Comparison of Oral Health–Related Quality of Life among Schoolchildren with and without Cleft Lip and/or Palate

Methods

Participants in this study were volunteer children who were aged 11 to 14 years old in the fall of 2006. Children were chosen from two groups. The first group (CLP sample) included all children of this age in the Oulu and Lapland regions who had had CLP. They were selected from the cleft lip and palate treatment register of Oulu University Hospital (N = 51), which treats all children with CLP in these two regions representing one-fifth of the population of Finland. The second group (school sample) included children of the same age without CLP from four school classes in the Oulu district (N = 82). This was a convenience sample that included two classes from an urban school and two classes from a rural school and represented the general child population of that age.

Data were collected using questionnaires mailed to the children from the CLP sample to their home addresses. Children from the school sample filled out the questionnaire during a lesson. This school group filled out the questionnaire again after 2 weeks to assess the reliability of the Finnish version of the CPQ_11-14. Parental written consent was obtained for all children. Completed questionnaires were received from 26 (51%) children in the CLP sample and 71 (87%) children in the school sample.

The Child Perception Questionnaire (CPQ) is part of the Child Oral Health-Related Quality of Life measure evaluating oral health–related quality of life from the viewpoint of children, parents, and family (Jocovic et al., 2002). The authors selected the age-specific version CPQ_11-14 that has been tested for validity and reliability in several countries and languages (Jocovic et al., 2002; Foster Page et al., 2005; Marshman et al., 2005; Brown and Al-Khayal, 2006; Goursand et al., 2008; McGrath et al., 2008) but not yet in Finland or in Finnish. The CPQ_11-14 contained 37 items and comprised four domains: oral symptoms (six items, such as “How often have you had pain in your teeth, lips, jaws, or mouth?”), functional limitations (13 items, such as “How often has it been difficult to open your mouth wide?”), emotional well-being (nine items, such as “How often have you felt irritable or frustrated?”), and social well-being (nine items, such as “How often have you not wanted to speak or read out loud in class?”). Reference time was the previous 3 months. Each question had five response alternatives varying from never = 0 to every day or almost every day = 4. The CPQ_11-14 was first translated to Finnish and then back-translated to English by a Finnish student and an English student who were not familiar with dentistry to help avoid expressions that would be difficult for the children to understand. The questionnaire was pilot-tested among 10 schoolchildren to find out if they understood the questions. The phrasings of some questions were then slightly modified based on what the children reported hard to understand (like biting or chewing corn on the cob) and these phrasings were cross-checked with one of the developers of the original scale.

The questionnaires also included questions about age and sex and a global rating of the respondent's oral health and its overall effect on the respondent's life. The response alternatives on questions about oral health varied from excellent = 0 to poor = 4. The question about the overall effect of oral health on life also had five response alternatives varying from not at all = 0 to very much = 4. The participants were also asked if they had received any of the following dental treatments: fillings, orthodontic treatment, extraction, or local anesthesia within the previous 3 months, earlier, or never. For the test-retest phase the authors also included information about whether the children thought their dental health had changed between the two surveys or not.

An overall CPQ_11-14 score was generated by adding the responses (range = 0-148). The sum scores for each of the four domains were also generated. These sums ranged as follows: oral symptoms (range = 0-24), functional limitations (range = 0-36), emotional well-being (range = 0-52) and social well-being (range = 0-36). In both surveys/time points, six children had missing data in the CPQ_11-14-items. These were handled using the mean imputation method (Fayers et al., 1998).

Differences according to age and sex between schoolchildren without CLP and children with CLP were evaluated with cross-tabulations and chi-square tests. Differences in CPQ_11-14 total and subscales between children with CLP and schoolchildren without CLP were evaluated using Mann-Whitney U tests. Associations between the CPQ_11-14 scale and global ratings of oral health and overall well-being were evaluated using Spearman's correlation coefficients, which also reflect convergent validity, for example, whether measures of constructs that theoretically should be related are in fact related.

The reliability of the questionnaire was assessed using internal consistency and test-retest measures. Internal consistency was evaluated by Cronbach alpha. The test-retest measurement was assessed by intraclass correlation coefficient (ICC) for consistency with two-way mixed model for single measures, first, among all children from the school sample and, second, among children from the school sample who reported that their dental health had not changed between the two surveys. For deciding analysis methods and evaluating potential clustering effect of students from the same class, multilevel modeling and discriminant and classification analyses were used. Multilevel models were fitted with the mixed models procedure separately for overall CPQ_11-14 score and each domain. The results showed no clustering effect. Class was estimated to explain between -3.5% and 5.6% of variance in the total sum and the domains (model and variance component estimation). Discriminant function analyses were used with cross-validation. The data were randomly assigned to two groups. Based on the first data set (n = 35), there were no statistically significant discriminant functions. When classifying the children in the second data set (n = 36), only 7 (19%) were classified to the category to which they actually belonged. According to all of these results, no clustering of scores among students in the same school class were found, so it was safe to perform the analyses without taking into account the potential clustering effect of children belonging the same school class. The statistical analyses were performed with SPSS 21.0 (IBM Corp., Armonk, NY).

This study was approved by the Ethics Committee of the Northern Ostrobothnia Hospital District, and written consent was obtained from the parents.

Results

No statistically significant differences were found between the CLP and school samples according to age and gender (Table 1). The CPQ_11-14 sum scores ranged from 1 to 43 among the school sample, and from 40 to 82 among the CLP sample. The children in the CLP sample reported poorer overall oral health–related quality of life in all dimensions than did the children in the school sample (Table 2).

OPEN IN VIEWER

Table 1

Distributions of Age and Gender Among Schoolchildren Without CLP and Children With CLP (Chi-Square Tests)

	School Sample (n = 71)	CLP Sample (n=26)	P
Sex
Boy	56	50	.579
Age, y
11	17	12	.768
12	28	31
13	39	34
14	16	23

OPEN IN VIEWER

Table 2

CPQ_11-14 Scores Among Schoolchildren Without CLP and Children With CLP (Mann-Whithey U Tests)

	School Sample (n = 71)			CLP Sample(n = 26)
	Median	Mean	SD	Median	Mean	SD	P
CPQ sum	13	15.0	11.0	52	55.5	12.3	<.001
Oral symptoms	5	5.1	2.7	11	11.9	3.5	<.001
Functional limitations	2	2.8	2.7	13	14.0	5.0	<.001
Emotional well-being	2	4.2	5.2	11.5	12.6	3.3	<.001
Social well-being	2	2.9	3.4	16	17.1	4.6	<.001

Overall oral health was correlated with the oral health–related quality-of-life total score as well as oral symptoms and social well-being scores among children with CLP, although this was not the case among the schoolchildren. For other dimensions and for overall well-being, significant correlations were not found (Table 3). However, when questionnaires were administered for the second time among the school sample, there were significant positive correlations between global rating for oral health and the total CPQ_11-14 scale (r = 0.36, P = .002) and the oral symptoms (r = 0.41, P < .001) and functional limitations (r = 0.29, P = .014) domains.

OPEN IN VIEWER

Table 3

Spearman Correlation Coefficients Between CPQ_11-14 Scores and Global Ratings of Oral Health and Overall Well-being, separately for Children With and Without CLP

	Oral Health				Overall Well-being
	School Sample		CLP Sample		School Sample		CLP Sample
	r	P	r	P	r	P	r	P
Total scale	0.19	.114	0.45	.022	-0.03	.811	0.17	.404
Oral symptoms	0.18	.137	0.44	.029	-0.08	.517	0.12	.558
Functional limitations	0.20	.104	0.37	.069	0.02	.906	0.05	.793
Emotional well-being	0.11	.366	0.15	.471	-0.07	.567	0.02	.930
Social well-being	0.05	.656	0.42	.038	0.11	.355	0.11	.596

As assessed with the Cronbach alpha value, internal consistency was 0.97 for the total scale and varied between 0.81 and 0.94 for the subscales. All items contributed to the scale and its domains. The test-retest reliability was acceptable. Among all children in the school sample, the ICC was 0.79 for the total scale and varied between 0.65 and 0.74 for subscales. The results were similar among children whose state of oral health had remained the same between surveys.

Discussion

These results indicate that children with CLP have significantly poorer overall oral health–related quality of life and in all dimensions compared with the schoolchildren without CLP. The difference was largest in the social well-being dimension.

Previous studies have used different oral health–related quality-of-life measures so the results of these studies cannot be directly compared (Bos and Prahl, 2011; Broder et al., 2012; Antonarkis et al., 2013; Dak-Albab and Dashash, 2013; Eslami et al., 2013; Ward et al., 2013). However, the results are considerably different from those among Danish children of same age using same outcome measure (Wogelius et al., 2009). The CPQ scores of the schoolchildren were higher among children in the current study compared with Danish schoolchildren (15.0 versus 10.5) and the differences between children with CLP were even greater (55.5 versus 10.2, respectively). This may be due to treatment variations between Danish and Finnish centers. On the other hand, the convergent validity varied across studies conducted in different languages (Jocovic et al., 2002; Foster Page et al., 2005; Marshman et al., 2005; Brown and Al-Khayal Z, 2006; McGrath et al., 2008), which suggests that there might be some cultural differences as well. This may not be the case, however, between Denmark and Finland.

The fact that overall oral health was correlated with the oral health–related quality-of-life total score as well as with the oral symptoms and social well-being scores among children with CLP but not among schoolchildren, possibly indicates that children with CLP can cope with their life in general but perceive handicaps in oral health–related issues. It may also indicate that the meaning of oral health in general differs in children with different levels of complex oral diseases and the varying spectrum of CLP. Additionally, this result indicated that convergent validity was moderate only among a group reporting high levels of oral impacts and was weak among all children as well as for the subscales.

According to this study, the Finnish version of the CPQ_11-14 was reliable. The retest validity of the total scale was more than 0.70, which is suggested as acceptable for research purposes; furthermore, the internal consistency of the total scale and the subscales ranged from good to excellent (Streiner and Norman, 2008) and were on the same or better level compared with other studies (Jocovic et al., 2002; Marshman et al., 2005; Brown and Al-Khayal., 2006; Goursand et al., 2008; McGrath et al., 2008). The construct validity was also good in terms of its ability to find differences between groups.

The weak convergent validity of the questionnaire among children in the school sample may be due to the low prevalence of oral diseases among them. On the other hand, children's responses may be influenced by other things, such as their physical or emotional state, social norms, or behavior (Jocovic et al., 2005). The convergent validity varied across studies conducted in different languages, which suggests that there might be some cultural differences as well (Jocovic et al., 2002; Foster Page et al., 2005; Marshman et al., 2005; Brown and Al-Khayal, 2006; Broder and Wilson-Genderson, 2007; McGrath et al., 2008). An observed association between oral health and the CPQ total scale and two subscales in the second survey may reflect that the children have become cognitively more aware about the concept of health and oral health or that they started to pay more attention to their oral health and its impacts on their lives after the first survey. Weak or varying convergent validity might also indicate that either global questions do not fully capture the multidimensional construct of oral health–related quality of life or that CPQ_11-14 measures only the impacts of oral diseases and disorders rather than oral health–related quality of life.

The strength of this study was asking children themselves about the impact of their condition on their daily lives. The questionnaire was developed for 11- to 14-year-olds and, thus, we chose all children with CLP of this age from the two northernmost regions of Finland representing one-fifth of the population of Finland. The control group consisted of a slightly higher number of children without CLP. As practically all children in Finland attend publicly funded schools we could easily reach children of the same age who filled out the questionnaires in their classrooms. This convenience sample of schoolchildren represented the general child population better than a clinical population would have as children with healthy mouths were included. With this approach, the response rate was good in the schoolchildren group (87%) though it was relatively low in the CLP group (51%), which may be due to the fact that questionnaires were mailed to the children from the CLP sample while the school sample filled out the questionnaire at school during a lesson. Some children with CLP might have been so severely handicapped that they were unable to respond. This could not be checked for ethical reasons, but the sample sizes and response rate were sufficient to ensure comparison. Possible nonresponse bias might have resulted due to better oral health–related quality of life among the responding children with CLP than among all children with CLP. However, the authors do not think the way the questionnaires were administered affected the results in other respects as guidance from adults was available for both groups if needed. Because patients with CLP undergo many more cleft clinic visits and cleft-related treatments for speech, hearing, dental, and orthodontic indications, they may have perceived treatment biases that the schoolchildren without CLP do not experience. Protocol differences in the treatment timing and delivery between Danish and Finnish centers may also amplify differences in bias.

Of the available methods for replacing missing CPQ_11-14-item responses, the authors used the mean imputation method (Fayers et al., 1998), which was adequate for this study as only a few children had missing item values, and each child only had a few missing items. Because the results of multilevel modeling and discriminant analyses revealed no clustering of scores among students in the same class, the methods used to assess the reliability and construct validity of the Finnish translation of the CPQ_11-14 were eligible for this study.

Conclusions

The oral health–related quality of life of Finnish children with CLP was considerably poorer than that of their peers in overall and all dimensions, especially social well-being. The Finnish version of the CPQ_11-14 showed good construct validity and reliability, thus indicating that it can be used for assessing oral health–related quality of life instead of developing disease-specific instruments.