INTRODUCTION

Oral health–related quality of life (OHRQoL) is a multidimensional construct that corresponds to the effect of oral health and disease on an individual's daily functioning, well-being, and overall quality of life.^1,2 Conditions affecting oral health such as dental caries, periodontitis,^3,4 and malocclusion are highly prevalent, and they have consequences not only for physical and economic well-being, but they can impair quality of life by affecting function, appearance, interpersonal relationships, socializing, self-esteem, and psychological well-being.^5,6

In 1988, Locker outlined a model of OHRQoL based on the World Health Organization's conceptualization of the impact of disease (Figure 1). Since then, much of the research, including that on malocclusion, orthodontic treatment, and OHRQoL has been based implicitly on this model.⁷

View Figure

• Download as Powerpoint
• Download Figure

This model states that there are five consequences of oral disease: impairment, functional limitation, pain/discomfort, disability, and handicapping. Further, the model proposes that these domains are sequentially related as shown in Figure 1. According to this model, impairment (structural abnormality, eg, malocclusion) leads to functional limitation (restriction of body functions, eg, difficulty chewing) and pain/discomfort (self-reported physical and psychological symptoms), which, in turn, leads to disability (limitations in performing daily activities such as ingesting an adequate diet). Disability may then lead to handicapping (social disadvantage such as isolation).⁸ Impairment and functional limitation may also lead directly to handicapping.

Locker's model has typically been viewed as a framework rather than as a scientific model to be empirically validated. There have been few studies attempting to test the model pathways explicitly.^8,9 These have included the oral health of the general adult population, the burden of being edentulous, and the dental caries experience.^8–11 However, to date it has not been determined whether, for malocclusion patients, the constructs in Locker's model relate to one another as hypothesized.

There are several reasons why the model may differ in this patient group. First, malocclusion and orthodontic treatment differ from most oral diseases, conditions, and interventions, in that malocclusion is a set of dental deviations rather than a disease, and orthodontic treatment does not cure a condition, but rather corrects variations from an arbitrary norm.¹² Further, it has been suggested that the majority of oral health measures developed in dentistry are not applicable to orthodontic patients because most malocclusions are asymptomatic and are related to esthetics, as opposed to the features of other “diseases.”^12,13 In addition, evidence suggests that social and psychological effects are the key motives for seeking orthodontic treatment rather than function limitation, pain, or discomfort.^14–16 However, few studies suggest other motivating factors such as temporomandibular disorders, headache, and reasons related to self-esteem and self-confidence.¹⁷ Therefore, we hypothesized that the pathways in Locker's conceptual model are not applicable to malocclusion patients in the manner suggested for other oral disorders and that, as a consequence, it is important to consider the different domains that can be affected and their relationships to malocclusion. In addition, the model sheds light on the effects of malocclusion on people's lives and provides more understanding of the reason for treatment demand.

Additionally, we treated the pain and discomfort domain separately in our model testing as it is vital to discriminate between these two in malocclusion patients. Developing knowledge of key pathways will help facilitate the design of intervention strategies by, for example, guiding clinicians as to where and how to intervene most effectively in patients with malocclusion.¹⁰ Finally, ongoing development of the OHRQoL, as in any field of inquiry, requires key concepts to the explored and disentangled. Only by testing the empirical validity of a model can alternatives be proposed to address any identified weaknesses.⁸

Therefore, the primary aim of the present research was to provide an empirical test of Locker's conceptual model of oral health for malocclusion patients using structural equation modeling (SEM) and to determine the OHRQoL domains affected by malocclusion and their relationship.

MATERIALS AND METHODS

Participants in the study were aged 15 to 25 years and were attending orthodontic clinics at the Faculty of Dentistry, Universiti Teknologi MARA, in Malaysia. The patients were recruited at their first visit for orthodontic screening before starting any treatment. The parents or participants signed an informed consent form, agreeing to participate in the study. To be eligible, participants had to be in good general health, that is, no known systemic disease, eg, diabetes. Participants requiring surgical intervention or who had chronic medical conditions, previous orthodontic treatment, severe dentofacial anomalies, untreated dental caries, or poor periodontal health status as indicated by a community periodontal index score of 3 or more were excluded. The Universiti Teknologi MARA Research Ethics Board approved all study procedures.

OHRQoL was measured using a Malay language version of the 14-item Oral Health Impact Profile (OHIP-14). The Malay version of this gauge has been found to be valid and reliable, and it has been used in a nationally representative survey to obtain population estimates for prevalence, extent, and severity.¹⁸ Summary scores and scores for each domain of the OHIP-14 were calculated by summing the ordinal values for the items. All participants completed the OHIP-14 before any orthodontic treatment was done. The internal consistency of the subscales of the profile was calculated using Cronbach's alpha, the values of which varied from 0.70 to 0.90 for OHIP-14 subscales, demonstrating a good level of internal consistency for orthodontic patients. It was not deemed feasible to assess test-retest reliability, as this would assume that there was no change in oral health status between tests. As orthodontic treatment covers a period of months or years, there was a high likelihood of change in oral health status.

The Dental Health Component (DHC) of the Index of Orthodontic Treatment Need (IOTN) was used to assess malocclusion. Treatment needs of the patients were categorized as grade 1 (no treatment need), grade 2 (little treatment need), grade 3 (borderline need), and grades 4 and 5 (high treatment need) (Table 1). The IOTN-DHC ratings were recorded by two trained and calibrated examiners. To assess intra- and interexaminer reliability, 20 subjects, not part of the study, were randomly selected and reexamined at 2- to 4-week intervals after their first examination. Intraexaminer reliability for the IOTN-DHC examiners was almost perfect, with kappa values for both the examiners, 0.91 and 0.96. Excellent agreement was also found for interexaminer reliability, with a kappa of 0.85.

Table 1. Characteristics of the Participants (n  =  323)

View Fullscreen

SEM with path analysis was used to analyze the data and to estimate the fit of Locker's conceptual model for the malocclusion data. This method evaluates the relative importance of different paths between the variables and also estimates the direct effects (paths directly from one variable to another, eg, functional limitations −> disability) and indirect effects (paths mediated through other variables, eg, functional limitations −> handicapping via disability). The strength of direct and indirect effects for different paths was measured by β-coefficients. Three models (a basic model, a modified model, and the final model) were developed to test different paths from malocclusion to handicapping. In the basic model, all the paths from malocclusion to handicapping as hypothesized by Locker's model were considered and analyzed. Figure 1 is the path diagram showing all paths hypothesized by Locker and used in the basic model.⁷ In the modified model, all the paths suggested by the modification index were relaxed. The final model was developed by removing all the nonsignificant paths from the modified model. AMOS 20.0¹⁹ was used for SEM and path analysis; it estimates model parameters with maximum likelihood estimation and bootstrapping. AMOS also estimates the total effects, which are made up of both direct and indirect effects. The bootstrap framework has been advocated as the best approach to testing direct and indirect effects in mediation models.^8,10,11 Following Shrout and Bolger's²⁰ techniques, 1000 bootstrap samples were created to derive less biased standard errors and 95% confidence interval (CI) bootstrap percentiles. These biased-corrected, or bootstrapped standard error and CIs, have been shown to be more accurate for calculating indirect effects.⁸ We also assessed whether mediation was present by testing the significance of the indirect effect using the bias-corrected bootstrap CIs. To assess the fit of the models, several indexes as given in Table 2 were generated and compared.

Table 2. Fit Indexes for the All Structural Equation Models^a

View Fullscreen

RESULTS

Table 1 summarizes the participants' characteristics (n  =  323). We developed three models to establish the best pathways in Locker's conceptual model for malocclusion patients: model 1— basic model, model 2—modified model, and model 3—final model.

In the basic model, only the pathways hypothesized by Locker in his conceptual model (Figure 1) were considered and analyzed. Following Locker's hypothesis, it was predicted that malocclusion is related to the patient's functional limitation, pain, discomfort, and handicapping. Functional limitation is related to disability and handicapping. Pain and discomfort only predict disability. Disability, in turn, is associated with handicapping. Fit indexes for this model indicated that it did not fit the data well (model 1, Table 2). Therefore, our next step was to determine whether a modified model incorporating additional paths between other levels would better fit the data.

The basic model was modified in the light of the modification index (ie, the expected change or decrease of the chi-square value, when relaxing a specific path) and theoretical considerations. In addition to the magnitude of the chi-square change, the interpretability of relaxing a specific path was also considered to improve model fit. Three additional direct paths were suggested by the modification index: functional limitation −> pain, functional limitation −> discomfort and pain −> discomfort. Therefore, the modified model (model 2) included all variables and paths suggested by Locker and those based on the modification index to provide a simultaneous test of all possible direct and mediation effects. The modified model fitted the data well compared with model 1 (Table 2). Model 2 accounted for 18% of the variance in functional limitation; the variance accounted for in this model's pain variable was on average 76%: 71% for discomfort, 71% for disability, and 82% for handicapping.

Table 3 shows the estimates for all direct paths in model 2. Four of the direct paths hypothesized by Locker were not significant in this model. The remaining five paths as hypothesized by Locker were significant and in the expected direction. Moreover, the additional paths that were added to the basic model on the basis of the modification index were significant and in the expected direction. In model 2, the presence of malocclusion was associated with functional limitation but not significantly associated with pain, discomfort, or handicapping. Higher functional limitation was directly associated with more pain. It was also significantly associated with nonadjacent levels of discomfort, disability, and handicapping. Pain predicts discomfort but does not directly predict disability. The direction of the coefficients indicates that a higher discomfort level was positively associated with a higher level of disability and, finally, disability predicted handicapping.¹⁹

Table 3. Hypothesis Testing of Direct Effects Between Adjacent and Nonadjacent Levels for the Full Structural Equation Model (Model 2) in Malocclusion Patients

View Fullscreen

A comparison of the direct path with the total effect (Table 3) indicates that the impact of functional limitation on nonadjacent level disability was mainly direct, accounting for 63% of the variance of the total effect. For the path malocclusion to handicapping, it was not possible to calculate effect proportions due to suppression,²⁰ which occurs when the direct and indirect effects have opposite signs, resulting in negative effect proportions or values >1.0.

Table 4 shows the estimates for the total indirect effects, with bootstrapped SE and bias-corrected 95% CI. As shown, all the indirect pathways were significant, showing that all paths have either full or partial mediation, which were checked. Full mediation is defined as occurring when a nonadjacent path has a nonsignificant direct effect and a significant indirect effect. If a path had both a significant direct effect and a significant indirect effect, it was categorized as partial mediation.

Table 4. Total Indirect Effects Between Nonadjacent Levels for the Full Structural Equation Model (Model 2) in Malocclusion Patients

View Fullscreen

In order to create a more parsimonious model, all nonsignificant direct paths were removed from model 2. In this model, the direct effects of malocclusion on pain, discomfort, and handicapping were nonsignificant, as was the direct effect of pain on disability. Therefore, these were dropped from model 3 (the final model). This model was then compared with model 2. There was no difference in the model fit indexes, suggesting that the dropped pathways were not important to the model, and that model 3 had good fit (Table 2). The squared multiple correlations for this final model were also similar to those of model 2. The standardized estimates, bootstrapped SE, and bias-corrected 95% CI and R² for this final model can be seen in Figure 2.

View Figure

• Download as Powerpoint
• Download Figure

DISCUSSION

The findings from this analysis suggest that a modified version of Locker's conceptual model of oral health is a better representation of the oral health of individuals with malocclusion. The final model shares some common paths with Locker's original model, ie, malocclusion predicts functional limitation, functional limitation predicts disability and handicapping, discomfort predicts disability, and disability predicts handicapping. However, the final model includes two additional paths: First, functional limitation directly predicts pain and discomfort, and second, pain directly predicts discomfort. The presence of malocclusion predicts functional limitation; there is no direct influence of malocclusion on pain, discomfort, disability, or handicapping. Functional limitation was associated directly with pain and discomfort; these pathways were both direct and indirect. In the final model, there were some direct paths that were deleted from Locker's model. The direct path from the impairing (in this case) malocclusion to handicapping was suppressed in the final model, which would support a view that handicapping and disability ultimately derive from the experience of other dimensions of impact. Removal of this path was consistent with the model proposed by Nuttall et al. (2006).⁹ Putative direct paths from malocclusion to pain and discomfort were also suppressed in the final model, suggesting that the pain and discomfort found in individuals with malocclusion is due to the functional limitation arising from the malocclusion—not from the malocclusion per se. However, it was not possible to ascertain from these results whether these suppression effects represent inconsistent mediation or are the result of sampling fluctuations (“spurious,” or empirical, suppression).²⁰ Shrout and Bolger²⁰ noted that spurious suppression would be expected in 50% of instances due to sampling variations.

The model proposed here for the impact of malocclusion differs from previous models based on modeling data for general population samples. Baker et al.¹⁰ modeled data from a general population sample with a full range of oral health and ill health. These authors argue that Locker's conceptual model provides a generic oral health model that is applicable across a broad range of settings. Compared with the model proposed here for individuals with malocclusion, two pathways exist in the Baker et al. model that were not found in the current model—a direct path between pain and handicapping and a direct path between discomfort and handicapping. Nuttall et al.,⁹ while not adopting a path analysis approach, suggested a model that is in many ways similar to that presented here for malocclusion. However, these investigators also identified a direct path between pain/discomfort and handicapping.⁹ As discussed above, it is possible that malocclusion exerts its effects through the pathway of disability rather than directly.

The complex relationship between malocclusion and domains of OHRQoL has a number of important implications. First, modeling indirect and mediated effects can help explain why relationships between malocclusion and disability and handicapping found in our study and previous research are often weak. Here, functional limitation was a key mediator in the malocclusion-to-disability (or -handicapping) pathway; as indicated, this path was fully mediated through functional limitation. Second, the findings reemphasise the importance of assessing subjective functional limitations alongside traditional clinical variables, both in research and clinical practice.^11,15 The data suggest that a full understanding of the effect of malocclusion cannot be captured by clinical assessment alone; it should be complemented by assessing functional limitation. Such data confirm that malocclusion can influence an individual's wider well-being by directly impacting everyday physical functioning, and indirectly, psychological and social functioning.¹¹

The findings of this study must be tempered by a consideration of its limitations. First, the participants were 15–25-year-old individuals seeking orthodontic treatment; however, a high prevalence of malocclusion was observed in different age groups, including an older age group, in most populations.²¹ Participants may have exaggerated the impact in order to access treatment. Additionally, the results from this study cannot be extrapolated to the entire youth population with varying levels of malocclusion and orthodontic treatment needs that might have different effects on their daily activities.¹⁵ Although various studies have assessed the impact of oral conditions on children's quality of life with convenience samples in hospitals or universities,¹² future studies should be based on a representative sample with or without normative need. Additionally, in future studies, the applicability of this modified model should be empirically tested for different segments of the population, eg, across gender, age, and socioeconomic status.

The OHIP-14 and IOTN were used in this study, both of which are valid and reliable, but they have limitations.²² The IOTN is a relatively insensitive instrument to measure minor occlusal traits and irregularities that mostly affect the patient's appearance or quality of life, about which a patient may be concerned.²¹ The OHIP-14 was developed for adults, but it has been successfully applied to adolescents by many authors²² because adolescents of 12 years of age and above are capable of abstract thinking, reasoning about the timing of past events, and relating them with good or bad experiences.²² Another limitation of the OHIP-14 is that it does not elicit the specific cause(s) of the effects recorded, which can be related to a variety of oral health conditions and not necessarily the subject's malocclusion. However, the participants in this study were selected to be free of untreated caries, periodontal disease, and other oral health issues, suggesting that the results from OHIP-14 were not confounded with other oral health conditions.

CONCLUSION

• The pathways identified in Locker's (1988) conceptual model of oral health may not be appropriate for describing the relationships between OHRQoL constructs in individuals with malocclusion. An alternative model is proposed.