1. INTRODUCTION

The Hiroshima University-Dental Behavioral Inventory (HU-DBI), developed by Kawamura in 1988 [1], has been shown to be useful for the assessment of oral health attitudes and behaviour and has effectively been utilized for cross-cultural comparisons [2]. Among other languages, the questionnaire has been successfully translated to English, French, Italian, Korean, and Chinese [3]. Both the original Japanese version and the later developed English version have shown good test–retest reliability [1,4]. The English version, which also displayed good translation validity [5], has been largely used in different countries.

Oral health awareness in Arab countries is generally poor, and dental health education is lacking. Moreover, little is known regarding the population’s knowledge and attitudes toward oral diseases and their prevention in comparison to other countries [6]. In a multinational comparison study using the HU-DBI, no Arabic-speaking countries were involved [3].

Although the utility of English as a second language is increasing in many Arabic speaking countries, socioeconomic and geographic disparities remain to create significant inequalities in English literacy. Therefore, although the English version of the HU-DBI may theoretically be used, the use of a non-native language to respond to a questionnaire may be suboptimal because of language and cultural barriers. There is thus a need for the translation of the HU-DBI into the Arabic language to allow researchers to assess oral health knowledge in Arab countries. Although the various Arab countries each use different dialects, standard Arabic is the official dialect utilized in writing and in official communication and is understood by all Arab populations.

The HU-DBI has been utilized in several Arab countries (Saudi Arabia, Jordan, Sudan, UAE), but the majority of studies have used the English version [7–10]. Although one recent study has used an Arabic translation from English [8], the translation does not appear to have been subjected to adequate validity testing according to recommended guidelines [11,12]. The aim of the present study was the translation, cultural adaptation, and validation of the HU-DBI into standard Arabic.

2. MATERIALS AND METHODS

3. RESULTS

4. DISCUSSION

The aim of this study was the translation and cultural adaptation of the (HU-DBI) into standard Arabic. In line with recommended guidelines [11,12], the process included translation, retranslation, semantic and literal adaptation, and pilot study validation on a target population. Focus group and cognitive interviewing ensured that all translated questions were being understood as intended by the authors of the original questionnaire. Testing in bilinguals confirmed understanding of the questions in responses in both English and Arabic forms. Finally, the high test–retest reliability scores achieved when using the final translated Arabic version attest to the excellent reliability of the translated instrument.

The vast majority of previous research using the HU-DBI questionnaire in Arab countries (Saudi Arabia, Jordan, Sudan, UAE, etc.) used the questionnaire in its English form [7–10]. Only one recent study used an Arabic version translated from English [8]. Although the questionnaire underwent forward and backward translation in addition to reliability testing in bilinguals, only two linguists were asked to answer the English and Arabic versions for reliability testing. It is very probable that the linguists’ knowledge of English and Arabic languages is not representative of the general population’s. Additionally, the researchers reported high percentage agreement and thus concluded a high level of reliability between the English and Arabic languages. Percentage agreement, however, is an unreliable method of testing reliability, and the kappa statistic is recommended instead [13].

In addition to our reporting of the recommended kappa values and confidence intervals, several other factors attest to the strength of the translation and adaptation process used in this study. Compared with previous reports of testing bilingual reliability on samples of 16–37 volunteers [5,14–20], our data are based on 58 bilingual patients fluent in both Arabic and English. In an article discussing the sample size required for pilot studies, the recommended sample size for test–retest pilot studies is 35–40 [21].

Several methods have been described in the recruitment of bilingual samples, ranging from the use of Likert scales for self-rating [22] to asking the participants one short question about their ability to use each of the two languages [18]. The authors of the original HU-DBI report attached a question to the actual questionnaire and requested the participants to describe their proficiency in each of the two languages being assessed using one of the following answers: (1) no difficulty using the language in everyday life; (2) no difficulty using the language in business and professional writing/reading; or (3) same as mother tongue. To ensure the selection of a truly bilingual sample, we opted for the first method and included only participants who rated their proficiency in both the English and Arabic languages with a score of at least 7 out of 10. In a first attempt without considering the degree of proficiency in English, results showed a large discrepancy in answers between the English and the Arabic versions. In this attempt, 10 questions out of 20 had an unsatisfactory kappa result (<0.8). A focus group discussion was performed with eight participants (two dentists, two dental students, three patients, and one translator) and minor changes were introduced to the formulation of the above 10 mentioned questions. The second attempt was then successfully performed on a new group mastering the two languages leading to kappa results greater than 0.8 for all questions.

In line with what is commonly performed in similar studies, bilingual participants were asked to first answer the English version and then the translated Arabic version [3]. The time interval between answering the two questionnaires, however, is debatable. The longer the time interval, the less the recall bias but also the greater the chance that the second questionnaire is not answered (attrition of the sample and losing participants). In our study, and according to patients’ convenience, time interval was at least 1 hour and up to 10 hours (same day), as recommended by several authors [17,20]. Although using a longer time interval may have reduced recall bias, it carries the risk that some habits may change if the person is affected by the information, especially in such a questionnaire regarding dental health.

Finally, focus groups and cognitive interviewing were carried out after ensuring that the reliability in bilinguals was high and that the translation is adequately accurate compared to the original language. The rationale behind this is that monolinguals do not necessarily behave as bilinguals. Bilinguals often experience cultural adaptation and may perceive an Arabic word in a way that is culturally similar to someone whose first language is English. Although in-depth personal interviews are one way of achieving this goal, we chose to organize focus groups in an attempt to improve the validity of our results because such methodology is based on open discussions which do not exist in individual interviews.

5. CONCLUSION

The HU-DBI was translated and adapted into the Arabic language. It was successfully tested for validity and reliability and showed high interrater reliability in both bilingual testing and in repeated Arabic testing. The translated Arabic inventory is believed to be a useful tool, necessary to fill the gap in the dental and oral health fields in Arab countries. The use of a validated Arabic version will enable the assessment of the knowledge and attitudes toward oral diseases and their prevention in Arab countries and will allow meaningful comparisons between these countries.

CONFLICTS OF INTEREST

None declared.