Psychometric assessment of the short‐form Child Perceptions Questionnaire: an international collaborative study

Abstract

Objective To examine the factor structure and other psychometric characteristics of the most commonly used child oral‐health‐related quality‐of‐life (OHRQoL) measure (the 16‐item short‐form CPQ11‐14) in a large number of children (N = 5804) from different settings and who had a range of caries experience and associated impacts. Methods Secondary data analyses used subnational epidemiological samples of 11‐ to 14‐year‐olds in Australia (N = 372), New Zealand (three samples: 352, 202, 429), Brunei (423), Cambodia (244), Hong Kong (542), Malaysia (439), Thailand (220, 325), England (88, 374), Germany (1055), Mexico (335) and Brazil (404). Confirmatory factor analysis (CFA) was used to examine the factor structure of the CPQ11‐14 across the combined sample and within four regions (Australia/NZ, Asia, UK/Europe and Latin America). Item impact and internal reliability analysis were also conducted. Results Caries experience varied, with mean DMFT scores ranging from 0.5 in the Malaysian sample to 3.4 in one New Zealand sample. Even more variation was noted in the proportion reporting only fair or poor oral health; this was highest in the Cambodian and Mexican samples and lowest in the German sample and one New Zealand sample. One in 10 reported that their oral health had a marked impact on their life overall. The CFA across all samples revealed two factors with eigenvalues greater than 1. The first involved all items in the oral symptoms and functional limitations subscales; the second involved all emotional well‐being and social well‐being items. The first was designated the ‘symptoms/function’ subscale, and the second was designated the ‘well‐being’ subscale. Cronbach's alpha scores were 0.72 and 0.84, respectively. The symptoms/function subscale contained more of the items with greater impact, with the item ‘Food stuck in between your teeth’ having greatest impact; in the well‐being subscale, the ‘Felt shy or embarrassed’ item had the greatest impact. Repeating the analyses by world region gave similar findings. Conclusion The CPQ11‐14 performed well cross‐sectionally in the largest analysis of the scale in the literature to date, with robust and mostly consistent psychometric characteristics, albeit with two underlying factors (rather than the originally hypothesized four‐factor structure). It appears to be a sound, robust measure which should be useful for research, practice and policy.

Rasch model of the Child Perceptions Questionnaire for oral health–related quality of life

“…The new split into 2 subscales, as we propose in this article, would produce 2 scores instead of 1 overall score and may reflect more accurately the properties of the instrument. This approach is consistent with the study done by Thomson and colleagues, 30 who suggested the same split of items to achieve psychometric accuracy. According to item response theory, 2 or 3 parameter models can take more parameters into account, such as guessing or the fact that items can vary in terms of location and discrimination.…”

“…As we could not fit the Rasch model because of strong indication of multidimensionality, which we describe in the Results section, we decided to split the instrument into 2 subscales according to the item development process published in literature: the results of a large international study (N ¼ 5,804) by Thomson and colleagues 30 in which the authors proposed 2 subscales instead of the originally hypothesized 4-factor structure, with social items and emotional items representing subscale 1, and functioning items and oral symptoms representing subscale 2. After rescoring 1 item, we reexamined the overall model fit, individual item functioning, and unidimensionality.…”

Assessing oral health-related quality of life in children and adolescents: a systematic review and standardized comparison of available instruments

“…This is important for longitudinal studies and clinical trials, where responsiveness and reproducibility are key attributes, as it cannot be assumed that a measure shown to be reliable and valid in cross-sectional studies will necessarily be sensitive to changes over time in a clinical intervention. Therefore, if responsiveness is not demonstrated prior to its application, it is not sure whether this change is real or generated by measurement error [43,44]. On the other hand, developing strategies to facilitate the interpretation of scores (such as estimating the minimal important difference by using anchor-based or distributionbased strategies) may help to extend the use of these instruments.…”

Relationship Between Orthodontic Treatment Need and Oral Health-Related Quality of Life among 11–18-Year-Old Adolescents in Lithuania

“…The most frequently investigated confounding factors were socio-demographic factors such as gender, household income and parents’ education. Because socio-demographic factors may influence OHRQoL, the nature and magnitude of impacts may in fact vary between populations of different cultural backgrounds [ 20 , 21 ]. Therefore, similar studies conducted in varying countries, including Lithuania, have their own inherent value and may be of interest.…”