Background
Planning dental treatment within a public health system requires information on the prevalence and distribution of oral diseases [
1]. However, normative treatment needs, reflected in clinical oral indicators, provide little information about the patients' self-perceived treatment needs. To overcome this limitation, oral-health-related quality-of-life (OHRQoL) instruments have been developed to assess the impact of oral health on daily life activities [
2]. According to Locker [
3], the subjective perception of oral health and treatment needs is considered to be the consequence of oral conditions, although studies that have investigated the relationship between subjective and clinical oral health indicators have shown both strong and weak significant associations and even the absence of any relationship [
4]. Numerous studies have identified a gap between professionally and self-defined oral health, suggesting that they document different dimensions of the human experience, which are conceptually and often empirically distinct, with different implications for treatment need [
5]. Consequently, OHRQoL instruments are recommended to supplement clinical measures and as adjuncts to them [
4].
Whereas clinical oral health indicators refer to specific oral conditions, such as dental caries, periodontal disease, and malocclusion, most OHRQoL indicators are generic in that they assess the overall impact of oral problems by considering numerous oral conditions. In contrast, condition-specific (CS) OHRQoL measures focus on particular diseases, conditions, symptoms, functions, or populations, and should be used when any of these attributes must be assessed [
1]. CS instruments provide information about the consequences of a specific, untreated oral condition and the corresponding benefits of its treatment. This might make CS instruments more sensitive to small but clinically relevant changes in oral diseases than both generic HRQoL and OHRQoL instruments [
1,
6]. Assuming that oral conditions have consequences for more widespread health issues, Allen et al. [
7] compared the validity of the Oral Health Impact Profile (OHIP) with a generic HRQoL instrument, SF36, in edentulous patients seeking implants or conventional dentures. Whereas OHIP discriminated between three clinically disparate groups, SF36 did not. Lee et al. [
8] compared the performances of the Pediatric Quality of Life Inventory and the Early Childhood Oral Health Impact Scale and showed that the latter instrument was superior in identifying those children affected by early childhood caries from those without caries. However, with few exceptions, the superiority of CS measures to generic HRQoL and OHRQoL instruments has yet to be established [
1,
9‐
11].
One of the most commonly used OHRQoL instruments, the Oral Impact on Daily Performances (OIDP), is designed to be used both as a generic and a CS instrument. As a CS instrument, it can link specific oral conditions to an individual's quality of life [
11]. The Child-OIDP [
12], derived from the adult OIDP version, has been shown to be applicable to school children across occidental and non-occidental socio-cultural contexts, when used as self-administered questionnaires or in face-to-face interviews [for a review, see [
13]]. However, there is little empirical evidence about the relationship between the Child-OIDP and various oral diseases or on whether those relationships vary across socio-cultural contexts. Few studies have compared the capacities of the generic and CS Child-OIDP inventories to discriminate between groups with different levels of normative treatment needs, as part of a construct validity assessment [
14].
In Tanzania, dental diseases have remained at moderate levels, and approximately 30%-40% of the population, irrespective of age, is reportedly free of dental caries. However, Tanzanian children have for many years demonstrated a high prevalence of untreated dentinal lesions, with a majority located in molars, which show relatively slow progression [
15]. Recently, 19.2% of a sample of rural school children was identified with normative treatment needs for dental caries [
16]. Periodontal problems have been reported to account for 80% of all oral diseases in the Tanzanian population [
17]. Poor oral hygiene at an age of 15 years or older is very common (65%-99%) and the prevalence of gingivitis is reported to range from 80% to 90% [
18,
19]. Previous studies have indicated a wide variation in the prevalence of malocclusion, ranging from 45% to 97% among school children [
20]. Exposure to dental services is low in this country, particularly in rural areas, and dental pain and discomfort have been cited as common reasons for seeking dental care [
17]. Information is needed about the generic and CS impacts of periodontal disease, dental caries, and malocclusion on children's quality of life, to guide the assessment of the dental treatment needs of Tanzanian school children.
Purpose
Focusing on school children, this study compared the discriminative ability of the generic Child-OIDP for dental caries and periodontal problems across socio-culturally different study sites (Arusha and Dar es Salaam) in Tanzania. The discriminative ability of the generic and CS Child-OIDP attributed to dental caries, periodontal problems, and malocclusion were then compared with respect to various oral conditions among school children in Dar es Salaam, as part of a construct validation.
Discussion
The assessment of OHRQoL in children is a relatively recent initiative and CS measures are yet to be applied [
30‐
32]. Because of the plethora of oral conditions that affect the quality of children's lives, the issue of describing the CS impact has remained a challenge [
1]. This study assessed for the first time the discriminative ability of the generic Child-OIDP across various socio-cultural contexts in Tanzania, and compared the discriminative abilities of the generic and CS Child-OIDP inventories with respect to normative treatment needs.
About half the school children in Arusha reported experience with any oral impacts on daily performances. This rate is higher than those reported previously in similarly aged groups of Tanzanian school children, but lower than those observed in Uganda and other developing countries [
33‐
35]. Not unexpectedly, the younger primary school children in Dar es Salaam had less caries experience and a lower prevalence of impacts as assessed by the generic Child-OIDP than their older counterparts in Arusha. Nevertheless, the performance of the generic Child-OIDP inventory in distinguishing between subjects with and without dental caries and periodontal problems did not vary across the study sites. Both the overall means and the generic prevalence scores revealed that oral problems had a greater impact on children suffering caries and periodontal problems than on their counterparts without these problems. This supports the construct validity of the Child-OIDP when used in Tanzanian school children. Although the generic Child-OIDP scores are less comparable to the specific normative treatment needs for dental caries and periodontal problems, the positive association observed might be explained by inferring that dental caries and periodontal problems contribute greatly to the burden of oral impacts on children's quality of life. In a previous study, toothache was recognized as the main cause of six of eight performance impacts of school children in Kinondoni district and four of eight impacts of school children in Temeke district, in Dar es Salaam [
13]. A mouth ulcer and bleeding and swollen gums were among the causes most frequently listed by those school children [
13]. Studies conducted elsewhere have shown similar results. Oral conditions related to dental caries, such as toothache and sensitive teeth, had the greatest reported impact on the quality of life in 11-12-year-old children from developing countries [
13,
36]. Despite differences in the prevalence of Child-OIDP and in the modes of administering the inventory across the study sites, neither the discriminative capacity of the generic instrument with respect to dental caries and periodontal problems nor its internal consistency (reliability) varied across the study sites. Previous studies that compared self- and interviewer-administered Child-OIDP inventories in the same study group found that the instrument showed acceptable psychometric properties irrespective of the mode of its administration [
37,
38].
As shown in Table
3,
4 and 5, the prevalence of oral impact obtained with the generic Child-OIDP was higher than that obtained with the CS Child-OIDP. Both the generic and CS Child-OIDP rates were relatively low compared with those obtained in children using other OHRQoL instruments. This might be attributable to the fact that the ultimate impacts assessed by OIDP are rare in most study populations [
30]. From the overall mean scores and the prevalence scores, both the generic and CS Child-OIDP inventories indicated that children with caries, periodontal problems, or malocclusion experienced a greater oral impact than those without these conditions. This corroborates previous studies that showed that children suffering from various dental diseases and clinical symptoms have a poorer OHRQoL [
13,
33]. Using the thresholds defined by Cohen [
29], the effect sizes for the generic Child-OIDP were small when children with normative treatment needs for dental caries and periodontal problems were compared with those without such treatment needs, and were almost negligible when children with and without orthodontic treatment needs were compared. In contrast, the effect sizes related to the mean differences in the CS Child-OIDP scores were negligible when children with and without periodontal problems were compared, moderate when children with and without malocclusion were compared, and large when children with and without dental caries were compared. The present findings agree with those of previous studies [
6,
14], indicating that the two forms of the Child-OIDP are complementary rather than alternative sources of information. Nevertheless, the CS Child-OIDP was better suited than the generic Child-OIDP to identifying school children according to their normative treatment needs for malocclusion and dental caries. When assessing the strength of the association between the clinical indicators and the prevalence of oral impact, the ORs were larger when the CS Child-OIDP attributed to dental caries and malocclusion was used than when the generic Child-OIDP was used, even after adjustments were made for socio-demographic factors (Tables
3 and
4). This finding corroborates some previous studies but is inconsistent with others. A recent study of Thai school children revealed that the generic and CS Child-OIDP inventories distinguished equally well the groups with and without normative treatment needs for dental caries [
14]. Comparing the generic and CS Child-OIDP assessments of malocclusion in Brazilian adolescents, Bernabé [
6] found that both inventories were able to discriminate between subjects with and without treatment needs. However, the CS Child-OIDP showed the largest effect size and therefore appeared to be the form best able to differentiate between groups of adolescents. Other studies have compared the discriminative abilities of generic HRQoL and OHRQoL instruments with respect to early childhood caries and found that the latter oral-specific instruments discriminated the clinical groups more efficiently [
8].
It should be noted that the two study groups considered were not age and sex matched, nor were they comparable with respect to their other socio-demographic characteristics (Table
1). The age and sex distributions of the school children with and without dental caries, periodontal problems, and malocclusions also differed, and might therefore have confounded the associations between the normative treatment needs or clinical indicators and the prevalence of oral impacts. Most of the confounding effects were probably accounted for when the site-specific multivariable analysis was adjusted for age, sex, and other socio-demographic factors. A comparison of the sample characteristics of the Dar es Salaam participants with the corresponding child population statistic on markers of gender and parental education suggested that this sample was representative of the populations of children aged 12-14 years in the two districts investigated. No similar analysis of the school children in Arusha was performed. Although both samples were randomized cluster samples, the possibility of selection bias cannot be overlooked. The structured self- and interviewer-administered questionnaires used in this study had certain limitations, with bias attributed to social desirability, acquiescence, and lack of recall frequently encountered, particularly in the younger age groups [
39]. Attempts were made to minimize these biases by informing the participants at both sites that their responses were confidential and that no-one could link their names to their responses. The estimates pertaining to the school children in Dar es Salaam might have been underestimated because social desirability bias is more pronounced with interviews than with self-administered questionnaires. Because the Child-OIDP was used as an interviewer-administered measure in Dar es Salaam, whereas the inventory was self-administered in Arusha, the comparability of the data collected across sites could be questioned [
12,
31,
32]. Nevertheless, previous studies of children from the general population and from specific disease groups have supported the comparability of the two modes of administration of the Child-OIDP inventory [
6,
14].
Acknowledgements
The work in Arusha was partly funded by a grant from the Norwegian Cooperation Programme for Development, Research and Education (NUFU), and partly by the Faculty of Medicine and Dentistry, University of Bergen. It was facilitated by the collaborating institutions: Muhimbili University of Health and Allied Sciences and the Centre for Educational Development in Health, Arusha, Tanzania, and the Universities of Oslo and Bergen, Norway. The authors acknowledge and thank the Arusha municipality, Arusha rural and Meru administrative council authorities, Muhimbili University of Health and Allied Sciences, the Ministries of Health and Social Welfare and Education of Tanzania, and REK Vest of Norway for their permission to conduct the study. The authors are indebted to the study participants, their parents, and their school administrations for making this study a reality. We thank Mrs Flora Mrita for her diligent assistance during the clinical field work.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
HSM: principal investigator, designed the study, collected the data (Arusha study site), performed the statistical analyses, and wrote the manuscript. MT: investigated, designed, and collected the data at the Dar es Salaam site. JRM: participated in the design of the study and provided valuable guidance in the data collection at both sites, and has been actively involved in writing the manuscript. PD: supervised, designed, and provided guidance for the study at the Dar es Salaam site. ANÅ: main supervisor, designed the study, and guided the statistical analyses. All authors have read and approved the final manuscript.