*Department of Paediatric Dentistry and Orthodontics, Faculty of Dentistry,Federal University of Minas Gerais, Belo Horizonte, Brazil

**Department of Paediatric Dentistry, Faculty of Dentistry,Federal University of Paraná, Curitiba, Brazil

***Division of Public Health and Society, Faculty of Dentistry,McGill University, Montreal, Canada

e-mail: smpaiva@uol.com.br

the influence of oral conditions on the lives ofindividuals [Slade, 1997].

There are as yet few instruments specificallydesigned for children, although recent studies havedemonstrated that children’s reports regarding theirhealth-related quality of life are valid and reliable[Jokovic et al., 2002; Varni et al., 2007a, b]. A searchof the Medline databank revealed that the instrumentsdesigned and validated specifically to measure oralhealth-related quality of life in children are the Child-OIDP (Child-Oral Impacts on Daily Performance),COHIP (Child Oral Health Impact Profile), ECOHIS(Early Childhood Oral Health Impact Scale) andCOHQoL (Child Oral Health Quality of LifeQuestionnaire) [Gerunpong et al., 2004; Jokovic et al.,2004; Jokovic et al., 2006; Pahel et al., 2007; Broderet al., 2007].

The following instruments have been validated foruse on Brazilian children: the Child-OIDP; ECOHIS;and one of the COHQoL – the Child PerceptionsQuestionnaire for the 11-to-14-year-olds (CPQ11-14)[Tesch et al., 2008; Goursand et al., 2008; Castro et al.,

IntroductionHealth-related quality of life is defined as an

emotional assessment one makes regarding one’s ownhealth status [Theunissen et al., 1998]. Dentists havetraditionally been prepared to diagnose orofacialconditions such as caries and malocclusion throughobjective criteria, establishing the clinical need fortreatment. Currently in dentistry, however, there isgrowing interest in assessing the impact of oralconditions on quality of life [Oliveira and Sheiham,2004; Feitosa et al., 2005; Marques et al., 2006; Broderet al., 2007]. As a result, a need has emerged for thedevelopment of assessment instruments for measuring

Preliminary validation of the Brazilianversion of the Child Perceptions

Questionnaire 8-10M.T. MARTINS*, F.M. FERREIRA*-**, A.C. OLIVEIRA*, S.M. PAIVA*-***,

M.P. VALE*, P.J. ALLISON***, I.A. PORDEUS*

ABSTRACT. Aim The aim of the study was to perform a preliminary validation of the Brazilian version of the ChildPerceptions Questionnaire 8-10 (CPQ8-10). Methods The sample used to test the measurement equivalence wasmade up of 59 children from 8 to 10 years of age at the Federal University of Minas Gerais, Brazil. The CPQ8-10

was administered, and clinical exams were performed by a single examiner, calibrated for the diagnosis of dentalcaries (Kappa= 0.98) and malocclusion (weighted Kappa and intraclass correlation coefficient = 0.66-1.00). Thechildren were divided into three groups according to their clinical condition (22 with cavitated carious lesions;20 with malocclusion; 17 with both oral conditions). Statistics The internal consistency of the instrument wasassessed by Cronbach’s alpha Coefficient. The Intraclass correlation coefficient (ICC) was used for the test-retestreliability (40 children). Discriminant validity was determined using the Kruskal-Wallis test. Results Anassessment of the psychometric properties revealed the instrument to be reliable (Cronbach’s alpha=0.92 for thetotal scale and 0.63-0.85 for the subscales), exhibiting excellent stability (ICC=0.96 for the total scale and 0.79-0.95 for the subscales). Construct validity was demonstrated through the significant correlations between theglobal indicators and subscales. There were statistically significant differences between the clinical groupsregarding the total scale, thereby demonstrating discriminant validity (p=0.03). Conclusions The preliminaryvalidation of the Brazilian version of the CPQ8-10 showed valid and reliable for use on Brazilian children.

Key words: Oral health; Quality of life; Validity; Child; Questionnaires.

2008]. The COHQoL is an instrument withmultidimensional scales drafted in English language inCanada and measures the negative effects of orofacialconditions on the social, emotional and functionalwellbeing of children between 6 and 14 years of age aswell as the perceptions of parents and the impact onfamily life. The COHQoL is made up of fiveinstruments: the P-CPQ (Parental–CaregiverPerceptions Questionnaire), which measures parent’sor caregiver’s perceptions regarding the impact ofchildren’s oral health on quality of life; the FIS(Family Impact Scale), which assesses the impact onthe family; and the CPQ, which assesses children’sperceptions regarding the impact of oral health onquality of life. There are three versions of the CPQ,one for each of three specific age groups: 6-7, 8-10 and11-14 years [Jokovic et al., 2004; Jokovic et al., 2006].The CPQ8-10 is not yet validated for use on Brazilianchildren.

The aim of the present study was to perform apreliminary cross-cultural adaptation of the ChildPerceptions Questionnaire 8-10 (CPQ8-10) as well astest its validity and reliability.

MethodsChild Perceptions Questionnaire 8-10 (CPQ8-10). The

CPQ8-10 is made up of 25 items distributed among 4subscales/domains: oral symptoms (5 items),functional limitations (5 items), emotional wellbeing(5 items) and social wellbeing (10 items). The itemsaddress the frequency events occurring in the fourweeks prior to the administration of the questionnaire.A 5-point Likert scale is used, with the followingoptions: ‘Never’=0; ‘Once/twice’=1; ‘Sometimes’=2;‘Often’=3; and ‘Every day/almost every day’=4.CPQ8-10 scores are calculated by summing all of theitem scores. Thus, the total score ranges from 0 (noimpact of oral condition on quality of life) to 100(maximal impact of oral condition on quality of life).There are also two questions on the patientidentification (gender and age) and two globalindicators asking the children for a global rating oftheir oral health and the extent to orofacial conditionaffects their overall wellbeing.

The original CPQ8-10 was developed and validated inCanada, having exhibited good construct validity,substantial internal consistency and acceptable test-retest reliability [Jokovic et al., 2004].

The COHQOL is based on Locker´s model, whichoffers a conceptual framework linking the concepts ofdisease, disability, functional limitation, inability andsocial disadvantage, thereby allowing the evaluation ofthe impact of oral health problems on the life of theindividuals to be performed progressively, frombiological to behavioral to social levels [Locker,1988].

Translation and cross-cultural adaptation of theCPQ8-10. The translation of the questionnaire wasperformed by two independent translators. A bilingualtranslator, whose native language is BrazilianPortuguese, translated the original questionnaire fromEnglish into Brazilian Portuguese. Another bilingualtranslator, whose native language is English,performed the back-translation. The two translatorsworked independently and the one responsible for theback-translation had no knowledge of the originalEnglish-language version of the questionnaire[Guillemin et al., 1993].

The original and back-translated English-languageversions were compared by a translation panel madeup of three individuals with knowledge on the subjectaddressed by the instrument. In choosing thetranslation panel, it was determined that the memberswere to be Brazilian, fluent in English and specialistsin paediatric dentistry due to the importance of thetranslation panel having experience and knowledge ofthe cultural context and development phase of thepopulation to be studied [Van Widenfelt et al., 2005].

For the cross-cultural adaptation, the Brazilianversion was first examined by a team of threespecialists in quality of life and oral health. Thisevaluation obeyed the criteria of clarity of theBrazilian version through the use of simple, easilyunderstood expressions; use of common language,avoiding the use of technical terms; and representationof the content of the original version (conceptualequivalence) [Herdman et al., 1998]. Thequestionnaire was then submitted to a pre-test at aschool through qualitative interviews with 15 boys and15 girls between 8 and 10 years of age (10 from eachage) in order to discuss the relevance of the items withthis group. Items could undergo slight changes or besubstituted in order to obtain item equivalence.Fundamental items were not to be excluded due todifficulties in the translation or complexity of the itemsin the Brazilian culture [Corless et al., 2001].

In order to determine whether the Brazilian versionachieved a similar effect as the original English-language version (semantic equivalence), thequestionnaire was once again evaluated by three otherexperts in quality of life and oral health who werefluent in English as well as by a group of threespecialists in the Brazilian Portuguese language[Herdman et al., 1998].

Validity and reliability of the Brazilian version of theCPQ8-10

A second pre-test was carried out at a school with 30children who did not take part of the final sample (10children at each age) so that the examiner couldobserve the possibility of employing the format of thequestionnaire, instructions and measurement methodsin a similar manner to the original instrument

MARTINS M. ET AL.

(operational equivalence) [Herdman et al., 1998].The option was made to administer the instrument as

an interview, as the authors of the original versionconcluded that eight-year-old children have difficultyin responding to the questionnaire as a self-completedinstrument. The interview also prevents parents frominterfering in their child’s answers [Brown and Al-Khayal, 2006].

The psychometric properties were assessed in thesame method as the original version of the CPQ8-10,including the adoption of an approximate sample sizein order to assess measurement equivalence (59children) (Fig. 1) [Jokovic et al., 2004].

The administration of the Brazilian version of theCPQ8-10 and the clinical oral exams of the 59children were held at the dental clinics of the Facultyof Dentistry of the Federal University of Minas Gerais(Brazil) by a single, previously calibrated examiner(Kappa = 0.98 for dental caries; weighted Kappa andintraclass correlation coefficient = 0.74-1.00 formalocclusion). The 59 children were divided into threegroups mutually exclusive according to their oralcondition: children with cavitated carious lesions (22),children with malocclusions (20) and children withboth conditions simultaneously (17). The childrenwere selected due to the fact that they had their centralincisors in occlusion, thereby allowing an adequateassessment of the occlusion. Another inclusioncriterion was that they were not yet in treatment,which was essential to discerning the impact on

quality of life due to the oral condition rather than dueto the treatment to which the child was submitted.

The procedures and diagnostic criteria for dentalcaries were those recommended by the World HealthOrganization (WHO) (1997). The Dental AestheticIndex (DAI) was used for the diagnosis ofmalocclusion [WHO, 1997].

The SPSS software program (version 12.0., SPSSInc., Chicago, IL, USA) was used for the data analysis.Descriptive analysis was carried out, obtaining meanvalues, standard deviation, total score and subscalescores. Internal consistency of the BrazilianPortuguese-language version of the CPQ8-10 was testedusing Cronbach’s alpha coefficient for the subscalesand total score. Test-retest reliability was determinedusing the intraclass correlation coefficient for thesubscales and total score; for such, 40 children of the59 participants of the final sample were interviewed asecond time within a period of seven to 14 days; thesechildren were chosen for not having undergone dentaltreatment since the initial interview and maintainingthe same clinical condition in both interviews.Construct validity was tested by calculatingcorrelations between the global indicators and totalscore as well as between the global indicators and eachsubscale, using Spearman’s correlation coefficient.Discriminant validity was tested through a comparisonof the clinical groups regarding the total CPQ scoreand score on each subscale, using the Kruskal-Wallistest at a significance level ≤5.

The study was approved by the Human ResearchEthics Committee of the Federal University of MinasGerais. Parents/guardians and children read andsigned an informed consent form prior to participationin the study.

ResultsThe results of the comparison between the original

CPQ8-10 and the back-translated version assessed by thetranslation panel demonstrate that the two versions areequivalent. The experts in quality of life found thesubscales of the questionnaire to be important toBrazilian culture. The item and operationalequivalence steps demonstrated that the instrument isunderstood by Brazilian children between 8 and 10years of age and that the questionnaire may beadministered to these children in the same format asthe original. Semantic equivalence was achieved aftera few changes in the grammatical structure of thequestionnaire.

For the assessment of measurement equivalence,59 children were examined and interviewed, 59.3% ofwhom were boys. Age distribution was homogeneous:39.0% were eight years of age; 32.2% were nine yearsof age; and 28.8% were 10 years of age. Regardingoral conditions, 37.3% had cavitated carious lesions,

VALIDITY OF THE BRAZILIAN CPQ8-10

33.9% exhibited malocclusions and 28.8% had bothconditions simultaneously.

ReliabilityCronbach’s alpha coefficient was 0.92 for the total

scale and ranged from 0.63 (oral symptoms) to 0.85(emotional and social wellbeing) for the subscales.Test-retest reliability was assessed using the intraclasscorrelation coefficient (ICC). The ICC was 0.96 for thetotal scale, 0.79 for oral symptoms, 0.88 for functionallimitations, 0.95 for emotional wellbeing and 0.92 forsocial wellbeing (Table 1).

Construct validitySpearman’s correlation was significant for the global

indicators and total scale (r = 0.27 and 0.26), functionallimitations (r = 0.23 and 0.26) and emotional wellbeing(r = 0.28 and 0.22). The oral symptoms subscale wasonly significantly associated with the global indicatorrelated to the wellbeing of the child (r = 0.30). Thesocial wellbeing subscale was not associated to eitherof the global indicators (Table 2).

Discriminant validityThere was a statistically significant different

between the clinical groups (children with cavitated

carious lesion; children with malocclusion; childrenwith both conditions) regarding the total score (p =0.03), oral symptoms (p = 0.01) and emotionalwellbeing (p = 0.02) (Table 3).

DiscussionThe psychometric properties of the questionnaire

were evaluated in a similar way to that of the originalinstrument so as to facilitate the comparison of theresults. As the objective of the present study was toassess the reliability and validity of the Brazilianversion of the CPQ8-10 through the establishment ofmeasurement equivalence, the comparison of thevalues of the psychometric properties with those of theoriginal questionnaire is more important than theobtained values [Reichenheim and Moraes, 2007]. Todate, the CPQ8-10 has not been completely validated foruse in other cultures or countries.

Internal consistency of a questionnaire showswhether all the items that make up the instrument arerelated to one another. The internal consistency of theBrazilian version of the CPQ8-10 proved satisfactory forthe social and emotional wellbeing subscales andacceptable for the oral symptoms and functionallimitation subscales, given that Cronbach’s alpha

MARTINS M. ET AL.

TABLE 1 - Reliability statistics for total scale and subscales (n=59).

Variable No. items Cronbach’s αα Intraclass correlation 95% confidence coefficient interval

Total scale 25 0.92 0.96 0.90-0.98

Subscales

Oral symptoms 5 0.63 0.79 0.60-0.89

Functional limitations 5 0.67 0.88 0.78-0.94

Emotional wellbeing 5 0.85 0.95 0.91-0.98

Social wellbeing 10 0.85 0.92 0.84-0.96

TABLE 2 - Construct validity: rank correlations between total scale and subscale scores; global rating of oral health and overallwellbeing (n=59).

Global Rating

Oral Health Overall Wellbeing

r* p-value r* p-value

Total scale 0.27 0.02 0.26 0.03

Subscales

Oral symptoms 0.21 0.06 0.30 0.01

Functional limitations 0.23 0.04 0.26 0.02

Emotional wellbeing 0.28 0.01 0.22 0.05

Social wellbeing 0.16 0.17 0.13 0.24

*Spearman’s Correlation Coefficient

coefficient should be at least 0.7 to 0.8 in order to beconsidered satisfactory when comparing groups[Bland and Altman, 1997]. Compared to the originalquestionnaire, the Brazilian version achieved highervalues for the total scale and the emotional and socialwellbeing subscales. On the oral symptoms andfunctional limitation subscales, the results were similarto those found for the original questionnaire.Cronbach’s alpha coefficient for the Brazilian versionof the CPQ11-14 (0.86) [Goursand et al., 2008] wassimilar to that found in the present study (0.92).

Test-retest reliability was assessed using the ICC,which is considered excellent when above 0.74; goodwhen between 0.6 and 0.74; and fair when between 0.4and 0.59 [Wilson-Genderson et al., 2007]. The resultsshowed excellent stability of the questionnaire for boththe total scale and subscales. Unlike the originalinstrument, which did not achieve a good test-retestresult for the social wellbeing subscale, the Brazilianversion obtained an excellent result for this subscale(0.92) as well as all the other subscales of thequestionnaire. When validated for use in the UnitedKingdom, the CPQ11-14 achieved an ICC of 0.83, whichthe authors considered nearly perfect agreement[Marshman et al., 2005].

Spearman’s correlation demonstrated the constructvalidity for the Brazilian version of the CPQ8-10,excepted for the social wellbeing subscale. The oralsymptoms subscale was only significantly associatedwith the global indicator related to the wellbeing of thechild. Both the original and the Brazilian versionsrevealed non-significant correlations when relating thesocial wellbeing and functional limitations subscalesto the global rating of oral health. A study carried outin Ireland with 270 eight-year-old children comparedthe scores of the CPQ8-10 with those from an instrumentof high validity and reliability that assesses children’sattitudes within their social context (Coopersmith Self-

Esteem Inventory-School Form) [Humphris et al.,2005]. However, the Irish study assessed only theconstruct validity of the CPQ8-10 and did not performthe other psychometric properties.

The original CPQ8-10 did not demonstratediscriminant validity between the groups studied. Theauthors state that this was likely due to the fact that thechildren had previously received clinical andpsychological treatment [Jokovic et al., 2004]. In thepresent validation study, none of the children had yetreceived treatment, which was an important criterionto demonstrate the statistically significant differencesbetween the groups regarding the total scale, oralsymptoms and emotional wellbeing subscales,especially when comparing the scores of the childrenwith one oral condition to those with both conditions.The CPQ11-14 has been validated in differentlanguages and has demonstrated discriminant validitybetween groups with different severities of oralconditions such as malocclusion and dental caries[Foster Page et al., 2005; Brown and Al-Khayal, 2006;O’Brien et al., 2006].

The Brazilian version of the CPQ8-10 exhibitedsimilar psychometric properties to those of the originalinstrument, thereby demonstrating the different typesof equivalence (conceptual, item, semantic andoperational) and achieving the final objective offunctional equivalence [Herdman et al., 1998].

ConclusionsThe preliminary validation of the Brazilian version

of the CPQ8-10 showed valid and reliable foradministration to Brazilian children from 8 to 10 yearsold. The psychometric properties were similar to theproperties of the original English language version andfunctional equivalence was proven. Thus, the validatedquestionnaire may assist healthcare professionals,

VALIDITY OF THE BRAZILIAN CPQ8-10

TABLE 3 - Discriminant validity: total and subscale scores for children with caries, children with malocclusion and childrenwith both conditions (n=59).

Cavitated caries (22) Malocclusion (20) Cavitated caries p-value*+ malocclusion (17)

Mean Median (SD) Mean Median (SD) Mean Median (SD)

Total scale 14.77 9.00 (±3.28) 10.70 6.50 (±12.30) 21.29 21.00 (±13.18) 0.02

Subscales

Oral symptoms 4.86 4.50 (±3.21) 3.20 2.50 (±2.72) 7.29 7.00 (±3.07) <0.001

Functional limitations 3.05 2.00 (±2.93) 2.30 1.50 (±2.95) 2.88 2.00 (±2.61) 0.40

Emotional wellbeing 3.00 1.00 (±3.79) 1.75 0.50 (±3.02) 4.65 3.00 (±4.15) 0.02

Social wellbeing 3.95 2.00 (±5.73) 3.40 1.00 (±5.09) 6.53 6.00 (±5.76) 0.10

* Kruskall-Wallis Test

especially paediatric dentists, in better understandingthe influence of oral health on the quality of life ofchildren. Based on such knowledge, effective healthpromotion measures targeting this segment of thepopulation can be developed and applied.

Acknowledgements

This study was supported by National Council forScientific and Technological Development (CNPq),Ministry of Science and Technology, Brazil.

References

Bland JM, Altman DG.Statistics notes:Cronbach´s alpha. BMJ1997; 314:572.

Broder HL, McGrath C, Cisneros GJ. Questionnaire development:face validity and item impact testing of Child Oral HealthImpact Profile. Community Dent Oral Epidemiol 2007;35:8-19.

Brown A, Al-Khayal Z. Validity and reliability of the ArabicTranslation of the Child oral-health-related quality of lifequestionnaire (CPQ11-14) in Saudi Arabia. Int J Paediatr Dent2006;16:405-11.

Castro RA, Cortes MI, Leão AT, et al. Child-OIDP index in Brazil:cross-cultural adaptation and validation. Health Qual LifeOutcomes 2008;6:68.

Corless IB, Nicholas PK, Nokes KM. Issues in cross-culturalquality-of-life research. Image J Nurs Sch 2001;33:15-20.

Feitosa S, Colares V, Pinkham J. The psychosocial effects of severecaries in 4-year–old children in Recife, Pernambuco, Brazil.Cad Saúde Pública 2005;21:1550-6.

Foster Page LA, Thomson WM, Jokovic A, Locker D. Validationof the Child Perceptions Questionnaire (CPQ11-14). J Dent Res2005;84:649-52.

Gerunpong S, Tsakos G, Sheiham A. Developing and evaluating anoral health-related quality of the index for children; the Child-OIDP. Community Dent Health 2004;21:161-9.

Goursand D, Paiva SM, Zarzar PM, et atl. Cross-culturaladaptation of the Child Perceptions Questionnaire 11-14(CPQ11-14) for the Brazilian Portuguese language. Health QualLife Outcomes 2008;6:2.

Guillemin F, Bombardier C, Beaton D. Cross-cultural adaptation ofhealth-related quality of life measures: literature review andproposed guidelines. J Clin Epidemiol 1993;46:1417-32.

Herdman M, Fox-Rushby J, Badia X. A model of equivalence inthe cultural adaptation of HRQoL instruments: the universalistapproach. Qual Life Res 1998;7:323-35.

Humphris G, Freeman R, Gibson B, Simpson K, Whelton H. Oralhealth-related quality of life for 8-10-year-old children: anassessment of a new measure. Community Dent Oral Epidemiol2005;33:326-32.

Jokovic A, Locker D, Stephens M, Kenny D, Tompson B, GuyattG. Validity and reliability of a questionnaire for measuring childoral-health-related quality of life. J Dent Res 2002;81:459-63.

Jokovic A, Locker D, Tompson B, Guyatt G. Questionnaire formeasuring Oral Health-related Quality of Life in eight- to ten-year-old children. Pediatr Dent 2004;26:512-8.

Jokovic A, Locker D, Guyatt,G. Short forms of the ChildPerceptions Questionnaire for 11-14-year-old children (CPQ11-14): development and initial evaluation. Health Qual LifeOutcomes 2006;4:4.

Locker D. Measuring oral health: a conceptual framework.Community Dent Health 1988;5(1):3-18.

Marques LS, Ramos-Jorge ML, Paiva SM, Pordeus IA.Malocclusion: esthetic impact and quality of life amongBrazilian schoolchildren. Am J Orthod Dentofacial Orthop2006;129:424-7.

Marshman Z, Rodd H, Stern M, et al. An evaluation of the ChildPerceptions Questionnaire in the UK. Community Dent Health2005,22:151-5.

O´Brien K, Wright JL, Conboy F, Macfarlane T, Mandall N. TheChild Perceptions Questionnaire is valid for malocclusions inthe United Kingdom. Am J Orthod Dentofacial Orthop 2006;129:536-40.

Oliveira CM, Sheiham A. Orthodontic treatment and its impact onoral health-related quality of life in Brazilian adolescents. JOrthod 2004;31:20-7.

Pahel BT, Rozier RG, Slade GD. Parental perceptions of children’soral health: the Early Childhood Oral Health Impact Scale(ECOHIS). Health Qual Life Outcomes 2007;5:6.

Reichenheim ME, Moraes CL. Operacionalização de adaptaçãotranscultural de instrumentos de aferição usados emepidemiologia. Rev Saúde Pública 2007;41:665-73.

Slade GD. Derivation and validation of a short- form oral healthimpact profile. Community Dent Oral Epidemiol 1997;25:284-90.

Tesch FC, Oliveira BH, Leão A. Equivalência Semântica da versãoem português do instrumento Early Childhood Oral HealthImpact Scale. Cad. Saúde Pública 2008;24:1897-909.

Theunissen NC, Vogels TG, Koopman HM, et al. The proxyproblem: child report versus parent report in health-relatedquality of life research. Qual Life Res 1998;7:387-97.

Van Widenfelt BM, Treffers PD, Beurs E, Siebelink BM, KoudijsE. Translation and cross-cultural adaptation of assessmentinstruments used in psychological research with children andfamilies. Clin Child Fam Psychol Rev 2005;8:135-47.

Varni J, Limbers C, Burwinkle T. How young can children reliablyand validly self-report their health-related quality of life? Ananalysis of 8591 children across age subgroup with the PedsQL4.0 Generic Core Scales. Health Qual Life Outcomes 2007;5:1.

Varni J, Limbers C, Burwinkle T. Parent proxy-report of theirchildren`s health-related quality of life: an analysis of 13878parents’ reliability and validity across age subgroups using thePedsQL 4.0 Generic Core Scales. Health Qual Life Outcomes2007;5:2.

Wilson-Genderson M, Broder HL, Philips C. Concordancebetween caregiver and child reports of children´s oral health-related quality of life. Community Dent Oral Epidemiol2007;35:32-40.

World Health Organization. Oral Health Surveys: basic methods.4th ed. Geneva: ORH/EPID; 1997.

MARTINS M. ET AL.