Validation of a Farsi version of the Early Childhood Oral Health Impact Scale (F-ECOHIS)

The original 13-item English version of the ECOHIS was translated into Farsi based on a linguistic translation exercise specifically standardized for translation of the health-related quality of life questionnaires [16, 17]. The translation process consisted of 6 stages which included content and face validity assessments:

1. The original English version was translated into Farsi by two bilingual people whose first language was Farsi. One of them was a health professional, familiar with quality of life questionnaires and the other was a lay person without previous knowledge in this field. Both translators rated the difficulty of translating each item, made comments if necessary and suggested alternatives for difficult words or phrases.

2. The two translators worked with one of authors (S.E.J.) to develop a common consensus translation (CCT). During this stage the readability of the questionnaire was tested [18, 19]. Alternative words were used to make it easy to read even by parents with only primary school education. The group tried to use short and simple words without making the questions shorter, as a longer question would give the respondent more time to think about the question while reading it [20].

3. The CCT version was then translated back into English using two bilingual lay people whose first language was English. The back-translations were compared with the original questionnaire by two English speaking health professionals to make sure that the differences were minor and all items had the same reflections as the originals.

4. An expert committee consisting of one of the authors (A.G.), forward-translators, two other health professionals and a linguistic professional developed the pre-final version of the translation by considering meaning of the words and sentences, applicability of the questions, and cultural issues. Some changes were necessary for the questionnaire to be applicable to the Iranian culture and lifestyle. For example, a high number of Iranian mothers, who will be the main respondents, were housewives with no official job. Therefore, their children's illness would not result in taking time off from work (item 12), but might disrupt their normal daily activities at home or limit the time they usually spend with their other children. The expert committee worked on this stage until all members were satisfied with the semantic, idiomatic, experimental, and conceptual equivalencies [16] of the pre-final questionnaire.

5. Ten parents of 2- to 5-year-old children attending a paediatric dental practice were asked to complete the pre-final version. They were then interviewed about their views of each item. No item was rated as difficult to understand by any of parents. Three items were left blank by one or two parents, and the other 10 items were answered by all 10 parents. This showed a reasonable distribution of responses. Some parents provided comments on font size and style.

6. The final translation was developed in appropriate font and design and stage 5 was repeated with another group of ten parents. No problem was reported by parents who completed the final version of the F-ECOHIS in the second pilot study. Only two respondents had left one question blank.

Adults who attended four clinics in Isfahan and two clinics in Tehran during a specific period in 2009 were asked if they had a 2- to 5-year-old child. They were not necessarily seeking dental or general treatment for the child included in this study. Those who were the main caregiver of a 2- to 5-year old child, and were native Farsi speaker were invited to complete the F-ECOHIS questionnaire. The process was continued until 260 parents were invited. Those who did not give consent for the main or follow up study and those who were not aware of the child's life circumstances in a period of more than three months were excluded.

Participants completed the F-ECOHIS questionnaire and were asked about child's age and sex, reason for their attendance to the clinic, and family's socio-economic status. They were also asked to answer two global self-rating questions about: (1) their satisfaction with their child's dental appearance, and (2) their perception of child's oral health (each answered in four categories). There were a few missing answers to some items of the F-ECOHIS. A value equal to the subject's average score in child or family section, whichever the missing answer belonged to, was given to the missing answer for validity analysis purposes. "I don't know" answers were dealt with in the same way.

The reason for attendance was re-categorized into three categories of 'seeking dental treatment for the subject child', 'routine dental check up for the subject child', and 'other reasons'. The difference between F-ECOHIS scores among the three categories was assessed. The hypothesis behind this test was that those parents who were seeking dental treatment for their children were more likely to report that their children had experienced dental problems, and therefore, would report higher F-ECOHIS scores than those in other two categories. And those who brought their children for a dental check up were more likely to score higher than those attended for other reasons. Kruskal-Wallis Test was used to test the difference between the three categories as the F-ECOHIS score was not normally distributed. Furthermore, the F-ECOHIS scores of those seeking dental treatment was tested against all others using Mann-Whitney Test. These two tests were repeated after adjustments were made for child's age, sex, and city, using Poisson regression model. The association between family's socio-economic status and the F-ECOHIS score was also assessed using the Kruskal-Wallis Test before any adjustment and Poisson regression after adjustments for age, sex, and reason of attendance.

In the next step, answers to the two self-rating questions were compared against the F-ECOHIS scores to test the concurrent (criterion-related) validity of the scale. This is a recommended way to assess the validity of a OHRQoL questionnaire [11, 21]. Kruskal-Wallis test was used for this purpose. It was hypothesized that higher level of satisfaction with dental appearance, better perceived oral health, and sum of these two (values of the two variables were added up and then re-categorized into 3 categories) were related with lower F-ECOHIS score and vice versa. Correlations between scores of the child (9 items) and family (4 items) sections of the F-ECOHIS were tested using Spearman's rank order to evaluate the convergent (construct-related) validity of the F-ECOHIS. The hypothesis related to this test was that the impacts of the child's oral health status on his/her life is closely related to its impacts on the family [11].

A follow up study was conducted to evaluate the test-retest reliability of the scale. One third of those participated in the main study were randomly chosen and invited to complete the F-ECOHIS questionnaire again after 2 weeks. The degree of agreement between the two responses was tested by the weighted kappa and intra-class correlation coefficient (ICC) [22].

Ninety five percent (N = 246) of the 260 invited parents participated in the main study. Those who did not participate either did not give consent for the main or follow up study (3%) or were unaware of the child's/family's events for a period longer than three months (2%). The average age of the children was 4 years (Table 1). Fifty seven percent of children (N = 140) were girls and 58% (N = 103) were from Isfahan. Reasons for attending the clinics were seeking dental treatment for the subject child (24%), routine dental check up for the subject child (19%), and other reasons (57%). F-ECOHIS scores ranged between 13 (the minimum possible score) and 58 (the maximum possible score was 65). The average score was 25.7 with standard deviation of 10.6. The average scores of child and family impact sections were 16.7 and 9.0 respectively (Table 2). The response rate to each item ranged from 96% to 100%.

There was a significant trend in average F-ECOHIS scores among those who attended for dental treatment, those attended for a routine dental check up and those attending for other reasons (P < 0.001). Also, those who attended clinics seeking dental treatment scored significantly higher than the other two categories combined (P < 0.001). The difference exceeded 7 scores (Table 3). The associations between F-ECOHIS score and reason of attendance in the two above-mentioned tests remained significant after adjustments were made for child's age, sex, and city (Table 3). There was no significant difference between F-ECOHIS of those from different socio-economic background.

In response to the self-rating questions, 11.4% (N = 28) of parents said they were very dissatisfied with their children's dental appearance. 12.6% (N = 31) were dissatisfied, 26% (N = 64) were satisfied, and 50% (N = 123) were very satisfied. On the other hand, 8.5% (N = 21) thought their child's oral health status was very bad, 51.2% (N = 126) thought it was bad, 25.6% (N = 63) thought it was good, and 14.6% (N = 36) thought it was very good. There was no missing or "I don't know" answer to these two questions. Answers to these two questions, separately (two tests) and combined (one test), were all significantly associated with F-ECOHIS scores (P < 0.001 for all three tests). These results showed that those parents who were less satisfied with their children's dental appearance and/or thought their children had worse oral health were more likely to give their children higher F-ECOHIS scores and vice versa. This showed that the concurrent validity of the instrument was good. In relation to convergent validity, scores of the child and family sections of the F-ECOHIS were highly correlated (P < 0.001).

The corrected item-total correlation coefficients of the F-ECOHIS, part of its internal consistency reliability measurement, ranged from 0.520 to 0.741. The lowest coefficients were related to "difficulty in pronouncing words" and "missed pre-school or school", and the highest value belonged to "trouble sleeping". The standardized Cronbach's alpha of the F-ECOHIS was 0.926. Cronbach's alpha ranged between 0.918 and 0.925 when items were deleted one by one, and did not increase by deleting any item (Table 4). The Cronbach's alpha coefficient of child and family sections of the F-ECOHIS, when tested separately, were 0.89 and 0.85 respectively. The inter-item correlation coefficients between each two of the 13 items of the F-ECOHIS ranged from 0.300 to 0.730. The weakest relationships were between "difficulty in pronouncing words" and three items of "oral/dental pain", "missed pre-school or school" and "financial impacts on family". On the other hand, the strongest correlation was found between "difficulty in drinking" and "difficulty in eating" items (Table 5). These results together suggested good internal consistency reliability for F-ECOHIS with no item being irrelevant.

Eighty two parents were invited for the follow up study. Nine of them refused, although they had previously agreed to participate in this part. Therefore, the response rate was 89%. Comparison between the test and retest responses revealed a weighted kappa of 0.81 and ICC of 0.82.