The findings above present the baseline psychosocial data from a large cohort of TMD-free controls compared to a smaller group of individuals meeting research diagnostic criteria for chronic TMD (i.e., cases). Significant case-control differences emerged across multiple psychosocial content domains, such that TMD cases reported higher levels of psychological and affective distress, greater perceived stress and catastrophizing, and increased somatic awareness compared to controls. The case-control differences of greatest magnitude emerged for measures of somatic awareness (i.e., PILL, SCL-90R Somatization), with standardized odds ratios exceeding 2.0. In general, case-control differences for other psychosocial variables were of more modest magnitude. These findings are consistent with previous results of case-control studies examining psychosocial variables among individuals with TMD8–10,13,29,47,48,50,51,61,67
, as well as in studies of people with other types of chronic pain.20,31,35
Less well-documented are the differences in pain coping that were observed in this sample. Specifically, consistent with recent evidence8,67
TMD cases reported higher levels of catastrophizing, a form of maladaptive cognitive appraisal that has been associated with increased pain and disability across multiple chronic pain conditions.22,79
Surprisingly, controls reported greater use of praying and hoping, which is typically construed as a passive coping strategy, as supported by its loading with catastrophizing in our PCA. It should be noted that the magnitude of this difference was quite small, and the practical significance of this finding requires further investigation.
From a methodological perspective, it is notable that cases and controls in the present study were recruited from the same communities using similar recruitment approaches. This avoids the sampling biases that can occur when cases are recruited from clinical settings and controls are recruited from non-clinical sources. Such biases tend to inflate case-control differences in psychological functioning, as many of these psychosocial variables are associated with health care seeking in chronic pain populations.36,85
This may also explain the modest magnitude of many of our case-control differences in psychological functioning, as compared to what might be expected when contrasting community-based controls to clinic cases.
Within controls, the association of demographic factors with psychosocial variables was examined independently and without adjustment for other factors. In general, age-related differences were small in magnitude and failed to follow a consistent pattern, perhaps owing to the relatively restricted age range in our cohort. However, gender differences emerged for multiple variables, with women reporting higher levels of psychological distress across several SCL90-R scales, along with greater neuroticism, perceived stress, negative affect, and somatic awareness. Women also reported greater use of catastrophizing, distraction and praying & hoping, while men reported higher scores on coping self statements, and ignoring and reinterpreting pain sensations. These gender differences are consistent with previously published results,3,4,23,76
and the increased expression of these potential psychosocial risk factors may contribute to the greater prevalence of TMD among women.42
Ethnic group differences were also observed for several psychosocial measures, such that non-whites reported higher psychological distress across all examined SCL90R subscales. Consistent with previous findings32
, ethnic differences in pain coping emerged, with significantly greater use of catastrophizing, praying & hoping, distraction, and reinterpreting pain sensations among non-whites. Regarding somatic awareness, ethnic differences emerged for both the PILL and the Kohn, similar to previous results.68
That psychosocial factors differed significantly between cases and controls raises the possibility that such psychosocial variables may represent predisposing risk factors for the development of chronic pain. Of course, our present cross-sectional analysis cannot determine the direction of the association; however, a limited literature has addressed whether premorbid psychosocial factors are associated with subsequent development of TMD. Members of this research group have reported that premorbid measures of depression, perceived stress, and mood state obtained at baseline were significant predictors of new onset TMD in a sample of healthy young females.75
Also, Aggarwal and colleagues1
found that preexisting health anxiety predicted future development of chronic orofacial pain. Similar findings supporting psychosocial variables as premorbid risk factors for development of chronic pain have been reported for other chronic pain conditions, including chronic widespread pain 53
, regional musculoskeletal pain,57
and low back pain.43,44,57
While we are not able to address this issue in the present analysis of our baseline data, the design of the OPPERA study will permit such analyses at the conclusion of the follow-up period. For example, Aggarwal and colleagues reported that four psychosocial measures distinguished cases from controls in univariate analyses; however, only health anxiety remained a significant predictor in the multivariate approach.1
Therefore, our future analyses to identify risk factors for new onset TMD will include multivariate approaches that concurrently take into account multiple psychosocial measures, as well as potential predictors from other domains.
Principal Component Analysis was implemented as an approach to operationalizing the psychosocial component of the OPPERA heuristic model17
PCA revealed multiple latent constructs based on the extensive battery of psychosocial measures. Among controls, the four components reflected: (1) stress, anxiety, and neuroticism; (2) global psychological symptoms; (3) passive pain coping; and (4) active pain coping. The model was quite stable, and the underlying constructs may reflect important sub-domains of the “enhanced psychological distress” intermediate phenotype previously proposed in the OPPERA conceptual model.49
Specifically, the PCA model indicates the presence of multiple aspects of psychosocial function, which go beyond the unitary concept of “psychological distress.” Indeed, the first two components seem to suggest that there are two different aspects of “psychological distress,” one comprised primarily of personality characteristics and affective distress (including anxiety and stress), and the other comprised of more global psychological adjustment, including both somatic and psychological symptoms. That state and trait anxiety load on the first component, while SCL90R Anxiety loads on the second component is likely due to method variance, given that all of the other SCL90R scales also loaded on component two. The last two components from the PCA model go beyond “distress” and reflect different aspects of pain coping. In contrast to the psychological symptoms reflected by the first two components, these coping components represent cognitive and behavioral processes that may modulate the experience of pain and associated psychological distress. Thus, the multiple constructs reflected by these components may reflect important psychosocial domains for identifying individuals at risk for new onset TMD pain. Among TMD cases, the underlying latent constructs identified in the present study included: (1) global psychological distress; (2) passive pain coping and reactivity; (3) active pain coping; and (4) psychological resilience. Given the difference in constructs identified within the controls versus the cases, it is tempting to speculate that the underlying structure of psychosocial constructs, as reflected by the component measures used in this study, differs for individuals with chronic pain compared to those without. However, given the instability of the PCA model in TMD cases, any potential case-control differences should be interpreted with caution. Also, the latent variables identified by our PCA should be viewed as exploratory, and further confirmation of these results will be needed. A logical next step would be to determine the replicability of these findings in another sample. Successful replication would increase confidence in the validity of the latent constructs reflected by these components, such that indices representing these underlying constructs could then be derived and used in future analyses designed to identify risk factors for new onset TMD.
The limitations of this study deserve mention. First, relative to our large control population, the sample size for cases is quite modest, which contributed to the instability of the PCA model in cases and limited our ability to compare PCA models for cases versus controls. Second, the analyses herein are restricted to the psychosocial data and do not examine potentially important associations between psychosocial variables and other phenotypic measures, such as those presented in other manuscripts in this volume. However, such analyses are planned for future manuscripts. Third, given the large number of analyses already presented, only univariate demographic comparisons were conducted, and the pattern of findings from psychosocial variables may change when multiple psychosocial measures are simultaneously considered using a multivariate approach. In future analyses, it also will be informative to examine psychosocial measures stratified across multiple demographic variables (e.g., age differences within each gender). Moreover, our analysis of ethnic differences only considered whites versus non-whites, and future investigation of different ethnic subgroups is warranted.
These limitations notwithstanding, the current findings demonstrate case-control differences across multiple domains of psychosocial functioning, with the most robust differences emerging for indices of somatic awareness. Moreover, our large sample of controls provided statistical power for demographic comparisons, which revealed gender and ethnic group differences for multiple psychosocial variables. Similar to recently reported findings from healthy subjects,56
our PCA suggests that these multiple psychosocial measures can be reduced to a smaller subset of latent constructs. Future analyses will examine associations of these psychosocial variables with other important phenotypes, such as quantitative sensory testing data and clinical examination findings. Further, genetic associations with psychosocial variables will be explored in future work. Finally, the ultimate goal of collecting these psychosocial measures is to determine their ability to predict new onset of TMD, and these analyses will be conducted upon completion of the follow-up phase of the OPPERA Study.