| Sign In to gain access to subscriptions and/or personal tools. |
The Association between Generalized Joint Hypermobility and Temporomandibular Joint Disorders: A Systematic Review
1 Dept. of Oral and Maxillofacial Surgery, Correspondence: *corresponding author, University Hospital Groningen, PO Box 30.001, 9700 RB, Groningen, The Netherlands; p.u.dijkstra{at}rev.azg.nl
To analyze conflicting evidence in the literature for the association between temporomandibular joint disorders (TMD) and generalized joint hypermobility (GJH), we performed a bibliographic search. The methodological quality of the 14 papers found was assessed according to 14 criteria. Papers were included in the analysis if the study population was clinically relevant, if range of motion of 2 or more joints was assessed on the left and right sides, and if cases had a TMD. Four studies fulfilled these selection criteria. Data from 3 studies, 113 cases and 95 controls, were available for analysis. Twenty-six cases and five controls were hypermobile (odds ratio, 5.4). In a sensitivity analysis, the odds ratio changed from significant to non-significant in 2 of 5 scenarios. As a result, it is not clear whether GJH is associated with TMD, and more rigorous studies are needed.
Key Words: temporomandibular joint disorder • generalized joint hypermobility • systematic review
The term "joint hypermobility" is generally used to indicate an excessively increased range of motion of a joint. If an individual has several joints with an excessive range of motion, he or she is referred to as having "generalized joint hypermobility" (GJH). GJH may be the result of a collagen defect, such as Ehlers-Danlos syndrome. However, GJH also occurs without a known underlying collagen defect, and in these cases is referred to as "benign GJH" (Beighton et al., 1989). The first scoring system for GJH was introduced by Carter and Wilkinson (1964) and has been modified several times (Beighton and Horan, 1969; Grahame and Jenkins, 1972; Beighton et al., 1973). In these modified scoring systems, the ranges of motion are usually visually assessed. Instrumental assessment of ranges of motion, to quantify hypomobility and/or hypermobility of one or several joints, has also been advocated (Kirk et al., 1967; Jobbins et al., 1979; Dubs et al., 1984; Fairbank et al., 1984; Haskard and Silman, 1985; Dijkstra et al., 1994). The assessment method and criteria according to Beighton are used in most studies (Beighton et al., 1973). In this method, the following ranges of motion are assessed: (1) passive hyperextension of fifth metacarpophalangeal joint, 1 point for each joint beyond 90°; (2) thumb apposition to the volar aspect of the forearm, 1 point for each thumb touching the arm, active hyperextension of the elbow, 1 point for each joint beyond 10°; (3) active hyperextension of the knee, 1 point for each joint beyond 10°; (4) trunk flexion with knees straight, 1 point when hand palm flat on floor. The maximum score is 9 points. It is generally accepted that an individual is considered "hypermobile" when at least 4 maneuvers exceed the predetermined range. The reported prevalence of benign GJH is dependent on the study population. In Caucasians, the number of subjects with benign GJH reduces rapidly during the first 10 years of life (Wynne-Davies, 1970; Dubs et al., 1984). Asians show a higher prevalence of benign GJH as compared with English Caucasians (Wordsworth et al., 1987). Additionally, benign GJH is more prevalent in Iraqi students than in Caucasians (Al-Rawi et al., 1985; Bird and Calguneri, 1986). With regard to gender, females show a higher prevalence of benign GJH than males (Dubs et al., 1984; Al-Rawi et al., 1985; Larsson et al., 1987; Wordsworth et al., 1987). GJH may be associated with a variety of complaints of the locomotor system, such as myalgia, arthralgia, traumatic synovitis, dislocation of joints, and soft tissue lesions (Kirk et al., 1967). An explanation for these associations might be that the locomotor system is more prone to mechanical overload, due to the excessive range of motion of the joints. An alternative explanation might be that the quality of collagen is poor in patients with GJH, resulting in early degeneration and associated clinical problems. Visceral complaints associated with GJH, rectal and uterine prolapse, support this latter hypothesis (Al-Rawi and Al-Rawi, 1982; Marshman et al., 1987). A relationship between temporomandibular joint disorders, such as internal derangement and osteoarthritis, and GJH had already been suggested more than a century ago and has been frequently supported since then (Annandale, 1887; Speck and Zarb, 1976; Dolwick et al., 1983; Solberg, 1986; Tanaka, 1986). The supposed sequence of events in this association is that the temporomandibular joint (TMJ) is involved in GJH, indicating that the TMJ is among the hypermobile joints. Due to this hypermobility, the joint is overloaded, resulting in degenerative changes which may become manifest as internal derangements and/or inflammation (Dijkstra, 1993). Several studies have been performed to analyze the association between TMJ disorders and GJH (Bates RE et al., 1984; Harinstein et al., 1988; Plunkett and West, 1988; Westling, 1989; Chun and Koskinen, 1990; Blasberg et al., 1991; Buckingham et al., 1991; Dijkstra et al., 1992; Westling and Mattiasson, 1992; Adair and Hecht, 1993; Khan and Pedlar, 1996; Conti et al., 2000). The results of these studies are conflicting: Some studies yielded an association between TMJ disorders and GJH, while other studies could not demonstrate an association. However, the studies differ with respect to recruitment of subjects, inclusion and exclusion criteria, the number of subjects included, and the number of joints assessed for the GJH, which may have contributed to the conflicting results. The aim of this systematic review was to analyze the available studies addressing the association between TMJ disorders and GJH with respect to methodology and outcome, and, if possible, to pool the data for risk estimation.
Paper Identification and Selection A bibliographic search was performed in the databases of Silverplatter, Medline, Psychlit, Cinahl, and Embase. For this search, the medical subject headings "joint instability" or "joint hypermobility" or "joint laxity" were combined with "temporomandibular joint" or "temporomandibular disorders" or "temporomandibular joint disorders" or "temporomandibular joint dysfunction syndrome" or "craniomandibular disorders" or "temporomandibular joint disk". No language restrictions were used. In this way, 78 records were found. Moreover, an on-line search was performed for congress abstracts. We manually searched the reference lists of the extracted papers to look for additional articles and abstracts. Included in this review were research papers analyzing the relationship between temporomandibular joint disorders and GJH. With respect to the research design, cohort studies and case-control studies were included, while case reports and case series were excluded. Papers describing a population without a control group were excluded from this review, because no risk estimation can be performed without a control group. Abstracts or congress papers were excluded from the review when the corresponding article was published. If a subsequent article could not be found, the abstract was included in the review. A total of 12 papers fulfilled these criteria.
Assessment Prior to the assessment, we extensively discussed all criteria to reach consensus about their content. In addition, all papers were blinded with respect to the title, the names of the authors, and the journal, to minimize observer bias. Nevertheless, it is clear that some papers could still be identified by their format, which may have introduced some observer bias. Cohen's Kappa was calculated as a measure of agreement between the observers. In case the observers disagreed on a specific score, consensus was reached subsequently by means of discussion. Two additional papers were found after the independent assessment was performed (Harkins et al., 1995; Winocur et al., 2000). The same observers assessed these papers. Because the observers had already discussed the assessment results of the other papers, the assessment of the additional papers could not be regarded as being independent, and, hence, Kappa was not calculated. Following the assessment by the observers, the corresponding author of each article was contacted, by letter or by e-mail, and requested to indicate whether or not he/she agreed with the assessment score. If no answer was received within 3 weeks, it was assumed that the author agreed with the assessment as indicated in the letter/e-mail. In case of disagreement, the author was invited to indicate why he/she disagreed with the assessment. Only if the corresponding author could convince the observers was the assessment score adjusted accordingly. Despite extensive efforts to reach the corresponding authors, two of them could not be traced. Studies were selected for further review and analysis if each of the following criteria was fulfilled:
Of the selected studies, the corresponding authors were requested to provide a copy of their original data on which the analysis and the results of their study were based. The following variables were entered into a database: author, age and gender of the subjects, the status of subjects (case or control), number of joints presenting hypermobility, and the Beighton-sum score.
Statistical Analysis
In Table 1  , characteristics of cases and controls are summarized, the assessment of peripheral joints is specified, and the overall results of the different studies are presented. The results of the assessment according to the criteria are specified in Table 2.
The agreement between the observers expressed as Cohen's Kappa was 0.60 for all criteria. The mean Kappa for the 3 major selection criteria (7, 8, and 11) was 0.83. Overall, 5 studies scored more than half of the maximally obtainable assessment points (i.e., > 7). Two authors contacted us about the assessment score of their papers, but no changes were made in the observers' assessment scores. Four studies fulfilled the major selection criteria. The original data from one of these were not available (20 cases and 20 controls), and the data from the 3 remaining studies were entered into a database.
In total, the records of 208 subjects (113 cases and 95 controls) were available for analysis. The mean ages were 28.6 yrs (SD 11.1) for the male subjects (n = 45) and 29.2 yrs (SD 8.6) for females (n = 163). A hypermobility score (Beighton score) of
The results of the logistic regression analysis are summarized in Table 3
A sensitivity analysis was performed by the addition of 20 cases and 20 controls, corresponding to the fourth study for which data were not available, to the database, so that the influence of these data on the stability of the odds ratio could be assessed (Blasberg et al., 1991). The following scenarios were analyzed and the corresponding odds ratios (OR) calculated:
To assess the influence of the methodological quality on the results, we performed an additional scenario. Of the selected studies, the one with the lowest methodological score was excluded (Perrini et al., 1997), and a 2 x 2 table was constructed. The odds ratio could not be calculated, because one of the quadrants was empty. However, a Fisher-exact test yielded no significance (p = 0.102).
The overall methodological quality of studies included in this review varied considerably (median score, 5.5; lower and upper quartiles, 4.5 and 9). Only 4 studies fulfilled the predetermined selection criteria, addressing clinical relevance (criterion 7), assessment of 2 or more joints on left and right sides (criterion 8), and requiring cases to have a temporomandibular joint disorder (criterion 11). Remarkably, the methodological quality of the abstract by Blasberg et al. (1991) was good compared with other studies (Table 2  ). The majority of criteria was adequately addressed in a few sentences. The data from only 3 studies could be entered into the analysis, because the data for one study were not available. This study was taken into account by means of a sensitivity analysis. From the analysis of the pooled data, the presence of generalized joint hypermobility appears to be associated with the presence of a temporomandibular joint disorder (OR = 5.4; 95% CI limits of 2.0 and 14.6, respectively). The results of the sensitivity analysis indicate that including or excluding studies from the meta-analysis may have considerable consequences for the results. Consequently, this review cannot be conclusive as to whether a relationship between TMJ disorders and GJH exists.
Pre-defined selection criteria were used because we were primarily interested in the clinical relevance of an association between TMJ disorders and GJH. If an association exists but cases do not seek help (i.e., treatment or advice) for TMJ disorders, there is no clinical relevance in the association. In several of the studies reviewed, different types of non-clinical cases (dental students, hypermobile children, an epidemiological sample of 17-year-old subjects, and children who were treated orthodontically) were included (Table 1 For the "generalized" character of GJH to be sufficiently addressed, the assessment of the joints was required to include at least 2 joints on left and right sides. In some studies, however, only 1 or 2 joints were assessed: the dominant elbow and thumb, or dominant thumb or wrist. Another basic requirement for this review was the involvement of the TMJ. Only 4 papers fulfilled our pre-determined criteria for TMJ involvement. From the other 10 papers, it could not be clearly deduced whether the TMJ was involved or whether the muscles of mastication were painful or causing the restriction. Furthermore, it is generally agreed that merely a clicking temporomandibular joint does not necessarily indicate that there is a clinically relevant TMJ disorder. With respect to selection bias, which is a major threat to case-control studies, only criterion #2 was addressed by more than 50% of the extracted papers. Other criteria addressing possible selection bias scored poorly in most studies. The range of joint motion was actually measured in only 7 studies. In the other studies, the range of motion was visually assessed, which may have introduced information bias of the exposure. In addition, in only 4 studies were the observers blinded, thus controlling observer bias. Although we used only 3 major criteria to select the papers for further review, only 4 papers fulfilled these criteria. Thus, in summary, the overall methodological quality of the studies published was low. To analyze the stability of the odds ratio, we performed a sensitivity analysis. In Scenario III (added cases are not hypermobile and added controls are hypermobile), the odds ratio passed the neutral value. However, this scenario is unlikely on the basis of the results of Blasberg et al. (1991), who reported to have found no significant differences in hypermobility scores between cases and controls, although no further details about the numbers of hypermobile cases and controls were reported in the abstract. Thus, on the basis of these analyses, GJH might be a risk factor for TMJ disorders, because the odds ratio passed the neutral value in only the most unlikely scenario (III). To assess a possible influence of the methodological quality of the studies, we excluded the data from the study by Perrini et al. (1997), methodological score 5, and re-analyzed the association between GJH and TMJ disorders. The association between GJH and TMJ disorders was no longer significant (Fisher exact test, p = 0.102). Thus, the methodological quality of the studies may influence the association between GJH and TMJ disorders, indicating that the results of this meta-analysis are unstable. Not until the exact relationship between TMJ disorders and GJH is established on the basis of methodologically sound research will it be clear whether clinical measures can and should be taken to treat and prevent TMJ disorders attributed to GJH. Based on the results of this study, it can be concluded that it is not clear whether GJH is clinically associated with TMJ disorders. This lack of conclusiveness may be caused by the poor methodological quality of the available studies. Therefore, further research of sound methodological quality is needed to analyze the clinical relationship between GJH and TMJ disorders.
We thank all authors of the reviewed papers for their cooperation with this study and for providing us with their original data.
A supplemental appendix to this article is published electronically only at http://www.dentalresearch.org. Received for publication July 2, 2001. Revision received January 22, 2002. Accepted for publication January 22, 2002.
Journal of Dental Research, Vol. 81, No. 3,
158-163 (2002) This article has been cited by other articles:
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
|
 |
|
Sign In to gain access to subscriptions and/or personal tools.
TOP
ABSTRACT
INTRODUCTION
4 was present in 26 cases and 5 controls, resulting in a crude odds ratio of 5.4 (95% confidence interval: 2.0 to 14.6). 
