The development, validation, and psychometric properties of the Japanese version of the Child Oral Health Impact Profile-Short Form 19 (COHIP-SF 19) for school-age children

Oral health is an important component of overall health, as oral disorders can have a significant impact on physical, social, and emotional well-being [1]. For example, among oral disorders, dental caries (tooth decay) [2], missing teeth [3], malocclusion [4, 5] and craniofacial anomalies [6] can adversely affect quality of life (QoL). Therefore, oral health-related quality of life (OHRQoL) is becoming increasingly important in both population research and clinical practice [1].

A variety of questionnaires measuring OHRQoL have been developed for adults [7‐9]. However, age-specific instruments are still required for better analysis of OHRQoL. As children and adolescents have different QoL-related issues compared to adults [10], various instruments for measurement of OHRQoL in pediatric populations have been developed over the past several decades, despite the difficulties associated with the development and validation of such instruments [11]. These include the Scale of Oral Health Outcomes for 5-year-old children [12], the Pediatric Oral Health-Related Quality of Life Measure [13] and the Child Oral Impacts on Daily Performances Index [14]. However, the most frequently used self-completed QoL scales for children are the Child Perceptions Questionnaire (CPQ) [15] and Child Oral Health Impact Profile (COHIP) [16].

The CPQ was the first validated questionnaire specifically designed to measure OHRQoL in children and adolescents [15]. The developers of the CPQ considered theories of child developmental psychology, and created different versions for specific age ranges, e.g. the CPQ_8–10 for 8–10-year-old children and CPQ_11–14 for adolescents [17, 18]. However, it is unclear whether the different versions are consistent, i.e. whether scores are comparable among them [19]. Therefore, care is required when conducting longitudinal research utilizing the CPQ [20].

The COHIP, which was originally developed to assess “oral–facial well-being” over a range of ages (8–15 years of age) [21] and ethnicities [22‐24], is also a well-validated and comprehensive questionnaire for determining children’s OHRQoL. It contains 34 questions and five subscales (oral health, functional well-being, socio-emotional well-being, school environment, and self-image). The COHIP questions cover the oral and maxillofacial areas. An important characteristic of this scale is the inclusion of positive aspects of OHRQoL (e.g. confidence and attractiveness). Child Oral Health Impact Profile-Short Form (COHIP-SF) 19 is a shortened version of the scale, developed in 2012, containing 19 items and three subscales (oral health, functional well-being, and socio-emotional well-being); the psychometric properties of the original version are well-maintained [25]. The short form can be administered more quickly, which facilitates QoL assessment in clinical studies [20, 25]. COHIP-SF 19 has been translated into Arabic [26], German [19] and Chinese (Mandarin) [27]. As models of OHRQoL should take account of cultural factors [1], the COHIP-SF 19 is a good instrument for international comparisons of OHRQoL in child and adolescent populations. In addition, because its target age has been extended from 8–15 to 7–18 years of age [25], the COHIP-SF 19 is suitable for longitudinal research, such as QoL studies in patients with cleft palate who require long-term treatment commencing after birth and continuing into adulthood [20].

The present study was performed to develop and validate a Japanese version of the COHIP-SF 19 to assess its psychometric properties in Japanese school-age children. The development of different language versions of the COHIP is beneficial for future international comparisons of children’s OHRQoL as an outcome of clinical interventions in dental and craniofacial/maxillofacial regions.

The development of a validated Japanese questionnaire is essential for the efficient and effective assessment of the OHRQoL of school-age children in Japan. This study was performed to adapt the original English COHIP-SF 19 to a Japanese cultural context and to test the psychometric properties of the COHIP-SF 19 JP in school-age children. In this study, the COHIP-SF 19 JP was developed in accordance with published guidelines [28], including the procedures, translation, back translation, conceptual equivalence confirmation by the original developer, and reliability and validity assessment with a population sample. Our results indicated that the COHIP-SF 19 JP possesses satisfactory psychometric properties for use in the targeted age group.

We have obtained 379 completed questionnaires for the assessments, which was sufficient for Cronbach’s alpha and covariance structure analyses, such as CFA. However, the response rate was relatively low. This may have been because the questionnaire was paper-based and answered at home rather than being completed at school; moreover, there was no compensation for participation, and each parent and child had to sign a consent form at home.

Interestingly, females had a significantly higher socio-emotional well-being subscore than males, which was opposite to the results obtained using the Dutch COHIP 34 [22] and Korean COHIP 34 [23]. A previous study using the Chinese COHIP-SF 19 found no sex differences in total or socio-emotional well-being subscores, but oral health and functional well-being subscores were higher in females than in males [27]. With regard to the differences that we noted according to age group, the increased functional well-being subscore in the older group may have been related to the fact that the younger (elementary school) group often had mixed dentition. In addition, the social-emotional well-being subscore was lower in the older age group, suggesting increased concern regarding appearance in Japanese adolescents. General health-related quality of life (HRQoL) has often been reported to be higher in adolescents than children, and females tend to show lower values than males [35, 36]. Future studies using COHIP-SF 19 JP should assess whether sex and age differences are due to specific characteristics of Japanese culture.

In this study, we investigated the internal consistency of COHIP-SF 19 JP reliability using Cronbach’s alpha coefficient. Cronbach’s alpha for the overall COHIP-SF 19 JP was 0.77, similar to Cronbach’s alpha in the original study (0.82–0.88) [25], the Chinese version (0.81) [27], and the German version (0.78–0.80) [19]; this finding indicated good internal consistency of the COHIP-SF 19 JP. Subscales showed relatively low Cronbach’s alpha values (0.45–0.68), although they tended to be similar to the Cronbach’s alpha values in the Chinese (0.59–0.74) and Arabic (0.57–0.67) versions. Notably, Cronbach’s alpha values for Oral health and Functional well-being subscales had lower scores than in previous studies, which may have been due to the small number of items included in these two subscales [37]. Three items (Q7, Missed school for any reason; Q8, Been confident; Q15, Felt that you were attractive [good-looking]) had relatively low item-rest correlations (< 0.2). These results were similar to the findings of the German version [19]; conversely, in the Chinese version [27], Cronbach’s alpha did not increase even if any of the items were deleted. Generally, items with inadequate psychometric properties may be either removed from the questionnaire or modified. However, these items were retained in our study, to allow comparison with international studies, as suggested by a study using the German version [19].

The test-retest reliability (ICC) for the overall scale of the COHIP-SF 19 JP was excellent and above the recommended threshold [38]. In addition, our ICC score for the overall scale was 0.82, which was higher than ICC scores observed in different language versions, such as the Arabic (0.76) and Chinese (0.77) versions [26, 27]; this finding suggested that the COHIP-SF 19 JP demonstrates sufficient stability over time.

Construct validity was assessed by examining discriminant and convergent validity [25]. The COHIP-SF 19 JP discriminant validity showed significant relationships with the statuses of dental caries and malocclusions. Children without dental caries reported a higher OHRQoL than children with dental caries; this finding was also observed with respect to malocclusion in the present study. Broder et al. [25] reported that US Latino children with dental caries in their permanent teeth had significantly lower scores in the overall COHIP-SF 19 and oral health subscale. Our data also showed that participants with malocclusion appeared to have lower COHIP-SF 19 JP scores than those without malocclusion, consistent with the findings of previous studies that used the Arabic and Chinese versions [26, 27]. Therefore, the COHIP-SF 19 JP exhibits sufficient discriminant validity for these oral clinical outcomes. However, gingivitis and plaque retention scores showed no significant associations with COHIP-SF 19 JP scores, which was presumably because the impacts of plaque and gingivitis on OHRQoL might have been insufficient for detection by the COHIP-SF 19 JP in our sample. The results regarding convergent validity of COHIP-SF 19 JP were satisfactory, as higher self-perceived oral health rating was associated with higher COHIP-SF 19 JP score, which was consistent with the findings of previous studies.

CFA is commonly used to examine the structure of instruments, such as OHRQoL [39‐42]. With CFA, it is possible to specify precisely which items should load onto which factor. All relationships between factors and variables can be specified in advance in the model. Subsequently, the fit of the model to the data can be tested. An overall test statistic, along with a number of descriptive fit measures, may be obtained to evaluate the degree to which the model fits the data [43]. The main test statistic is the chi-square statistic. However, chi-square is an extremely sensitive statistical test, which is not interpretable in a standardized way and is not a practical test of model fit. Therefore, we used the chi-square/degree of freedom (χ²/DF) ratio, which is less affected by sample size [44]. In accordance with previous reports [34, 44], several different model indices were also used to evaluate the degree of fit of the data to the model, including RMSEA, GFI, AGFI, CFI, and AIC. RMSEA describes how closely the model fits the population, with lower values indicating better fit. GFI is a different type of measure where the model of interest is assigned a score between 0 and 1, with higher values indicating better fit. AGFI adjusts for the GFI model’s degrees of freedom relative to the number of observed variables, and therefore favors less complex models with fewer parameters. CFI is derived from comparison of the χ²/DF ratios between the null and alternative hypotheses. AIC adjusts the chi-square statistic for the number of parameters estimated, and can be used to compare competing models that do not need to be nested [34]. Our CFA demonstrated that the data collected with the three-factor model showed acceptable fit values, according to the previous report [34]; the three-factor model is the same structure as the original COHIP-SF 19. However, the CFA for the four-factor model provided better factor loadings and slightly better acceptable fit values, suggesting that the COHIP-SF 19 JP should include an additional factor. The two items, Q8 and Q15, composed the potential additional factor in our four-factor model. Exploratory factor analysis of the Arabic version of COHIP-SF 19 suggested that the four-factor model exhibited better fit, with the same two questions composed the additional factor [26]. These two items showed small factor loadings, strong correlation between error covariances in the three-factor model (Fig. 1), and low item-rest correlations (Table 4). Only these two items were actually positively worded questions, which originally belonged to a separate subscale, “Self-image,” with four other items in the full version of the COHIP-34. When developing the short version, only these two items remained, and were integrated into a single “Socio-emotional well-being” subscale. Although it appears inappropriate to include them in the Socio-emotional well-being subscale, we eventually chose to retain them for the purpose of international comparisons, as discussed above. However, care is needed when analyzing data from Q8 and Q15, as some characteristics of Japanese culture would have an influence. Indeed, modesty/humility is still one of the most important virtues in Japanese culture, so the positive questions regarding self-image may not reflect the QoL of Japanese children as well as those in other cultures.

Q7: Missed school for any reason because of your teeth, mouth, or face, also showed the smallest factor loading and the lowest item-rest correlation among all items. This was presumably due to differences in cultural background or the inclusion/exclusion criteria in our study. Because we excluded children who had current or prior orthodontic treatment, the majority of children requiring periodic visits to a dental clinic/hospital during the day had been omitted; this may have led to the high score with minimal variation. However, this question appears to be very important, especially when assessing the outcomes of patients with cleft lip and palate who frequently visit dental hospitals in Japan.

Use of the COHIP-SF 19 provides a number of opportunities for researchers and clinicians. Because of its short length, it places less burden on the patient, relative to the original 34-item version or other long instruments, such as the 37-item CPQ_11–14 [15]. Another advantage of the COHIP is the wide age range, which allows longitudinal assessments that involve the COHIP in long-term prospective studies of treatment effects and prognoses. Moreover, the COHIP asks questions regarding the oral area and the maxillofacial region. For example, one question in the CPQ_8–10 is “In the past 4 weeks, how often have you: Been concerned what other people think about your teeth or mouth?”; a similar question in the COHIP is “In the past 3 months, how often have you: Been worried about what other people think about your teeth, mouth, or face?”. Therefore, the COHIP may enable more effective evaluation when researchers and/or clinicians assess patients with diseases that affect the maxillofacial area (e.g. cleft lip and palate), which require a very long treatment period.

The International Consortium for Health Outcomes Measurement (ICHOM), has recently focused on the development of standardized datasets of subjective and objective outcomes and case-mix factors for use in clinical practice [45]. To measure the outcomes of patients with cleft lip and palate, the COHIP is recommended by the ICHOM for assessment of the oral area, along with CLEFT-Q [46]. Therefore, the development and use of the COHIP-SF 19 JP are likely to be especially beneficial for longitudinal studies or international comparisons of children’s OHRQoL.

This study had some limitations: the sampling was unbalanced and was conducted in a non-random manner. The difference between DMFT and malocclusion indices had a significant but modest impact on OHRQoL values. The sampling bias associated with the lower prevalence of oral diseases in urban schools could have led to the relatively low discriminant validity. In addition, the quality of clinical dental examination and the low response rate may also have affected the discriminant validity. As we did not use the results from the clinical examination specifically conducted for the purpose of this study, it was not possible to implement a calibration process for the examinations. In addition, the low response rate may have made the survey vulnerable to nonresponse bias. Evaluation of discriminant validity in additional pediatric populations is warranted. Further longitudinal and interventional studies may be required to better evaluate longitudinal validity and the sensitivity of the measurements. It should be noted that COHIP-SF 19 JP had better model fit indices when using the four-factor model.

The Japanese version of the COHIP-SF 19 was validated in a representative community sample of 7–18-year-old Japanese schoolchildren. The COHIP-SF 19 JP was successfully developed in accordance with the standard procedure for cross-cultural adaptation of a self-reported instrument; it showed sufficient psychometric properties for use in Japanese school-age children.