Validación de la versión brasilera del Youth Risk Behavior Survey 2007
Dartagnan Pinto Guedes; Cynthia Correa Lopes
Centro de Educação Física e Esporte. Universidade Estadual de Londrina. Londrina, PR, Brasil
OBJECTIVE: To validate the psychometric properties of the Brazilian version of the 2007 Youth Risk Behavior Survey (YRBS) questionnaire.
METHODS: The original version of the 2007 YRBS was translated into Portuguese and back-translated into English. The questionnaire versions were analyzed by a committee of experts. The committee used semantic, idiomatic, cultural and conceptual equivalences as criteria of analysis. The final version of the translated 2007 YRBS questionnaire was administered in two occasions, with an interval of two weeks, in a sample of 873 high school students of both sexes, in the city of Londrina, Southern Brazil, to identify the psychometric properties. Test-retest reliability was analyzed by calculating the Kappa index of agreement and prevalence rate of each type of risk behavior when application was repeated. Chi-square test was used to identify statistical differences between the first and second questionnaire applications.
RESULTS: After minor changes identified in the translation process, the committee of experts concluded that the Portuguese version of the 2007 YRBS showed semantic, idiomatic, cultural and conceptual equivalences. Significant differences between the prevalence rates of both applications were found in 23.4% of items. The identification of 91% of items with moderate-to-substantial Kappa index of agreement and mean value of this index of 68.6% indicated the quality of psychometric properties of the Portuguese version of the 2007 YRBS.
CONCLUSIONS: The translation, cross-cultural adaptation and psychometric qualities of the 2007 YRBS questionnaire were satisfactory, thus enabling its application in epidemiological studies in Brazil.
Descriptors: Translations. Questionnaires. Adolescent Behavior. Psychometrics. Reproducibility of Results. Brazil.
OBJETIVO: Validar las propiedades psicométricas de la versión brasileña del cuestionario Youth Risk Behavior Survey (YRBS), versión 2007.
MÉTODOS: Se realizó traducción al portugués de la versión original del YRBS-2007 y retrotraducción al inglés. Las versiones del cuestionario fueron analizadas por comité de árbitros. El comité utilizó como criterio de análisis las equivalencias semántica, idiomática, cultural y conceptual. Para identificación de las propiedades psicométricas, la versión final del cuestionario YRBS-2007 traducida fue administrada en dos ocasiones con intervalo de dos semanas en muestra de 873 estudiantes de ambos sexos, de la enseñanza de secundaria, en Londrina, Sur de Brasil, en 2007. La confiabilidad prueba-re-prueba fue analizada por el cálculo del índice de concordancia kappa y de la tasa de prevalencia de cada comportamiento de riesgo en las réplicas de aplicación. Se empleó la prueba de chi-cuadrado para identificar diferencias estadísticas entre la primera y la segunda aplicación del cuestionario.
RESULTADOS: Posterior a las discretas modificaciones señaladas en los procesos de traducción, el comité de árbitros consideró que la versión en el idioma portugués del YRBS-2007 presentó equivalencias semántica, idiomática, cultural y conceptual. Diferencias significativas entre las tasas de prevalencias de ambas aplicaciones fueron observadas en 23,4% de los ítems. La identificación del 91% de los ítems con índice de concordancia kappa entre moderado-a-sustancial y valor promedio de dicho índice de 68,6% indicaron la calidad de las propiedades psicométricas del YRBA-2007 traducido al idioma portugués.
CONCLUSIONES: La traducción, la adaptación transcultural y las calidades psicométricas del cuestionario YRBS-2007 fueron satisfactorias, lo que viabiliza su aplicación en estudios epidemiológicos en Brasil.
Descriptores: Traducción (Producto). Cuestionario. Conducta del Adolescente. Psicometría. Reproductibilidad de Resultados. Brasil.
The Youth Risk Behavior Surveillance System (YRBSS) is a surveillance program, originally developed by the Centers for Disease Control and Prevention (CDC) in the late 1980s. It is aimed at monitoring risk behaviors that contribute to the main causes of social problems, morbidities and mortality in American adolescents. The data guiding this survey are obtained from the application of the self-administered Youth Risk Behavior Survey (YRBS) questionnaire, which involves questions about six behavior categories: a) non-intentional injuries and violence; b) tobacco use; c) alcohol and other drug use; d) sexual behaviors that contribute to unintended pregnancy and sexually transmitted diseases; e) dietary habits; and f) physical activity practice.7
The first version of the YRBS questionnaire resulted from extensive research and field testing. It was used by the CDC to conduct the first national survey, in 1991. Surveys are conducted biannually, the questionnaires are revised and possible changes are proposed before each data collection. In 1997, the CDC established the objectives of the Healthy People 2010 and, to guarantee the information required by this program, it suggested changes to the 1999 YRBS version, which led to the creation of 16 new questions, elimination of 11 and adaptation of other 14.3 The 2001, 2003, 2005 and 2007 versions included minor changes.
In Brazil, there is not a systematized health risk behavior surveillance program for the adolescent population. Studies concentrate results on isolated behaviors, such as alcohol and tobacco use,11,13,18,21 illicit drug use,8,15,19 dietary habits1,6,16,20 and physical activity practice.9,14,16,17 Instruments to gather information that have been standardized to the Portuguese language are not easily identified. Specific questionnaires for each study are designed from adaptations of different instruments, which can compromise the quality of psychometric properties and hinder comparisons among studies.
Carlini-Cotrin et al5 translated the 1995 YRBS questionnaire and applied it to primary education students of two 7th and 8th grade classrooms of a state school of the city of São Paulo, Southeastern Brazil, aiming to test its clarity and adequacy. They made minor changes to adapt the questionnaire to the Brazilian context. Despite its relevance, this study did not give due attention to the methodological strictness of the process of translation and adaptation of questionnaires to other cultures. Such limitations suggest caution when using the translated and adapted questionnaire in these conditions.
The present study aimed to validate the psychometric properties of the Brazilian version of the 2007 Youth Risk Behavior Survey.
The translation and cross-cultural adaptation protocols followed the procedures suggested by Guillemin et al.10 The initial translation from the original language (English) to Portuguese was made independently by two researchers, specialists in YRBS. These researchers spoke Portuguese as their native language and were fluent in English, including experiences with translations of academic texts. In addition to the translation, they were requested to record expressions which could have ambiguous interpretation.
A bilingual group composed of three researchers in the area of health compared the translated texts, standardizing the use of ambiguous expressions, and a single version of the questionnaire was designed, summarizing the two previous versions. Next, the questionnaire was back-translated by two other translators in an independent way. The translators selected for this stage were native speakers of English and fluent in Portuguese, working as professors in Brazilian universities. Translators were requested to record expressions that could be unclear in the back-translation process. The bilingual group compared both back-translated texts, designing one single version.
A committee analyzed the translation process and the results achieved in the previous stages. This committee was comprised of nine members, including the authors of this study, the translators who participated in the translation/back-translation process and three university professors in the area of health, all bilingual in English and Portuguese. The committee reviewed the seven available versions of the YRBS questionnaire: the original version in English, two versions translated to Portuguese, summary version of both translations to Portuguese, two back-translated versions, and summary version of both back-translations.
This committee evaluated the types of equivalences between the original questionnaire and the Portuguese version. Members received written guidance on the objective of the study and the definitions adopted for such equivalences. They each completed an analysis form individually, which compared each question with the respective choice of response of the original questionnaire, of the summary version translated to Portuguese and of the summary back-translated version, in terms of semantic, idiomatic, cultural and conceptual equivalences. The analysis form was structured using a differential scale with discrete alternatives: "unaltered", "slightly altered", "greatly altered" and "completely altered".
The Portuguese version of the questionnaire was tested, aiming to investigate the difficulties and suggestions of adolescents in terms of their understanding of questions. Authors of this study applied the questionnaire experimentally to 140 adolescents aged between 14 and 18 years (81 females and 59 males), who were secondary education students of schools of Londrina, Southern Brazil, in April 2007. The results were considered after a new evaluation by the bilingual analysis committee members.
The aspect of reliability associated with reproducibility was investigated to identify psychometric properties, by repeating the application of the translated 2007 YRBS questionnaire. The target population of this study was comprised of students of both sexes, aged between 14 and 21 years, enrolled in public secondary schools of the city of Londrina. According to information from the State of Paraná Department of Education, this population was estimated to be approximately 20,000 students, in the 2007 academic year. The representative sample was obtained through a cluster probabilistic process, using the number of students, in terms of sex, age, school grade and period in which they were enrolled, in each school, as reference. The sample size was established, including 50% prevalences, 95% confidence intervals (95% CI), 3% sampling error and an additional 10% for possible refusals and losses during data collection. The design effect of the sample (DEFF) was defined at 1.5, predicting a minimum sample of 820 students. The final sample was comprised of 873 adolescents. Data were collected between August and November 2007. Demographic characteristics are shown in Table 1.
The classroom selected was visited by researchers and the research objectives and principles of confidentiality, anonymity and non-influence on school performance were clarified to students.
The classroom was revisited and the students who handed in the signed informed consent form received an envelope containing two copies of the Portuguese version of the 2007 YRBS questionnaire with the same codification. Students took one copy of the questionnaire, wrote their respective names on the outer side of the envelope and returned it to the researchers who followed the application. Next, they received instructions on how to self-complete the questionnaire and information about researchers' availability to clarify any questions. After completing and returning the codified questionnaires, they were put in a ballot box with the remaining ones.
After 14 days, the application of the questionnaire was repeated. The classroom was revisited and students who participated in the study received the envelope with their name and were instructed to take out the questionnaire, tear it and throw away the envelope. The following instructions were identical to those given in the first application.
In the analysis of reproducibility, the following two procedures were used: the calculation of the traditional kappa agreement index and prevalence rates of the first and second questionnaire applications. Calculation of the kappa agreement index, followed by the respective 95% CI, was determined for each of the questionnaire items, except for items 6 and 7, classified as numerical variables, which resulted in the calculation of the intraclass correlation coefficient. Agreement among groups of items of the Portuguese version of the 2007 YRBS, according to demographic indicators, reference period of time and risk behavior category, was analyzed using median and mean values of kappa indices. Prevalence rates of the first and second applications were determined from one of the choices of responses to each questionnaire item, following the YRBS methodological recommendations to divide responses into "with risk" and "without risk".4 Statistical differences between the prevalence rates of the first and second applications were estimated using the chi-square test (χ2). Data were analyzed with the SPSS statistical package, version 15.0.
The present study was approved by the Human Research Ethics Committee of the Universidade Estadual de Londrina (Official Opinion 073/07).
Minor divergences in the use of expressions were found in the stages of the translation process. These divergences were discussed by the analysis committee and the expressions that were more easily understood and more frequently used among adolescents were maintained to facilitate understanding. When establishing questions whose literal translation raised doubts about the use of the expression "quão freqüentemente" ["how often"], which is grammatically correct in Portuguese, although rarely used by adolescents, the expression "com que frequência" was chosen. The difficulties and suggestions reported by the group of adolescents did not require changes.
Of all the 87 items that comprise the YRBS, 79% of them were indicated as "unaltered" by the analysis committee, in terms of semantic, idiomatic, cultural and conceptual equivalences. The remaining 21% were indicated by this committee as having at least one of the equivalences "slightly altered". Table 2 shows the summary of the adjustments made to these items. No items of the translated version of the 2007 YRBS questionnaire had the "greatly altered" or "completely altered" choices marked, when compared to the original version.
The statistical indicators associated with the psychometric properties are shown in Table 3. Kappa agreement index varied from 31.6% to 100%, with a mean value of 68.6% and median of 68.5%. According to the qualitative considerations suggested by Landis & Koch,12 68.3% of the items achieved at least "substantial" reproducibility (kappa>61%), and 90.6% achieved at least "moderate" reproducibility (kappa>41%). A total of eight items (9.4%) showed a kappa agreement index<41%. Based on the χ2 test, 23.4% of the items showed significantly different prevalence rates when the application of the questionnaire was repeated. Of all the 19 items that showed significant differences between the prevalence rates of the first and second applications, nine items had a kappa agreement index lower than 61%. The items associated with self-reported weight and height measurements had an intraclass correlation coefficient of 0.98 and 0.97, respectively.
Table 4 shows the median, mean and 95% CI values of kappa agreement indices of the items of the Portuguese version of the 2007 YRBS, according to demographic indicators, reference period of time and risk behavior category. The analysis of reproducibility indicates that day-time students of both sexes tended to show a higher kappa agreement index than boys and night-time students; however, the differences found were not statistically significant. Kappa scores among the grades and age groups were similar. With regard to the reference period of time, the items involving "seven days" showed kappa agreement index scores significantly lower than those of items for longer periods of time. The items indicating reference periods of time equivalent to "30 days" and "12 months" had similar kappa scores, whereas those with periods of time equivalent to "throughout life" and "without time reference" showed the highest kappa index scores.
Dietary habits and physical activity practice showed kappa index scores significantly lower than those of the remaining categories. The sexual behavior category achieved the highest kappa score, followed by tobacco use, alcohol and other drug use, other health-related topics and non-intentional injuries and violence. Of all items that mention a period of time equivalent to "seven days", eight refer to dietary habits and two to physical activity practice. Except for the reference period of time of "seven days", dietary habits and physical activity practice, groups showed scores equivalent to substantial kappa indices (> 61%).
The process of questionnaire translation did not present difficulties due to the methodology adopted and the simple and objective structure of the 2007 YRBS items. The initial translation made by two translators was slightly changed in the subsequent stages. The back-translation, when compared to the original questionnaire, showed minor discrepancies, resulting from the adjustments made to meet the specifications of certain items. The analysis of semantic, idiomatic, cultural and conceptual equivalences, comparable to the cross-cultural adaptation, such as the translation stage, indicated that the questionnaire was easily translated.
The analysis of equivalences showed that the 2007 YRBS domains are appropriate and the attributes used in the original version of the questionnaire are equally valid for the target culture, which meets the cultural equivalence. The conceptual equivalence indicated that few items required adjustments. Items could be considered in a way similar to the original format, once again showing that the structure of development of 2007 YRBS items was well designed. With regard to the idiomatic equivalence, the translated version revealed that approximately 80% of the questions were evaluated as "unaltered" and the remaining ones as "slightly altered" between the original and back-translated versions of the questionnaire. None of the analysis committee members considered any items as "slightly altered", when the original, translated and back-translated questionnaire versions were compared, thus suggesting semantic equivalence.
Brener et al2 administered the 1991 YRBS version to American students in two occasions, with an interval of 14 days. The results indicated kappa agreement index scores between 14.5% and 91.1%, with 72% of items showing kappa>61%. There were no significant differences between the prevalences of repetitions of questionnaire application. In this same study, the analysis of age group showed that better consistency of responses was found among adolescents aged>12 years. A study with a similar design, involving the 1999 YRBS,3 was performed, with results showing kappa index scores between 23.6% and 90.5%, a mean of 60.7%, and 47.2% of items with kappa > 61%. Approximately 22% of items showed statistically significant differences between the prevalences of the first and second questionnaire applications.3
The kappa agreement index in the second application of the Portuguese version of the questionnaire indicated from moderate to substantial reproducibility in the majority of items. The proportion of 68.3% of items with kappa agreement index > 61% was lower than that found in the analysis of the 1991 YRBS (72%) and higher than that found in the analysis of the 1999 YRBS (47.2%). This suggests that this proportion is within the expected limit. Brenner et al2,4 considered the kappa values of the studies as adequate for instruments such as questionnaires. The mean kappa index found in the translated 2007 YRBS (68.6%) was higher than the mean of kappa index of the original version of the 1999 YRBS (60.7%). Data from the original version of the translated 1999 YRBS and 2007 YRBS were similar: 23.4% of the items of the 2007 YRBS in Portuguese and 22% of the items of the 1999 YRBS in the original version showed significant differences between the prevalences of the first and second applications. The items that had significant differences between the prevalence rates of the repetitions of questionnaire application and kappa agreement index < 61% showed questionable reproducibility. Thus, caution is required when interpreting this information.
There were no statistically significant differences in terms of sex, school grade and age group in the indicators of reproducibility between the translated version and the original version of the 2007 YRBS. With regard to the reference period of time, "seven days" showed a kappa index significantly lower than the other reference periods. The reference periods of "30 days" and "12 months" showed similar values, while the reference periods of "throughout life" and "without time reference" showed the highest kappa index scores. The 1999 YRBS version does not include the reference period of "seven days"; in both studies, the reference period of "throughout life" stood out in terms of the kappa index level. Dietary habits and physical activity practice had the lowest kappa indices, such as in the 1999 YRBS. Whereas "tobacco use" had the highest kappa index in the 1999 YRBS version, sexual behavior showed the highest kappa index in the translated 2007 YRBS, followed by tobacco use, alcohol and other drug use, other health-related topics and non-intentional injuries and violence. In the group of items that showed significant differences between the prevalence rates of repetitions of questionnaire application and kappa index < 61%, more than half of the items were associated with dietary habits, physical activity practice and other health-related topics.
The analysis of reproducibility of the original version of the 1999 YRBS and translated 2007 YRBS were expected to indicate lower agreement between repetitions of questionnaire application for items associated with dietary habits, physical activity practice and other health-related topics. Behavior related to use of substances, such as tobacco, alcohol and drugs; sexual activity; and behavior that involves rebelliousness, such as reckless driving and violence, can be more prevalent and considered to have greater importance in adolescents than behaviors associated with routine (dietary habits and physical activity practice). The low kappa index scores (< 41%) and the inconsistency among prevalence rates of repetitions of questionnaire application can reflect behavioral changes in the period of 14 days.
Adolescents aged more than 17 years, who went to school during the day, showed higher kappa index scores than those of adolescents aged between 14 and 17 years, who went to school in the evening. Studies involving the YRBS indicated greater reproducibility in older adolescents.2,3 With regard to the period of study, there appear to be differences between secondary education students who were in day-time schools and those in night-time schools, as a result of paid work performed by these students. In the sample selected for the present study, 22.1% of students who went to school during the day reported performing a certain type of paid job, compared to the 64.9% of students who went to school in the evening, of which 37% performed work full-time (40 hours/week). Night-time students are older and, due to their being in the job market, they remain closer to a non-familial adult environment, have some independence, and can behave in a less usual way, thus having an influence on the change of behaviors in relatively short periods of time, such as the 14-day interval between questionnaire applications.
The Portuguese version of the 2007 YRBS was found to be a questionnaire capable of gathering information about health risk behaviors in Brazilian adolescents. The methodology of translation adopted enabled it to have quality and safety. The simple and objective structure of design of items of the original 2007 YRBS version contributed to the success of this stage.
The cross-cultural adaptation showed that the domains dealt with by the questionnaire are suitable for the Brazilian adolescents' culture and that certain items needed minor adjustments. The identification of the psychometric properties of the Portuguese version of the 2007 YRBS indicated high reproducibility of items.
1. Andrade RG, Pereira RA, Sichieri R. Consumo alimentar de adolescentes com e sem sobrepeso do município do Rio de Janeiro. Cad Saude Publica. 2003;19(5):1485-95. DOI:10.1590/S0102-311X2003000500027
2. Brener ND, Collins JL, Kann L, Warren CW, Williams BI. Reliability of the Youth Risk Behavior Survey Questionnaire. Am J Epidemiol. 1995;141(6):575-80.
3. Brener ND, Kann L, McManus T, Kinchen SA, Sundberg EC, Ross JG. Reliability of the 1999 Youth Risk Behavior Survey Questionnaire. J Adolesc Health. 2002;31(4):336-42. DOI:10.1016/S1054-139X(02)00339-7
4. Brener ND, Kann L, Kinchen SA, Grunbaum JA, Whalen L, Eaton D, et al. Methodology of the Youth Risk Behavior Surveillance System. MMWR Recomm Rep. 2004;53(RR-12):1-13.
5. Carlini-Cotrim B, Gazal-Carvalho C, Gouveia N. Comportamentos de saúde entre jovens estudantes das redes pública e privada da área metropolitana do Estado de São Paulo. Rev Saude Publica. 2000;34(6):636-45. DOI:10.1590/S0034-89102000000600012
6. Dalla Costa MC, Cordoni Jr L, Matsuo T. Hábito alimentar de escolares adolescentes de um município do oeste do Paraná. Rev Nutr. 2007;20(5):461-71. DOI:10.1590/S1415-52732007000500002
7. Eaton DK, Kann L, Kinchen S, Shanklin S, Ross J, Hawkins J, et al. Youth risk behavior surveillance-United States, 2007. MMWR Surveill Summ. 2008;57(4):1-131.
8. Galduróz JCF, Noto AR, Nappo SA, Carlini EA. Trends in drug use among students in Brazil: analysis of four surveys in 1987, 1989, 1993 and 1997. Braz J Med Biol Res. 2004;37(4):523-31. DOI:10.1590/S0100-879X2004000400009
9. Guedes DP, Guedes JERP, Barbosa DS, Oliveira JA. Níveis de prática de atividade física habitual em adolescentes. Rev Bras Med Esporte. 2001;7(6):187-99. DOI:10.1590/S1517-86922001000600002
10. Guillemin F, Bomabardier C, Beaton D. Cross-cultural adaptation of health-related quality of life measures: literature review and proposed guidelines. J Clin Epidemiol. 1993;46:1417-32. DOI:10.1016/0895-4356(93)90142-N
11. Horta RL, Horta BL, Pinheiro RT, Morales B, Strey MN. Tabaco, álcool e outras drogas entre adolescentes em Pelotas, Rio Grande do Sul, Brasil: uma perspectiva de gênero. Cad Saude Publica. 2007;23(4):775-83. DOI:10.1590/S0102-311X2007000400005
12. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1997;33(1):159-74. DOI:10.2307/2529310
13. Machado Neto AS, Cruz AA. Tabagismo em amostra de adolescentes escolares de Salvador Bahia. J Pneumol. 2003;29(5):264-72. DOI:10.1590/S0102-35862003000500004
14. Matsudo SMM, Araújo TL, Matsudo VKR, Andrade DR, Vaquer W. Nível de atividade física em crianças e adolescentes de diferentes regiões de desenvolvimento. Rev Bras Ativ Fis Saude. 1998;3(4):14-26.
15. Muza GM, Bettiol H, Muccillo G, Barbieri MA. Consumo de substâncias psicoativas por adolescentes escolares de Ribeirão Preto, SP (Brasil). I Prevalência do consumo por sexo, idade e tipo de substância. Rev Saude Publica. 1997;31(1):21-9. DOI:10.1590/S0034-89101997000100005
16. Nunes MMA, Figueiroa JN, Alves JGB. Excesso de peso, atividade física e hábitos alimentares entre adolescentes de diferentes classes econômicas em Campina Grande (PB). Rev Assoc Med Bras. 2007;53(2):130-4. DOI:10.1590/S0104-42302007000200017
17. Silva RCR, Malina RM. Nível de atividade física em adolescentes do município de Niterói, Rio de Janeiro, Brasil. Cad Saude Publica. 2000;16(4):1091-7. DOI:10.1590/S0102-311X2000000400027
18. Souza DPO, Silveira Filho DX. Uso recente de álcool, tabaco e outras drogas entre estudantes adolescentes trabalhadores e não trabalhadores. Rev Bras Epidemiol. 2007;10(2):276-87. DOI:10.1590/S1415-790X2007000200015
19. Tavares BF, Béria JU, Lima MS. Fatores associados ao uso de drogas entre adolescentes escolares. Rev Saude Publica. 2004;38(6):787-96. DOI:10.1590/S0034-89102004000600006
20. Toral N, Slater B, Cintra IP, Fisberg M. Comportamento alimentar de adolescentes em relação ao consumo de frutas e verduras. Rev Nutr. 2006;19(3):331-40. DOI:10.1590/S1415-52732006000300004
21. Vieira PC, Aerts DRGC, Freddo SL, Bittencourt A, Monteiro L. Uso de álcool, tabaco e outras drogas por adolescentes escolares em município do sul do Brasil. Cad Saude Publica. 2008;24(11):2487-98. DOI:10.1590/S0102-311X2008001100004
Dartagnan Pinto Guedes
Campus Universitário CEFE/UEL
Rod. Celso Garcia Cid (PR 445) Km 380
86051-990 Londrina, PR, Brasil
The authors declare that there are no conflicts of interest.