with backward direction) was then performed.
Statistical Analysis Statistical analysis involved use of R v13.0 (http://www.Rproject.org, the R Foundation for Statistical Computing, Vienna, Austria). P < 0.05 was considered statistically significant.
TABLE 1. Characteristics of Adolescents With NCSS and AIS and Healthy Adolescents Who
Completed the SRS-22 fcv Questionnaire NCSS n = 25
AIS n = 175
Healthy Adolescents n = 60
Male
5 (20%)
32 (18.3%)
13 (21.7%)
Female
20 (80%)
143 (81.7%)
47 (78.3%)
13.8 (12.8–14.6) (12–18.6)
14.4 (13.7–15.7) (10.1–18.9)
14.7 (13.7–16.2) (12.7–18.3)
11 yr Body mass index (kg/m ) median (Q1, Q3) (range) 2
17.8 (16.8–19.2) (13.2–28.7) 19.3 (17.8–21.3) (13.0–30.9)
19.2 (17.6–20.4) (13.3–27.1)
Type of scoliosis Simple thoracic
71 (40.6%)
Thoracolumbar
46 (26.3%)
Lumbar
38 (21.7%)
Double main
20 (11.4%)
Cobb angle (°) median (Q1, Q3) (range)
8 (7–9) (6–10)
28 (19–42) (11–75)
0
9 (36%)
32 (18.3%)
1
5 (20%)
16 (9.1%)
2
1 (4%)
13 (7.4%)
3
2 (8%)
32 (18.3%)
4
7 (28%)
60 (34.3%)
5
1 (4%)
22 (12.6%)
25 (100%)
42 (24%)
Cast
0
102 (58.3%)
Surgery scheduled
0
31 (17.7%)
Risser score
Treatment Observation
NCSS indicates nonclinically significant scoliosis; AIS, adolescent idiopathic scoliosis; SRS-22 fcv, French-Canadian version of the Scoliosis Research Society 22.
E28
www.spinejournal.com
January 2014
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE130351.indd E28
06/12/13 1:40 PM
HEALTH SERVICES RESEARCH
SRS-22 fcv in France • Lonjon et al
TABLE 2. Descriptive Statistics for Domains of the SRS-22 fcv NCSS n = 25
AIS n = 175
Healthy Adolescents n = 60
4.46 (0.43)
4.19 (0.63)
4.67 (0.46)
% floor
0
0
0
% ceiling
24
19.4
41.7
3.79 (0.64)
3.54 (0.64)
4.18 (0.63)
% floor
0
0
0
% ceiling
0
1.1
15
4.62 (0.62)
4.43 (0.63)
4.67 (0.41)
% floor
0
0
0
% ceiling
0
0
0
3.98 (0.69)
3.77 (0.77)
3.97 (0.68)
% floor
0
0
0
% ceiling
8
8
6.7
4.07 (0.63)
3.68 (0.82)
…
% floor
0
1.2
…
% ceiling
13
9.4
…
4.21 (0.43)
3.98 (0.5)
4.37 (0.44)
% floor
0
0
0
% ceiling
0
0.6
1.7
4.2 (0.41)
3.96 (0.47°
…
% floor
0
0
…
% ceiling
0
0.6
…
Domains (Items) Pain (items 1, 2, 8, 11, 17) Mean score (SD)
Self-image (items 3, 7, 13, 16, 20) Mean score (SD)
Function (items 4, 6, 10, 14, 19) Mean score (SD)
Mental health (items 5, 9, 12, 15, 18) Mean score (SD)
Satisfaction with management (items 21, 22) Mean score (SD)
Subtotal (items 1–20) Mean score (SD)
Total mean (SD) (items 1–22) Mean score (SD)
The subtotal score is calculated with items 1 to 20. NCSS indicates nonclinically significant scoliosis; AIS, adolescent idiopathic scoliosis; SRS-22 fcv, French-Canadian version of the Scoliosis Research Society 22.
RESULTS Patients Demographic characteristics of participants are summarized in Table 1. Among 218 eligible patients seen for scoliosis, 200 patients agreed to complete the questionnaire: 175 with AIS and 25 with NCSS. Among the 150 patients who received the SF-12, 136 completed it (119 AIS and 17 NCSS) and of the 50 patients who received a second SRS-22 fcv, 38 sent it back (33 AIS and 5 NCSS). Thanks to random, subgroups of patients selected to receive the SF-12 or the second SRS-22 Spine
fcv were representative (P > 0.05) of general population in terms of sex ratio, age, body mass index, Risser score, Cobb angle, type of curve, type of scoliosis and results (total score). In addition, 60 healthy adolescents completed the SRS-22 fcv. Girls represented 210 of the 260 participants. The mean age was 14.4 years (range, 10.1–18.9 yr), and the mean body mass index was 19.1 kg/m2 (range, 13–30.8 kg/m2). The mean magnitude of the main spinal curve (Cobb angle) for patients with AIS was 28° (range, 6°–75°). Of the 200 patients with AIS or NCSS, 31 were scheduled for surgical correction, 102 had received a brace, and 67 were under observation. www.spinejournal.com
E29
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE130351.indd E29
06/12/13 1:40 PM
HEALTH SERVICES RESEARCH
SRS-22 fcv in France • Lonjon et al
TABLE 3. Convergent Validity Between Relevant Domains of the SRS-22 fcv and the SF-12 for Patient
With AIS or NCSS (SF-12)
SRS-22 Function
Pain
Self-image
Mental Health
Satisfaction With Management
Physical function
0.27
0.29
0.30
0.10
0.08
Role-physical
0.36
0.39
0.21
0.30
0.44
Pain index
0.32
0.41
0.17
0.29
0.08
n = 136 SF-12
General health perceptions
0.21
0.16
0.38
0.29
0.23
Vitality
0.12
0.23
0.22
0.30
0.19
Social functioning
0.34
0.24
0.36
0.46
0.17
Role emotional
0.35
0.32
0.39
0.48
0.22
Mental health
0.20
0.26
0.29
0.62
0.37
Pearson correlation coefficients in boldface are for conceptually corresponding domains. Correlations italicized are not statistically significant (P > 0.05). NCSS indicates nonclinically significant scoliosis; AIS, adolescent idiopathic scoliosis; SRS-22 fcv, French-Canadian version of the Scoliosis Research Society 22; SF-12, Short-Form 12.
Analysis of the domain scores for the SRS-22 fcv (Table 2) revealed a ceiling effect for pain, not expressing the clinical diversity of patients with little or no pain.
Reliability Internal consistency of the SRS-22 fcv was good, with Cronbach α 0.82 for patients with AIS, 0.79 for patients with NCSS, and 0.93 for healthy adolescents. For patients with AIS, good reliability was found for all domains: pain, 0.71; self-image, 0.61; function, 0.60; mental health, 0.73; and satisfaction with management, 0.60. In all, 34 of 38 patients completed the second SRS-22 fcv. The ICC for the 4 domains combined and separate demonstrated excellent test-retest reproducibility (total score ICC = 0.90, SEm = 0.31; pain domain ICC = 0.93, SEm = 0.50; self-image domain ICC = 0.89, SEm = 0.53; function ICC = 0.86, SEm = 0.47; mental health domain ICC = 0.85, SEm = 0.53).
Validity For convergent validity, the correlations between the SRS-22 fcv domains and the relevant SF-12 domains (in bold in
Table 3) was on the range from 0.27 to 0.62 (P < 0.001), considered at least fair. Consequently, all of the convergent validity hypothesizes were supported by the correlation results. For construct validity, questions related to satisfaction with treatment (items 21 and 22) were not included in the factor analysis because this dimension was not considered an aspect of the disease. Factor analysis of the 20 items produced 4 factors that explained 44% of the variance (Table 4). These factors corresponded to the pre-established dimensions of the original questionnaire, with some exceptions. Items 7 (factor loading = 0.63), 12 (factor loading = 0.80), 14 (factor loading = 0.78), 17 (factor loading = 0.71), 18 (factor loading = 0.51), and 20 (factor loading = 0.53) originally located in a domain, were related to another factor on factor analysis. Item 11 was not located in a domain. For construct validity, univariate analysis found several variables statistically associated with a poor score (see Supplemental Digital Content Appendix 2 available at http://links. lww.com/BRS/A840 ). The total score was greater for healthy adolescents than for patients with AIS (P = 0.018) (Figure 2) and for NCSS than for patients with AIS (P < 0.001). Older
TABLE 4. Factorial Structure of the SRS-22 fcv With 4 domains for Patients With AIS Factors
Variance, %
Factor Items*
SRS Domains
SRS Items
1
19
4-6-10-18-20
Self-image
4-6-10-14-19
2
14
3-12-13-14-16-17
Mental health
3-7-13-16-20
3
11
1-2-8
Pain
1-2-8-11-17
4
10
5-7-9-15
Function
5-9-12-15-18
*Items with factor loading > 0.45. Items in boldface were not found in the relevant factors; AIS, adolescent idiopathic scoliosis. SRS-22 fcv indicates French-Canadian version of Scoliosis Research Society 22.
E30
www.spinejournal.com
January 2014
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE130351.indd E30
06/12/13 1:40 PM
HEALTH SERVICES RESEARCH
Figure 2. Box plot figure of SRS-22 total score of healthy patient and patient with AIS. AIS indicates adolescent idiopathic scoliosis; SRS-22, Scoliosis Research Society 22.
patients, patients with obesity, patients with a large Cobb angle, patients with a double main curve, and patients scheduled for surgery had minor scores. Figures 3 and 4 show in a box plot format the mean scores for type of curve and type of treatment. Patients with a double main curve and patients scheduled for surgery had minor score. After multivariate analysis, only 2 clinical variables were independently associated with SRS-22 fcv total score (Cobb angle curve: P = 0.03; patients scheduled for surgery: P = 0.04) (see Supplemental Digital Content Appendix 3 available at http://links.lww .com/BRS/A840).
DISCUSSION To the authors’ knowledge, use of a previously adapted and validated French-Canadian version of a health-quality questionnaire in France has never been investigated. The aim of this study was to determine whether the SRS-22 fcv could be used in France and specifically in an adolescent population. In France, the SRS-22 fcv showed good internal reliability (Cronbach α = 0.81), good temporal reliability (ICC
Figure 3. Box plot figure of SRS-22 total score for different types of curve. SRS-22 indicates Scoliosis Research Society 22. Spine
SRS-22 fcv in France • Lonjon et al
Figure 4. Box plot figure of SRS-22 total score for different types of treatment. SRS-22 indicates Scoliosis Research Society 22.
0.90), and significant convergent validity between SRS-22 fcv domains and corresponding SF-12 domains (P < 0.001). For construct validity, factor analysis revealed 4 coherent factors and multivariate regression analysis showed good correlation between Cobb angle of the main spinal curve and total SRS-22 fcv score, and between type of treatment and total SRS-22 score. However the correlation between Cobb angle and SRS-22 score was not very strong (β = −0.06; P = 0.03). This might be explained by a nonlinear relationship.37 For divergent validity, as expected unfavorable characteristics (double main curve and patient scheduled for surgery) are associated with poor score. Results were consistent with prior observations,17–22 and internal reliability of this study was consistent with similar evaluations of the SRS-22 in other languages in the literature (Table 5). Thus, the SRS-22 fcv can be used in France. Nevertheless, as for other adapted versions,17–19,21,22 validity and reliability of the SRS-22 fcv were slightly inferior to those of the original version. Regarding internal reliability, although the coefficient values were lower than those of the original SRS-22 (0.21 less for pain, 0.14 less for self-image, 0.26 less for function, 0.17 less for mental health, and 0.28 less for satisfaction with management domains), they still equaled or exceeded the acceptability threshold of 0.6. Several reasons might explain these differences. First, item 15 in the function domain (Cronbach α = 0.60), representing financial difficulties, is not relevant in France because the French health care system provides 100% of the fees for AIS. This question, with previously observed difficulties,17,18,21 should not be presented to patients in France so as not to overestimate the mean score. Second, item 19, attractiveness in the self-image domain (Cronbach α = 0.61), was not adapted for patients younger than 13 years.24 In this study, 20% of the participants were in this age group. Patients younger than 13 years should not complete this question to avoid soliciting random answers. Regarding convergent validity, all correlations among related domains of the SF-12 were statistically significant www.spinejournal.com
E31
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE130351.indd E31
06/12/13 1:40 PM
HEALTH SERVICES RESEARCH
SRS-22 fcv in France • Lonjon et al
TABLE 5. Reliability (Cronbach α Coefficients) for the SRS-22 fcv in France (This Study) and Previous
Adaptations
Current Study (10––18 yr)
Quebec (10––21 yr) (Beauséjour et al)24
Original (
HEALTH SERVICES RESEARCH
Reliability and Validity of the French-Canadian Version of the Scoliosis Research Society 22 Questionnaire in France Guillaume Lonjon, MD,* Brice Ilharreborde, MD, PhD,* Thierry Odent, MD, PhD,† Sébastien Moreau, MD,* Christophe Glorion, MD,† and Keyvan Mazda, MD*
Study Design. Outcome study to determine the internal consistency, reproducibility, and concurrent validity of the FrenchCanadian version of the Scoliosis Research Society 22 (SRS-22 fcv) patient questionnaire in France. Objective. To determine whether the SRS-22 fcv can be used in a population from France. Summary of Background Data. The SRS-22 has been translated and validated in multiple countries, notably in the French-Canadian language in Quebec, Canada. Use of SRS-22 fcv seems appropriate for evaluating adolescent idiopathic scoliosis in France. However, French-Canadian French is noticeably different from the French spoken in France, and no study has investigated the use of a FrenchCanadian version of a health-quality questionnaire in another French population. Methods. The methods used for validating the SRS-22 fcv in Quebec were adopted for use with a group of 200 adolescents with idiopathic scoliosis and 60 healthy adolescents in France. Reliability and reproducibility were measured by the Cronbach α and intraclass correlation coefficient (ICC), construct validity by factorial analysis, concurrent validity by the Short-Form of the survey, and discriminant validity by analysis of variance and multivariate linear regression. Results. In France, the SRS-22 fcv showed good global internal consistency (Cronbach α = 0.87, intraclass correlation coefficient = 0.92), a coherent factorial structure, and high correlation coefficients between the SRS-22 fcv and Short-Form of the survey (P < 0.001). From the *Peditaric Orthopaedic Department, Robert Debré Hospital, Assistance Publique-Hôpitaux de Paris, Paris Diderot University, Paris, France; and †Peditaric Orthopaedic Department, Necker Hospital, Assistance Publique-Hôpitaux de Paris, Paris Descartes University, Paris, France. Acknowledgment date: March 15, 2013. First revision date: August 10, 2013. Acceptance date: August 23, 2013. The manuscript submitted does not contain information about medical device(s)/drug(s). No funds were received in support of this work. Relevant financial activities outside the submitted work: consultancy and royalties. Address correspondence and reprint requests to Guillaume Lonjon, MD, Pediatric Orthopaedic Department, Robert Debré Hospital, AP-HP, ParisDiderot University, 48 bd Serurier, 75019 Paris, France; E-mail: dr.guillaume [email protected] DOI: 10.1097/BRS.0000000000000080
E26
www.spinejournal.com
However, reliability and validity were slightly less than that for the instrument’s original validation and the validation of the SRS-22 fcv in Quebec. These differences could be explained by language and cultural differences. Conclusion. The SRS-22 fcv is relevant for use in France, but further development and validation of a specific French questionnaire remain necessary to improve the assessment of functional outcomes of adolescents with scoliosis in France. Key words: idiopathic scoliosis, health-related quality of life, SRS-22, questionnaire, validity, French, French-Canadian. Level of Evidence: N/A Spine 2014;39:E26–E34
A
dolescent idiopathic scoliosis (AIS) is the most common spinal disease in the pediatric population, negatively affecting patient quality of life.1–7 Haher et al8 and the Scoliosis Research Society (SRS) developed and validated a simple, practical questionnaire to facilitate assessment of the outcome of idiopathic scoliosis. The latest version of this questionnaire (SRS-22) contains 22 questions covering 5 domains: pain, self-image, function, mental health, and satisfaction with management. The SRS-22 has been shown to have high reliability and concurrent validity in patients with scoliosis.9–13 After the initial English version of the SRS-22 was validated, several authors published reliability and validity results for translated and adapted versions in other languages.14–23 In 2009, Beauséjour et al24 developed a French-Canadian version that was validated in Quebec. That version was used by several centers for clinical evaluation. However, Beauséjour et al24 did not validate this version of SRS-22 for Frenchspeaking countries in Europe (France, Switzerland, Belgium) and the literature reports no French-Canadian version of other clinical assessment tools in France. Actually, for native French speakers in France, particularly adolescents, Canadian French is noticeably different from their French. Moreover, besides language differences, cultural differences exist between Quebec and France in terms of health care resources, systems, and insurance coverage.25–28 January 2014
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE130351.indd E26
06/12/13 1:40 PM
HEALTH SERVICES RESEARCH In this background, the aim of this study was to determine whether the French-Canadian version of the SRS-22 (SRS-22 fcv) could be used for adolescents in France.
MATERIALS AND METHODS The methods described by Beauséjour et al24 in validating the SRS-22 fcv in Quebec were used, except for translation and cross-cultural adaptation step. The validation of the SRS-22 has been well described in several foreign validation studies and was based on study of reliability and validity.14–23,29 Reliability concerns the extent to which an experiment, test, or any measuring procedure yields the same results on repeated trials. Validity defines the degree to which an experiment, text or any measuring procedures measures what it is supposed to measure (Figure 1).30,31
Study Population All consecutive adolescents (11–18 yr) seen for follow-up of AIS, in 2 outpatient orthopedic clinics in Paris between March and September 2011 were included. Patients with previous scoliosis surgery and non-native French speakers were not included. Patients were seen in outpatient clinics for follow-up of conservative treatment (observation or brace) or to schedule surgery. All eligible patients completed the SRS-22 fcv in the waiting room of the clinic (see Supplemental Digital Content Appendix 1 available at http://links.lww.com/BRS/A840). According to the protocol put forth by Beauséjour et al,24 a subgroup of patients completed the Short-Form of the survey (SF-12) quality-of-life questionnaire.32,33 To test temporal reliability, another subgroup of patients received a second identical questionnaire (SRS-22 fcv) with a stamped return
Figure 1. Schema of validity and reliability explanations. Spine
SRS-22 fcv in France • Lonjon et al
envelope for return within 2 weeks. To limit the number of questionnaires for each patient and increase concentration, an unequal random selection was made to designate which patients would complete the SF-12 and which patients would complete a second SRS-22 fcv. Patients received an explanation of the origin of the questionnaire and the study purpose but no help in answering the questionnaire. Patients were distributed into 2 groups: (1) adolescents with “nonclinically significant scoliosis” (NCSS) (curve angle ≤10° according to the Cobb angle on anteroposterior radiographs), and (2) adolescents with AIS according to the SRS criteria34 (lateral deviation of the spine more than 10° according to the Cobb angle on anteroposterior radiographs). A third group of similar-aged healthy adolescents, recruited from a public high school in Paris and without back problems, completed the SRS-22 fcv in the classroom without assistance.
Analysis First, each item of the SRS-22 fcv was scored from 1 (worst) to 5 (excellent). For each domain, for the subtotal score (the subtotal score is calculated with items 1 to 20, without satisfaction management domains) and for the total score, the average value and floor and ceiling effects (percentage of participants showing the minimum and maximum possible scores, respectively) were calculated. Second, the reliability of the SRS-22 fcv was assessed (internal consistency and temporal reliability). Internal consistency was assessed by calculating the Cronbach α for each domain and for the total score for the 3 groups. The internal consistency was considered good with Cronbach α value more than 0.6.35 Then in the test-retest method, temporal reliability was measured by calculating the intraclass correlation coefficient (ICC) with an estimate of measurement error (Standard Error of measurement: SEm). An ICC value less than 0.2 indicates poor agreement, 0.2 to 0.3 fair agreement, 0.4 to 0.5 moderate agreement, 0.6 to 0.8 strong agreement, and more than 0.8 almost perfect agreement. Third for validity, construct validity was assessed. Convergent and discriminant validity are the 2 subtypes of validity that make up construct validity. Convergent validity refers to the degree to which 2 measures of constructs that theoretically should be related, are in fact related. Discriminative validity, refer to the ability of a tool to distinguish among different categories of subjects.36 For convergent validity, Pearson correlation coefficients were calculated for the corresponding domain scores of the SRS-22 fcv and SF-12. According to previous studies, the domains “function” and “pain” of the SRS-22 correspond to the domains “physical function,” “role-physical,” and “pain” of the SF-12, the domain “self-image” of the SRS22 corresponds to the domain “general health perceptions” of the SF-12 and the domain “mental health” of SRS-22 to domains “role emotional” and “mental health” of SF-12.15,20,24 Correlation values of 0.20 or more were considered at least fair. Then, a factor analysis was performed. Factor analysis investigating the correlation between items involved principal component analysis and varimax rotation factor to determine whether the SRS-22 fcv revealed a structure similar to the www.spinejournal.com
E27
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE130351.indd E27
06/12/13 1:40 PM
HEALTH SERVICES RESEARCH
SRS-22 fcv in France • Lonjon et al
original version of the instrument (items selection for loading >0.45). Here, we used data for the AIS group for 4 factors only and the items representing each factor and the explained variance. For discriminative validity, a univariate analysis (student test, analysis of variance) of the relationship between clinical variables and questionnaire scores was conducted under the assumption that unfavorable characteristics are associated with poor scores. The most relevant general characteristics for all participants were used for this univariate analysis: sex, age, age at diagnosis, body mass index, and radiographical indices (Cobb angle of the main spine curve, type of scoliosis, Risser
score for skeletal maturity [0–3, immature; 4 and 5, mature] and type of treatment [patients under observation, patients treated by a brace, and patients scheduled for surgery]). A multiple regression analysis of independent variables (P < 0.2) (stepwise with backward direction) was then performed.
Statistical Analysis Statistical analysis involved use of R v13.0 (http://www.Rproject.org, the R Foundation for Statistical Computing, Vienna, Austria). P < 0.05 was considered statistically significant.
TABLE 1. Characteristics of Adolescents With NCSS and AIS and Healthy Adolescents Who
Completed the SRS-22 fcv Questionnaire NCSS n = 25
AIS n = 175
Healthy Adolescents n = 60
Male
5 (20%)
32 (18.3%)
13 (21.7%)
Female
20 (80%)
143 (81.7%)
47 (78.3%)
13.8 (12.8–14.6) (12–18.6)
14.4 (13.7–15.7) (10.1–18.9)
14.7 (13.7–16.2) (12.7–18.3)
11 yr Body mass index (kg/m ) median (Q1, Q3) (range) 2
17.8 (16.8–19.2) (13.2–28.7) 19.3 (17.8–21.3) (13.0–30.9)
19.2 (17.6–20.4) (13.3–27.1)
Type of scoliosis Simple thoracic
71 (40.6%)
Thoracolumbar
46 (26.3%)
Lumbar
38 (21.7%)
Double main
20 (11.4%)
Cobb angle (°) median (Q1, Q3) (range)
8 (7–9) (6–10)
28 (19–42) (11–75)
0
9 (36%)
32 (18.3%)
1
5 (20%)
16 (9.1%)
2
1 (4%)
13 (7.4%)
3
2 (8%)
32 (18.3%)
4
7 (28%)
60 (34.3%)
5
1 (4%)
22 (12.6%)
25 (100%)
42 (24%)
Cast
0
102 (58.3%)
Surgery scheduled
0
31 (17.7%)
Risser score
Treatment Observation
NCSS indicates nonclinically significant scoliosis; AIS, adolescent idiopathic scoliosis; SRS-22 fcv, French-Canadian version of the Scoliosis Research Society 22.
E28
www.spinejournal.com
January 2014
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE130351.indd E28
06/12/13 1:40 PM
HEALTH SERVICES RESEARCH
SRS-22 fcv in France • Lonjon et al
TABLE 2. Descriptive Statistics for Domains of the SRS-22 fcv NCSS n = 25
AIS n = 175
Healthy Adolescents n = 60
4.46 (0.43)
4.19 (0.63)
4.67 (0.46)
% floor
0
0
0
% ceiling
24
19.4
41.7
3.79 (0.64)
3.54 (0.64)
4.18 (0.63)
% floor
0
0
0
% ceiling
0
1.1
15
4.62 (0.62)
4.43 (0.63)
4.67 (0.41)
% floor
0
0
0
% ceiling
0
0
0
3.98 (0.69)
3.77 (0.77)
3.97 (0.68)
% floor
0
0
0
% ceiling
8
8
6.7
4.07 (0.63)
3.68 (0.82)
…
% floor
0
1.2
…
% ceiling
13
9.4
…
4.21 (0.43)
3.98 (0.5)
4.37 (0.44)
% floor
0
0
0
% ceiling
0
0.6
1.7
4.2 (0.41)
3.96 (0.47°
…
% floor
0
0
…
% ceiling
0
0.6
…
Domains (Items) Pain (items 1, 2, 8, 11, 17) Mean score (SD)
Self-image (items 3, 7, 13, 16, 20) Mean score (SD)
Function (items 4, 6, 10, 14, 19) Mean score (SD)
Mental health (items 5, 9, 12, 15, 18) Mean score (SD)
Satisfaction with management (items 21, 22) Mean score (SD)
Subtotal (items 1–20) Mean score (SD)
Total mean (SD) (items 1–22) Mean score (SD)
The subtotal score is calculated with items 1 to 20. NCSS indicates nonclinically significant scoliosis; AIS, adolescent idiopathic scoliosis; SRS-22 fcv, French-Canadian version of the Scoliosis Research Society 22.
RESULTS Patients Demographic characteristics of participants are summarized in Table 1. Among 218 eligible patients seen for scoliosis, 200 patients agreed to complete the questionnaire: 175 with AIS and 25 with NCSS. Among the 150 patients who received the SF-12, 136 completed it (119 AIS and 17 NCSS) and of the 50 patients who received a second SRS-22 fcv, 38 sent it back (33 AIS and 5 NCSS). Thanks to random, subgroups of patients selected to receive the SF-12 or the second SRS-22 Spine
fcv were representative (P > 0.05) of general population in terms of sex ratio, age, body mass index, Risser score, Cobb angle, type of curve, type of scoliosis and results (total score). In addition, 60 healthy adolescents completed the SRS-22 fcv. Girls represented 210 of the 260 participants. The mean age was 14.4 years (range, 10.1–18.9 yr), and the mean body mass index was 19.1 kg/m2 (range, 13–30.8 kg/m2). The mean magnitude of the main spinal curve (Cobb angle) for patients with AIS was 28° (range, 6°–75°). Of the 200 patients with AIS or NCSS, 31 were scheduled for surgical correction, 102 had received a brace, and 67 were under observation. www.spinejournal.com
E29
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE130351.indd E29
06/12/13 1:40 PM
HEALTH SERVICES RESEARCH
SRS-22 fcv in France • Lonjon et al
TABLE 3. Convergent Validity Between Relevant Domains of the SRS-22 fcv and the SF-12 for Patient
With AIS or NCSS (SF-12)
SRS-22 Function
Pain
Self-image
Mental Health
Satisfaction With Management
Physical function
0.27
0.29
0.30
0.10
0.08
Role-physical
0.36
0.39
0.21
0.30
0.44
Pain index
0.32
0.41
0.17
0.29
0.08
n = 136 SF-12
General health perceptions
0.21
0.16
0.38
0.29
0.23
Vitality
0.12
0.23
0.22
0.30
0.19
Social functioning
0.34
0.24
0.36
0.46
0.17
Role emotional
0.35
0.32
0.39
0.48
0.22
Mental health
0.20
0.26
0.29
0.62
0.37
Pearson correlation coefficients in boldface are for conceptually corresponding domains. Correlations italicized are not statistically significant (P > 0.05). NCSS indicates nonclinically significant scoliosis; AIS, adolescent idiopathic scoliosis; SRS-22 fcv, French-Canadian version of the Scoliosis Research Society 22; SF-12, Short-Form 12.
Analysis of the domain scores for the SRS-22 fcv (Table 2) revealed a ceiling effect for pain, not expressing the clinical diversity of patients with little or no pain.
Reliability Internal consistency of the SRS-22 fcv was good, with Cronbach α 0.82 for patients with AIS, 0.79 for patients with NCSS, and 0.93 for healthy adolescents. For patients with AIS, good reliability was found for all domains: pain, 0.71; self-image, 0.61; function, 0.60; mental health, 0.73; and satisfaction with management, 0.60. In all, 34 of 38 patients completed the second SRS-22 fcv. The ICC for the 4 domains combined and separate demonstrated excellent test-retest reproducibility (total score ICC = 0.90, SEm = 0.31; pain domain ICC = 0.93, SEm = 0.50; self-image domain ICC = 0.89, SEm = 0.53; function ICC = 0.86, SEm = 0.47; mental health domain ICC = 0.85, SEm = 0.53).
Validity For convergent validity, the correlations between the SRS-22 fcv domains and the relevant SF-12 domains (in bold in
Table 3) was on the range from 0.27 to 0.62 (P < 0.001), considered at least fair. Consequently, all of the convergent validity hypothesizes were supported by the correlation results. For construct validity, questions related to satisfaction with treatment (items 21 and 22) were not included in the factor analysis because this dimension was not considered an aspect of the disease. Factor analysis of the 20 items produced 4 factors that explained 44% of the variance (Table 4). These factors corresponded to the pre-established dimensions of the original questionnaire, with some exceptions. Items 7 (factor loading = 0.63), 12 (factor loading = 0.80), 14 (factor loading = 0.78), 17 (factor loading = 0.71), 18 (factor loading = 0.51), and 20 (factor loading = 0.53) originally located in a domain, were related to another factor on factor analysis. Item 11 was not located in a domain. For construct validity, univariate analysis found several variables statistically associated with a poor score (see Supplemental Digital Content Appendix 2 available at http://links. lww.com/BRS/A840 ). The total score was greater for healthy adolescents than for patients with AIS (P = 0.018) (Figure 2) and for NCSS than for patients with AIS (P < 0.001). Older
TABLE 4. Factorial Structure of the SRS-22 fcv With 4 domains for Patients With AIS Factors
Variance, %
Factor Items*
SRS Domains
SRS Items
1
19
4-6-10-18-20
Self-image
4-6-10-14-19
2
14
3-12-13-14-16-17
Mental health
3-7-13-16-20
3
11
1-2-8
Pain
1-2-8-11-17
4
10
5-7-9-15
Function
5-9-12-15-18
*Items with factor loading > 0.45. Items in boldface were not found in the relevant factors; AIS, adolescent idiopathic scoliosis. SRS-22 fcv indicates French-Canadian version of Scoliosis Research Society 22.
E30
www.spinejournal.com
January 2014
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE130351.indd E30
06/12/13 1:40 PM
HEALTH SERVICES RESEARCH
Figure 2. Box plot figure of SRS-22 total score of healthy patient and patient with AIS. AIS indicates adolescent idiopathic scoliosis; SRS-22, Scoliosis Research Society 22.
patients, patients with obesity, patients with a large Cobb angle, patients with a double main curve, and patients scheduled for surgery had minor scores. Figures 3 and 4 show in a box plot format the mean scores for type of curve and type of treatment. Patients with a double main curve and patients scheduled for surgery had minor score. After multivariate analysis, only 2 clinical variables were independently associated with SRS-22 fcv total score (Cobb angle curve: P = 0.03; patients scheduled for surgery: P = 0.04) (see Supplemental Digital Content Appendix 3 available at http://links.lww .com/BRS/A840).
DISCUSSION To the authors’ knowledge, use of a previously adapted and validated French-Canadian version of a health-quality questionnaire in France has never been investigated. The aim of this study was to determine whether the SRS-22 fcv could be used in France and specifically in an adolescent population. In France, the SRS-22 fcv showed good internal reliability (Cronbach α = 0.81), good temporal reliability (ICC
Figure 3. Box plot figure of SRS-22 total score for different types of curve. SRS-22 indicates Scoliosis Research Society 22. Spine
SRS-22 fcv in France • Lonjon et al
Figure 4. Box plot figure of SRS-22 total score for different types of treatment. SRS-22 indicates Scoliosis Research Society 22.
0.90), and significant convergent validity between SRS-22 fcv domains and corresponding SF-12 domains (P < 0.001). For construct validity, factor analysis revealed 4 coherent factors and multivariate regression analysis showed good correlation between Cobb angle of the main spinal curve and total SRS-22 fcv score, and between type of treatment and total SRS-22 score. However the correlation between Cobb angle and SRS-22 score was not very strong (β = −0.06; P = 0.03). This might be explained by a nonlinear relationship.37 For divergent validity, as expected unfavorable characteristics (double main curve and patient scheduled for surgery) are associated with poor score. Results were consistent with prior observations,17–22 and internal reliability of this study was consistent with similar evaluations of the SRS-22 in other languages in the literature (Table 5). Thus, the SRS-22 fcv can be used in France. Nevertheless, as for other adapted versions,17–19,21,22 validity and reliability of the SRS-22 fcv were slightly inferior to those of the original version. Regarding internal reliability, although the coefficient values were lower than those of the original SRS-22 (0.21 less for pain, 0.14 less for self-image, 0.26 less for function, 0.17 less for mental health, and 0.28 less for satisfaction with management domains), they still equaled or exceeded the acceptability threshold of 0.6. Several reasons might explain these differences. First, item 15 in the function domain (Cronbach α = 0.60), representing financial difficulties, is not relevant in France because the French health care system provides 100% of the fees for AIS. This question, with previously observed difficulties,17,18,21 should not be presented to patients in France so as not to overestimate the mean score. Second, item 19, attractiveness in the self-image domain (Cronbach α = 0.61), was not adapted for patients younger than 13 years.24 In this study, 20% of the participants were in this age group. Patients younger than 13 years should not complete this question to avoid soliciting random answers. Regarding convergent validity, all correlations among related domains of the SF-12 were statistically significant www.spinejournal.com
E31
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE130351.indd E31
06/12/13 1:40 PM
HEALTH SERVICES RESEARCH
SRS-22 fcv in France • Lonjon et al
TABLE 5. Reliability (Cronbach α Coefficients) for the SRS-22 fcv in France (This Study) and Previous
Adaptations
Current Study (10––18 yr)
Quebec (10––21 yr) (Beauséjour et al)24
Original (
