| Sign In to gain access to subscriptions and/or personal tools. |
Testing a Conceptual Model of Oral Health: a Structural Equation Modeling ApproachDepartment of Oral Health and Development, School of Clinical Dentistry, University of Sheffield, UK; s.r.baker{at}sheffield.ac.uk
Locker’s (1988) multidimensional model of oral health provides a scientific model for the understanding of oral disease and its consequences. To date, there have been no studies that have explicitly tested the model with empirical evidence. This study aimed to: first, test the model in a general population sample using data from the UK adult dental health survey (N = 5268); and, second, to cross-validate these results in two different and diverse samples—edentulous elders (N = 133) and a clinical sample of xerostomia patients (N = 85). Structural equation modeling indicated support for the model as applied to each of the samples. All of the direct pathways hypothesized by the model were significant, in addition to several indirect or mediated pathways between key variables. Further conceptual development of the model is discussed, particularly the role of individual difference factors, and theoretical and methodological issues in oral-health-related quality-of-life research are highlighted. Abbreviations: oral health quality of life (OHQoL); structural equation modeling (SEM); Adult Dental Health Survey (ADH survey); Oral health Impact Profile-short form (OHIP14).
Key Words: oral-health-related quality of life • structural equation modeling • Locker • population study • xerostomia • edentulous
Oral-health-related quality of life (OHQoL) is a multidimensional construct that refers to the extent to which oral disorders disrupt an individual’s normal functioning. OHQoL has become an important focus for assessing the impact of a range of oral conditions on an individual’s quality of life and well-being (e.g., Gift et al., 1997), together with the outcomes of clinical care, such as the effectiveness of treatment interventions (e.g., Awad et al., 2000; Allen et al., 2001). To date, much OHQoL research has been based implicitly on Locker’s (1988) conceptual model of oral health. This model states that there are five consequences of oral disease—impairment, functional limitation, pain/discomfort, disability, and handicap—and that these are sequentially related. Impairment (structural abnormality, e.g., edentulousness) leads to functional limitation (restrictions in body functions, e.g., difficulty chewing) and pain/discomfort (self-reports of physical and psychological symptoms), which, in turn, lead to disability (limitations in performing daily activities, e.g., unsatisfactory diet) and then to handicap (social disadvantage, e.g., social isolation). Functional limitation may also lead directly to handicap. Locker’s multidimensional model provides a scientific model for the understanding of oral disease and its consequences. Yet, there have been few studies that have attempted to test the model explicitly with empirical evidence. As such, the model has been seen more as a framework rather than as a scientific model to be empirically validated. Nagi (1991) suggested that any conceptual framework needs to meet certain criteria: Does the framework "fit the facts?", and what does it do to advance scientific development and guide further action? In relation to the first, there are numerous key questions that remain unanswered. Are functional limitations, discomfort/pain, disability, and handicap distinguishable? If so, are they related to one another in the way hypothesized within Locker’s model? There are several reasons why providing answers to such questions is important. First, the validity of OHQoL as an outcome measure in clinical trials is partly dependent on an understanding of the causal processes linking oral conditions to patient-reported outcomes. Before the pathways underlying such effects can be understood, any proposed model needs to be valid and empirically tested. Second, 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. Finally, ongoing development of the OHQoL field, as in any field of inquiry, requires key concepts to be explored and disentangled. Only by testing of the empirical validity of a model can alternatives be proposed to address any identified weaknesses. The primary aim of the present research was to provide an empirical test of Locker’s (1988) conceptual model of oral health. The model was tested in a general adult population with data from the UK Adult Dental Health Survey (Kelly et al., 2000), by means of structural equation modeling (SEM). SEM is a powerful statistical technique that allows for the simultaneous testing of complex interrelationships between variables specified within a priori models (Kline, 2005). As such, it is currently the best technique for assessing and modifying theoretical models. A second aim was to test, by cross-validation analysis, whether the model generalized to different samples. The aim was to examine whether separate conceptual models might be necessary for specific patient/population groups. Two additional samples were tested: edentulous elders and a clinical group of xerostomia patients.
The research questions were examined with data collected as part of three studies.
Samples
(2) Edentulous Elders
(3) Xerostomia Patients
Measure
Statistical Modeling Procedure Structural equation modeling (SEM) with maximum likelihood estimation was used to estimate the models with AMOS 6.0 (Arbuckle, 2005). The main a priori hypotheses were that functional limitations would predict disability, which, in turn, would predict handicap. Functional limitations would also predict handicap. Both pain and discomfort would predict disability, and pain would be associated with discomfort. AMOS estimates the total effects, which are made up of both the direct effects (a path direct from one variable to another, e.g., functional limitations  disability) and indirect effects (a path mediated through other variables, e.g., functional limitations handicap via disability). We assessed whether mediation was present by testing the significance of the indirect effect using the bias-corrected bootstrap confidence intervals. The bootstrap framework has been advocated as the best approach for SEM with small-moderate sample sizes (Samples 2 & 3), and for testing mediation models (MacKinnon et al., 2002; Shrout and Bolger, 2002).
The adequacy of model fit was assessed by the chi-square test statistic, together with the Goodness of Fit Index (GFI), Normed Fit Index (NFI), Comparative Fit Index (CFI), and the root-mean-squared error of approximation (RMSEA) with 90% confidence intervals (90%CI); a non-significant chi-square—values > than 0.95 for GFI, CFI, and NFI—and a RMSEA
General Adult Population The model was not an acceptable fit on any of the indices: GFI, NFI, and CFI indices were 0.88–0.92, the RMSEA was 0.24 (90% CI = 0.23–0.25), and the chi-square was significant (  2 [4] = 1202.26, p < 0.001). Inspection of the modification indices indicated that if functional limitation was allowed to predict both discomfort and pain, and discomfort predicted handicap, this may improve the fit of the model. When the model was re-run with these modifications, the chi-square indicated a significantly improved fit. The data supported the Locker model on 5 out of 5 of the a priori model-fitting criteria [2 (1) = 0.156, p = 0.212, GFI = 1.000, NFI = 1.000, CFI = 1.000, RMSEA = .010 (90% CIs = 0.00–0.04)]. This model accounted for 27% of the variance in discomfort, 12% in pain, 62% in disability, and 43% in handicap. The standardized parameter estimates for this model can be seen in the Fig .
Direct Effects All of the direct paths hypothesized within Locker’s model were significant and in the expected direction (Table 2 ). There were also three additional paths; more functional limitation predicted greater discomfort and pain, and more discomfort predicted higher handicap scores. The proportion of the total effects accounted for by the direct paths varied greatly; e.g., the impact of functional limitation on discomfort was primarily direct (70%), compared with that between functional limitation and disability (34%) (Table 2).
Indirect Effects There were numerous significant indirect effects between non-adjacent levels in the model (Table 3 ). These total indirect effects represent all possible paths. Specific indirect effects have to be calculated by multiplication of the estimates of the direct effects involved in the total pathways. In summary, these were: (i) functional limitation pain discomfort (β = 0.12), (ii) functional limitation pain disability (β = 11) and functional limitation discomfort disability (β = 0.14), (iii) functional limitation disability handicap (β = 0.08), (iv) pain discomfort disability (β = 0.18), (v) pain disability handicap (β = 0.15), and (vi) discomfort disability handicap (β = 0.26).
Cross-validation Analysis In line with the general adult population sample, the Locker model fit the data on five out of five of the a priori criteria for both the edentulous elders sample [ 2 (1) = 0.873, p = 0.350, GFI = 0.997, CFI = 1.000, NFI = 0.997, RMSEA = 0.000 (90% CIs = 0.00–0.22)] and xerostomia patients [2 (1) = 0.914, p = 0.339, GFI = 0.996, CFI = 1.000, NFI = 0.996, RMSEA = 0.000 (90% CIs = 0.00–0.28)]. For the edentulous and xerostomia samples, respectively, the model accounted for 20% and 35% of the variance in pain, 32% and 29% in discomfort, 69% and 70% in disability, and 34% and 59% in handicap. As in the population sample, all of the direct pathways hypothesized within Locker’s model were significant, the only difference being that pathways between functional limitation-handicap and discomfort-handicap were indirect (compared with direct in the population sample). The standardized estimates for the two cross-validation samples can be seen in Appendix Figs. 1 and 2.
Findings reported here support Locker’s (1988) conceptual model of oral health as applied to three different and diverse samples. These results indicate that Locker’s model may be considered to represent a generic ’oral health’ model, which is applicable at individual, group, and population levels. This is in contrast to recent research suggesting that it may be useful only for understanding patient outcomes in populations (Nuttall et al., 2006). Nevertheless, to confirm the findings, investigators should test the relative fit of Locker’s model against other models. One model is that by Wilson and Cleary (1995), which has been used extensively in health-related quality-of-life research outside of dentistry, and has recently been applied to oral health (Baker et al., 2007). The statistical modeling techniques used here highlight the importance of testing complex interrelationships, including both direct and indirect (mediated) effects. Only by testing such mediation models will it be possible to gain a greater understanding of the complexity of the causal processes underpinning oral health impacts. For example, functional limitations had an impact on handicap, but this effect was primarily indirect via disability. Translating this result into intervention strategies aimed at minimizing handicap resulting from oral disease would suggest targeting patient’s levels of physical, psychological, and social disability (e.g., unsatisfactory diet, social embarrassment). The original Locker (1988) model did not include individual and environmental factors, although it was noted that they were likely to play an integral role in oral health. Here, the models explained 12–35% of the variance in pain, 27–32% of that in discomfort, 62–70% in disability, and 34–59% in handicap. It could be that the inclusion of key contextual factors may have increased the model’s explanatory power. A host of factors has been identified in the psychology literature, upon which dentistry could draw. These include coping strategies (Lazarus and Folkman, 1984), social support (Cohen and Wills, 1985), sense of coherence (Schnyder et al., 1999), and negative affectivity (Kolk et al., 2003). There is extensive evidence that negative affectivity (predisposition to experience chronic negative emotions, for example) influences symptom perception, as well as physical health reports generally (Pennebaker, 1982; Baker, 2007) and, more specifically, OHQoL (Kressin et al., 2001). Similarly, sense of coherence (a measure of the salutogenic resources available to an individual) has been shown to act as a mediator between disability and handicap (Schnyder et al., 1999).
Further research is needed to identify key contextual factors, and examine their exact role in oral health pathways. There has been much terminological, conceptual, and statistical confusion within the general literature on the role of such factors (Holmbeck, 1997). This is because contextual factors may influence the relationships between variables (e.g., functional limitation In conclusion, the present study is the first to provide support for Locker’s multidimensional model as a scientific framework for the understanding of patient outcomes in different and diverse samples. The findings have several theoretical and clinical implications. Published two decades ago, Locker’s model represented a fundamental shift in dentistry from a paradigm emphasizing disease and a medical model of care to one that incorporated a patient-centered perspective. Since then, OHQoL research has burgeoned and has become accepted as a valid outcome in clinical practice and research. However, most OHQoL research has been hindered by the lack of a systematic application of a theoretical framework. Theoretically driven research, such as that presented here, is important to facilitate a greater understanding of the dynamics of an individual’s experiences of his/her oral health and, in turn, how oral health influences longer-term systemic health and well-being. Understanding such processes will facilitate the conceptual development of OHQoL, and help translate the clinical relevance of this research, most notably by facilitating the design of treatment strategies aimed at improving patient outcomes.
Thanks to Peter G. Robinson, Caroline Pankhurst, and Nicola Pearson for allowing use of their data for the cross-validation analysis. The community-based randomized control trial of a domiciliary denture service was supported by a grant from the Community Fund. The randomized control trial of an intervention device for the management of xerostomia was supported by a grant from Action Research. Thanks go to David Locker and Barry Gibson for the many useful discussions relating to points raised in this paper. The UK Adult Dental Health Survey was carried out by the Office for National Statistics, Social Survey Division, Northern Ireland Statistics and Research Agency, Central Survey Unit. The sponsors of the research were the Department of Health, National Assembly for Wales, Scottish Executive, and Department of Health and Social Security (Northern Ireland), and the survey was conducted in collaboration with the Dental Schools of the Universities of Birmingham, Dundee, Newcastle, and Wales. The data are distributed by the UK Data Archive, University of Essex, Colchester. Neither the original data creators, depositors or copyright-holders, the funders of the Data Collections, nor the UK Data Archive bear any responsibility for the analysis or interpretation of the results reported herein.
A supplemental appendix to this article is published electronically only at http://www.dentalresearch.org. Received for publication November 16, 2006. Revision received March 28, 2007. Accepted for publication April 17, 2007.
Journal of Dental Research, Vol. 86, No. 8,
708-712 (2007)
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
|
 |
|
Sign In to gain access to subscriptions and/or personal tools.
TOP
ABSTRACT
INTRODUCTION
0.06 indicate that the model is a good fit (