Sage Journals: Discover world-class research

Abstract

French

Objective:

Attention-deficit hyperactivity disorder is one of the most frequent neurodevelopmental disorders. Its diagnosis requires reference questionnaires such as the Conners’ Parent Rating Scale (CPRS). Presently, in French-speaking countries, a few translations of the revised short CPRS have been put to use without previous formal validation. We sought here for the validation of a French version (Lausanne, Switzerland) of the revised short CPRS regarding construct validity, internal consistency, and item reliability in a sample of French schoolchildren.

Method:

The study involved 795 children and adolescents aged 9 to 19 years from a single school. The factorial structure and item reliability were assessed with a confirmatory factor analysis for ordered categorical variables. The dimension internal consistency was assessed with Guttman’s lambda 6 coefficient.

Results:

The results confirmed the original and strong 3-dimensional factorial structure (Oppositional, Cognitive Problems/Inattention, and Hyperactivity), showed satisfactory item reliability, and indicated a good dimension internal consistency (Guttman’s lambda 6 coefficient: 0.87, 0.90, and 0.82, respectively, to the 3 dimensions).

Conclusions:

Thus, the Lausanne French version of the revised short CPRS may be considered validated regarding construct validity and item and dimension reliability; it can be now more confidently used in clinical practice.

Keywords

attention-deficit hyperactivity disorder child adolescent validation studies confirmatory factor analysis validity reliability questionnaire

Attention-deficit hyperactivity disorder (ADHD) is one of the most frequent neurodevelopmental disorders. It affects 1% to 20% of school-aged children, and its global prevalence is 5.3% (95% CI, 5.0 to 5.6).¹

The current knowledge is orienting the understanding of this disorder toward a multifactorial cause: genetic, biological, environmental, and psychosocial.² The main clinical presentations are difficulties with concentration, organization, and planning; impulsivity; and, very often, motor hyperactivity.^3
–5

For ADHD diagnosis, the international guidelines recommend first a clinical examination, then the use of reference questionnaires^4,5 (Conners’ Parent Rating Scale [CPRS]⁶ or ADHD Rating Scale⁷). These questionnaires have been scientifically validated and investigate the overall functioning of a given subject in various environments (school, family, etc.).

The long form of the revised CPRS (CPRS-R: L) includes 48 items divided into 7 dimensions. The short form of the CPRS-R (the CPRS-R: S) includes 27 items, of which 18 quantify 3 dimensions: Oppositional, Cognitive Problems/Inattention, and Hyperactivity. The 9 remaining items and 3 items from the Cognitive Problems/Inattention dimension provide an ADHD index.

A number of studies have demonstrated the relevance of the various CPRSs and underlined the advantages of the CPRS-R in investigating ADHD.⁸ Some Conners’ scales have been already translated and validated in various languages (Turkish,⁹ Swedish,¹⁰ Spanish,¹¹ or rural Bengali¹²). A few French versions of the CPRS-R: S are also available; however, to our knowledge, although widely used, none has been validated yet. In 2006, the American CPRS-R: S⁶ was translated into French and then back-translated in Lausanne, Switzerland¹³ (Table 1) in collaboration with Multi-Health Systems (New York, NY, and Toronto, Canada).

Table 1.

French Version of the Conners’ Parent Rating Scale,^a Short Form (CPRS-R: S) Used for the Study (Lausanne Translation).

Questions concernant l’enfant
Vous trouverez ci-dessous une liste des comportements que les enfants peuvent présenter.	Basez-vous sur les attitudes de votre enfant durant le mois écoulé.	Pour répondre à chaque question, entourez d’un cercle la meilleure réponse.	Pour des raisons de lisibilité, nous utilisons le masculin généralement.

		Un peu vrai (occasion nel)	Assez vrai (souvent, passable ment)	Très vrai (Très souvent, très fréquem ment)
1.	Inattentif, facilement distrait	1	2	3
2.	Colérique et rancunier	1	2	3
3.	Difficulté à faire ou à terminer ses devoirs à la maison	1	2	3
4.	Bouge sans arrêt, ne tient pas en place	1	2	3
5.	A une durée d’attention réduite	1	2	3
6.	Se dispute avec les adultes	1	2	3
7.	Bouge sans arrêt ses mains ou ses pieds, se tortille sur son siège	1	2	3
8.	N’arrive pas à finir ses devoirs	1	2	3
9.	Difficile à contrôler dans un magasin	1	2	3
10.	Désordonné à la maison ou à l’école	1	2	3
11.	Perd facilement son calme	1	2	3
12.	A besoin d’être très encadré pour faire ses devoirs	1	2	3
13.	S’implique dans quelque chose seulement si ça l’intéresse beaucoup	1	2	3
14.	Court ou grimpe partout quant il ne devrait pas	1	2	3
15.	Facilement distrait ou n’arrive pas à se concentrer longtemps	1	2	3
16.	Irritable	1	2	3
17.	Evite, est réticent, ou a des difficultés à s’engager dans les activités demandant une attention soutenue (p. ex. devoirs scolaires)	1	2	3
18.	Se tortille tout le temps	1	2	3
19.	Vite distrait quand on lui explique quelque chose à faire	1	2	3
20.	Défie les adultes ou refuse de leur obéir	1	2	3
21.	A de la peine à se concentrer en classe	1	2	3
22.	A de la peine à rester dans une file ou à attendre son tour dans les jeux ou en groupe	1	2	3
23.	Quitte sa place dans la classe ou dans d’autres situations où il devrait rester assis	1	2	3
24.	Fait exprès des choses qui agacent les autres	1	2	3
25.	Ne suit pas jusqu’au bout les instructions et n’arrive pas à terminer ses devoirs ou ce qui lui est demandé (non provoqué par un problème de compréhension)	1	2	3
26.	A de la peine à rester tranquille pendant les jeux et les loisirs	1	2	3
27.	Facilement frustré quand il doit faire un effort	1	2	3

^aBased on Conners CK 1997. Translation: Michel Bader, Blaise Pierrehumbert, Raphaële Miljkovitch et Olivier Halfon.

The present article reports the validation of this French version regarding its construct validity, internal consistency of the dimensions, and reliability of the items.

Methods

Study Population

By the end of 2011, the study team solicited the parents of 1218 schoolchildren (40 mid-level classes of a private school in Lyon, France) via the school principal and then the head teachers. The approach included an explanatory note and a whole questionnaire (see the 27-item form in Table 1) to fill out freely and anonymously.

Among the 901 returned questionnaires (74% of those sent), 106 were left blank; thus, the response rate was 65% (795 of 1218). The responses corresponded to 366 boys and 374 girls; the sex was not mentioned in 55 forms. The mean ± SD age of the children was 12.7 ± 1.3 years (range: 9-19). Data on all 795 children/adolescents were kept for statistical analysis. The parents of the children under study belonged to the rather well-educated French middle class. In other words, 79% of the children had at least one parent with higher education (nearly 70% of the mothers and 60% of the fathers), and 87% of the mothers and 97% of the fathers were in paid full-time or part-time employment.

Data

The data were the responses to the 18 items that quantify the 3 dimensions of the American version of the CPRS-R: S.^6,14,15 These dimensions were 1) Oppositional (items 2, 6, 11, 16, 20, and 24), 2) Cognitive Problems/Inattention (items 3, 8, 12, 17, 21, and 25), and 3) Hyperactivity (items 4, 9, 14, 18, 22, and 26). Each item is rated on a 4-point Likert scale (0 = not true at all/Pas du tout vrai to 3 = very much true/Très vrai; see Table 1).

Statistical Methods

Confirmatory factor analysis

A confirmatory factor analysis (CFA)¹⁶ checked the 4 hypotheses of the baseline CFA model (or model 1) applied to the French CPRS-R: S: 1) ADHD responses can be explained by 3 factors (also called latent variables, or dimensions): Oppositional, Cognitive Problems/Inattention, and Hyperactivity; 2) each item has a nonnull coefficient of regression on the dimension it was designed to measure (loading factor) and null loading factors on the 2 other dimensions; 3) the 3 factors are correlated; and 4) the residual errors associated with each item are uncorrelated (Figure 1).

As the items are ordinal variables, the parameters of the CFA models were estimated from polychoric correlation matrices,¹⁷ using mean- and variance-adjusted weighted least squares (WLSMV).

Figure 1.

Baseline model (or model 1) for the French version of Conners’ Parent Rating Scale, Short Form (CPRS-R: S). In this diagram, each circle corresponds to a single dimension and each rectangle to a single item. The circles are linked to the rectangles by single-headed arrows only for items that reflect the dimension (i.e., items 2 and 6 for Oppositional and not for the other dimensions). The lack of arrows means that the loading factor is set at 0. The relationships between dimensions are represented by double-headed arrows between circles. This relationship is either a covariance or a correlation if the latent variables are standardized.

Items and latent variables were standardized leading to standardized parameters, thus standardized loading factors. This enables interpreting and comparing these loading factors as correlations.

Several indexes (with their distinct rules for good fit) were used to assess the fit of the CFA models: 1) the comparative fit index (CFI)^18,19 and the Tucker-Lewis index (TLI),²⁰ with “good fit” if >0.95; 2) the root mean square error of approximation (RMSEA) and its 90% confidence interval (CI) with “close fit” if <0.05 and fair fit in the 0.05 to 0.08 range;²¹ 3) the weighted root mean square residual (WRMR), with “good fit” if ≤1.0;²² and 4) the modification indices (MIs).²³

The MIs were used to select additional parameters to improve the fit. A high MI indicates an important improvement. Alternative models may thus be obtained by adding to the initial model (model 1) one parameter at a time, starting with the parameter with the highest MI. This purely statistical approach was associated with pragmatic fit-improvement criteria: 1) the completely standardized expected parameter change (CS-EPC) associated with each MI is ≥0.3 (an item cannot reflect a given dimension if its standardized loading factor is <0.30^24,25 or <0.40),²⁶ and 2) the modifications and parameters (loading factors or residual correlations between items) are clinically meaningful (the wordings of the items are consistent with knowledge behind the development of the scale).

Item reliability

Item reliability was estimated through R², that is, the part of the item variance that may be explained by its underlying dimension. It is equal to the square of the standardized loading factor of the item.¹⁶

Dimension internal consistency

The internal consistency of each dimension was estimated using Guttman’s lambda 6 reliability coefficient.²⁷ Usually, the items of each dimension are considered heterogeneous, consistent, or redundant at values <0.7, close to 0.8, or >0.9, respectively.

Software Programs

CFA and item reliability were estimated using Mplus, version 7.11.²⁸ All other analyses used R, version 3.0.1 (http://www.r-project.org/). All tests were 2-tailed, and P < 0.05 was considered for statistical significance.

Results

Validation of the Factorial Structure

The statistical indexes used for the fit of model 1 (Table 2) were heterogeneous: the CFI and the TLI indicated a good fit (>0.95; precisely, 0.972 and 0.968, respectively), the RMSEA indicated a fair fit (0.059 [90% CI, 0.053 to 0.064], RMSEA ≤0.05 was rejected, P = 0.004), and the WRMR indicated a poor fit (1.321, so >1). Thus, model 1 could be slightly improved using the MIs.

Table 2.

Model Fit on the Basis of the Statistical Indexes.

Model Number and Characteristics	χ²	df	CFI	TLI	RMSEA		WRMR	Model Improvement
Model Number and Characteristics	χ²	df	CFI	TLI	Value (90% CI)	P Value	WRMR	Parameter to Add	MI	CS-EPC
Model 1: baseline model	495.932	132	0.972	0.968	0.059 (0.053 to 0.064)	0.004	1.321	Residual correlation between items 4 and 18	106.085	1.148
Model 2: model 1 + residual correlation between items 4 and 18	398.790	131	0.980	0.976	0.051 (0.045 to 0.056)	0.409	1.155	Residual correlation between items 6 and 20	58.984	0.618
Model 3: model 2 + residual correlation between items 6 and 20	342.287	130	0.984	0.981	0.045 (0.040 to 0.051)	0.903	1.054	Item 24 reflecting hyperactivity	44.231	0.327
Model 4: model 3 + item 24 reflecting hyperactivity	302.595	129	0.987	0.984	0.041 (0.035 to 0.047)	0.993	0.972	Item 25 reflecting hyperactivity	35.167	0.267

CFI, comparative fit index; CI, confidence interval; CS-EPC, completely standardized expected parameter change; MI, modification index; RMSEA, root mean square error of approximation; TLI, Tucker-Lewis index; WRMR, weighted root mean square residual.

The MIs suggested adding, in a first step, a residual correlation between items 4 and 18 of the Hyperactivity dimension and, in a second step, another residual correlation between items 6 and 20 of the Oppositional dimension. This led successively to models 2 and 3 (Table 2), with improved fit at each step. Model 3 was very well fitted according to the CFI, the TLI, and the RMSEA (0.045 [90% CI, 0.040 to 0.051], RMSEA ≤0.05 not rejected, P = 0.903), and the WRMR was much closer to 1 (1.054). No modification of model 3 was relevant regarding the statistical and pragmatic criteria: the MIs suggested that item 24 would also reflect the Hyperactivity dimension (model 3 in Table 2). This led to a weak improvement of the fit of model 4 (as per the WRMR) and did not satisfy the pragmatic fit-improvement criteria (standardized loading factor of item 24 <0.30).

Model 3 was finally retained for the French version of the CPRS-R: S. The standardized parameter estimates are shown in Figure 2. The fit of this model was very good according to the pragmatic criteria: all 18 standardized loading factors were >0.70, of which 12 were even >0.80. The loading factors, the correlation coefficients between the 3 dimensions, and the residual correlations were all statistically significant. In addition, the correlations between the latent variables were not too high: 0.70 between Hyperactivity and Oppositional, 0.56 between Cognitive Problems/Inattention and Oppositional, and 0.71 between Cognitive Problems/Inattention and Hyperactivity. This indicates that the model with 3 dimensions is relevant.

Figure 2.

Final model for the French version of Conners’ Parent Rating Scale, Short Form (CPRS-R: S). Standardized parameter estimates and their standard errors.

Item Reliability

Table 3 shows the estimated reliabilities of all items in decreasing order within each dimension. These values ranged between 0.487 and 0.841 and were all, statistically speaking, significantly different from 0 (P < 0.0001). The lowest values belonged to items 18, 4, and 6 that had residual correlations.

Table 3.

Reliability of the Items per Dimension.

Dimensions and Items	Estimated Reliability	Standard Error	Residual Variance
Oppositional
Item 11	0.770	0.029	0.230
Item 16	0.755	0.029	0.245
Item 24	0.677	0.040	0.323
Item 20	0.667	0.036	0.333
Item 2	0.600	0.037	0.400
Item 6	0.487	0.043	0.513
Cognitive Problems/Inattention
Item 17	0.841	0.021	0.159
Item 3	0.837	0.021	0.163
Item 8	0.825	0.027	0.175
Item 25	0.778	0.030	0.222
Item 21	0.739	0.034	0.261
Item 12	0.646	0.029	0.354
Hyperactivity
Item 14	0.685	0.057	0.315
Item 26	0.675	0.055	0.325
Item 9	0.617	0.067	0.383
Item 22	0.594	0.058	0.406
Item 18	0.553	0.053	0.447
Item 4	0.552	0.046	0.448

Internal Consistency

The coefficient of internal consistency was 0.87 for Oppositional, 0.90 for Cognitive Problems/Inattention, and 0.82 for Hyperactivity. The 3 dimensions are thus internally consistent.

Discussion

The existence of ADHD and its diagnosis and management are still the objects of lively debates. The tools recommended for the diagnostic approach are already widely used but sometimes inadequately or suboptimally. In testing the factorial validity of the Lausanne French version of the CPRS-R: S and in estimating item and dimension reliability, we wanted to provide French-speaking physicians and researchers a valid, reliable, and recognized tool.

The present results confirmed the 3 assumptions checked by the study: 1) a strong 3-dimensional factorial structure identical to that of the 1997 American version, 2) a satisfactory dimension internal consistency, and 3) a good item reliability. These properties make this rating scale a fairly objective behaviour assessment tool in ADHD.

One strength of the present study is that it takes into account the ordered categorical items into the CFAs. In a previous study, Kumar and Steer²⁹ did not use a robust categorical least squares estimator and thus could not confirm the 3-dimensional factorial structure stated by Conners et al.^6,14,15 Other explanations is that Kumar and Steer observed American psychiatric outpatients, whereas Conners et al. observed American and Canadian schoolchildren. This emphasizes the need for a study of the invariance of the scale between different populations using multigroup CFA for categorical data.

The present results showed that allowing for 2 residual correlations, one between items 6 and 20 and another between items 4 and 18, led to a better-fitted model but violated the fourth hypothesis of the baseline CFA model. This was not surprising because, in French, as in English, items 6 and 20 on one hand and items 4 and 18 on the other hand have very close meanings; however, to our knowledge, no residual correlations were considered in the models used with the American version. These residual correlations may stem from the specific feature of our sample (a single French school); they should be confirmed in other independent samples. Anyway, because each residual correlation belongs to items of the same dimension, this does not question the use of CFA models or the strong 3-dimensional factorial structure of the French version.

The suggested modification to model 3 (item 24 may reflect Oppositional as well as Hyperactivity) did not meet the pragmatic criteria (no evidence in French that item 24 does not suggest the Hyperactivity dimension). However, this suggestion was already present in childhood cancer survivors³⁰; this item should thus be carefully examined in further analyses.

Conclusions

To our knowledge, this is the first validation study on the French version of the CPRS-R: S. The next validation study will investigate the scale invariance, first by sex and age classes, then by clinical status (cases vs. controls). In case of invariance, scores by sex and age will be proposed that will allow reliable comparisons between same sex and age classes in similar populations. Other investigations such as test-retest reliability, interrater reliability, external validity, and, especially, sensitivity to change would be necessary. If these properties prove satisfactory, this French version of the CPRS-R: S will represent more than a complementary diagnostic tool; it will allow following children’s progress and using the version in international clinical trials.

Footnotes

Acknowledgments

Special thanks to Charles de Foucauld School Director for providing help and its students for their participation. Study conducted with financial support from Fonds de Perfectionnement du Centre Hospitalier Universitaire Vaudois, Société Académique Vaudoise, Fondation d’Entreprise Laboratoire Urgo, and Shire Pharmaceuticals Group.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Ethics

Study carried out with agreements from 1) Comité de Protection des Personnes Lyon Sud Est II (France), obtained on January 5, 2012; 2) Comité Consultatif sur le Traitement de l’Information en matière de Recherche dans le domaine de la Santé (France), Agreement Nr. 12.185bis, issued on September 6, 2012; and 3) Commission Nationale de l’Informatique et des Libertés (France), Agreement Nr. DR-2012-524, issued on November 7, 2012.

References

Polanczyk

de Lima

Horta

. The worldwide prevalence of ADHD: a systematic review and metaregression analysis. Am J Psychiatry. 2007;164(6):942–948.

National Collaborating Centre for Mental Health. Attention deficit hyperactivity disorder: the NICE guideline on diagnosis and management of ADHD in children, young people and adults. Leicester: The British Psychological Society & London: The Royal College of Psychiatrists. 2009.

American Psychiatric Association. The diagnostic and statistical manual of psychiatric disorders. 4th ed. Washington (DC): American Psychiatric Association; 1994.

Committee on Quality Improvement, Subcommittee on Attention-Deficit/Hyperactivity Disorder. Clinical practice guideline: diagnosis and evaluation of the child with attention-deficit/hyperactivity disorder. Pediatrics. 2000;105(5):1158–1170.

Subcommittee on Attention-Deficit/Hyperactivity Disorder, Steering Committee on Quality Improvement and Management. ADHD: clinical practice guideline for the diagnosis, evaluation, and treatment of attention-deficit/hyperactivity disorder in children and adolescents. Pediatrics. 2011;128(5):1–16.

Conners

. Conners’ Rating Scales–Revised technical manual. North Tonawanda, NY: Multi-Health Systems; 1997 (ordering information available at http://www.mhs.com).

DuPaul

Power

, Anastopoulos. ADHD Rating Scale IV: checklists, norms, and clinical interpretations. New York (NY): Guilford; 1998.

Gianarris

Golden

Greene

. The Conners’ Parent Rating Scales: a critical review of the literature. Clin Psychol Rev. 2001;21(7):1061–1069.

Dereboy

Şenol

Şener

. Validation of the Turkish versions of the short-form Conners’ Teacher and Parent Rating Scales. Turk Psikiyatri Derg. 2007;18(1):1–11.

10.

Westerlund

Holmberg

. The Conners’ 10-item scale: findings in a total population of Swedish 10-11-year-old children. Acta Paediatr. 2009;98:828–833.

11.

Farré-Riba

Narbonna

. Conners’ rating scales in the assessment of attention deficit disorder with hyperactivity (ADHD): a new validation and factor analysis in Spanish children. Rev Neurologia. 1997;25(138):200–204.

12.

Pal

Chaudhury

Das

. Validation of a Bengali adaptation of the Conners’ Parent Rating Scale (CPRS-48). Br J Med Psychol. 1999;72(4):525–533.

13.

Pierrehumbert

Bader

Thévoz

. Hyperactivity and attention problems in a Swiss sample of school-aged children: effects of school achievement, child gender, and informants. J Atten Disord. 2006;10(1):65–76.

14.

Conners

Wells

Parker

JDA

. A new self-report scale for assessment of adolescent psychopathology: factor structure, reliability, validity, and diagnostic sensitivity. J Abnorm Child Psychol. 1997;25:487–497.

15.

Conners

Sitarenios

, Parker . The revised Conners’ Parent Rating Scale (CPRS-R): factor structure, reliability, and criterion validity. J Abnorm Child Psychol. 1998;26:257–268.

16.

Bollen

. Structural equations with latent variables. New York (NY): John Wiley; 1989.

17.

Muthén

du Toit

SHC

Spisic

. inference using weighted least squares and quadratic estimating equations in latent variable modeling with categorical and continuous outcomes [Internet]. 1997 [accessed February 16, 2016]. Available from: http://www.statmodel.com/bmuthen/articles/Article_075.pdf.

18.

Bentler

. Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives. Structural Equation Modeling. 1999;6(1):1–19.

19.

Bentler

. Comparative fit indices in structural models. Psychol Bull. 1999;107:238–246.

20.

Tucker

Lewis

. A reliability coefficient for maximum likelihood factor analysis. Psychometrika. 1973;38(1):1–10.

21.

Browne

Cudeck

. Alternative ways of assessing model fit. In: Bollen

Long

, editors. Testing structural equation models. Beverly Hills (CA): Sage; 1993. p. 136–162.

22.

. Evaluating cutoff criteria of model fit indices for latent variable models with binary and continuous outcomes [Internet]. Los Angeles: University of California; 2002 [last accessed February 16, 2016]. Available from: http://statmodel2.com/download/Yudissertation.pdf.

23.

Sörbom

. Model modification. Psychometrika. 1989;54:371–384.

24.

Kim

Mueller

. Factor analysis: structural methods and practical issues. Beverly Hills (CA): Sage; 1978.

25.

Brown

. Confirmatory factor analysis for applied research. New York (NY): Guilford; 2006.

26.

Ford

MacCallum

Tait

. The application of factor analysis in applied psychology: a critical review and analysis. Personnel Psychol. 1986;39:291–314.

27.

Guttman

. A basis for analyzing test-retest reliability. Psychometrika. 1945;10:255–282.

28.

Muthén

. Mplus user’s guide. 7th ed. Los Angeles (CA): Muthén & Muthén; 1998-2012.

29.

Kumar

Steer

. Factorial validity of the Conners’ Parent Rating Scale–Revised: Short Form with psychiatric outpatients. J Pers Assess. 2003;80(3):255.

30.

Helton

Corwyn

Bonner

. Factor analysis and validity of the Conners Parent and Teacher Rating Scales in childhood cancer survivors. J Pediatr Psychol. 2006;31(2):200–208.

Validation of the French Version of Conners’ Parent Rating Scale Revised,Short Version: Factorial Structure and Reliability

Abstract

Objective:

Method:

Results:

Conclusions:

Keywords

Methods

Study Population

Data

Statistical Methods

Confirmatory factor analysis

Item reliability

Dimension internal consistency

Software Programs

Results

Validation of the Factorial Structure

Item Reliability

Internal Consistency

Discussion

Conclusions

Footnotes

Acknowledgments

Declaration of Conflicting Interests

Funding

Ethics

References