Introduction

Muscle pain has been known for some time to be the leading cause of discomfort in the head and neck area [5], and masticatory muscle pain (myofascial pain) is the most common diagnosis among the various conditions encompassed by the term temporomandibular disorders (TMDs) [58, 81]. For about 40 years [76], stabilization splints have been one of the preferred modalities in the management of TMDs [72], although a great variety of other treatments are currently in use among clinicians [3, 96].

In the early 1990s, David Sackett and his colleagues from McMaster University, Hamilton (Ontario, Canada), introduced the concept of evidence-based medicine (EBM), “the conscientious, explicit and judicious use of current best evidence in making decisions about the care of individual patients” [82]. The practice of EBM relies on three equally important elements: the best research evidence from a systematic search of the literature (mostly from patient-centered clinical research), individual clinical expertise (clinical skills and past experience), and patient values (preferences, concerns, expectations) [83]. The application of EBM comprises several steps: formulation of an important and answerable clinical question derived from a clinical problem; selection of the most appropriate information source(s) and the most appropriate search strategy; and appraisal and application of the evidence found. In many editorials and articles, the introduction of EBM into dentistry has been encouraged. However, the incorporation of EBM into clinical decision-making depends heavily on the availability of external clinical evidence.

Over the past two decades, a few reviews about the effect of occlusal appliances for the management of TMDs have been published (e.g., [2, 10, 11, 30, 31, 52]). In these publications, however, no distinction among specific TMD diagnoses such as masticatory muscle pain was made. The present paper, therefore, focuses on the most common clinical scenario in the management of TMDs by dentists: the management of a patient, who suffers from pain in the masticatory muscles, with a bite splint. Specifically, by systematically searching the literature, we want to provide an answer to the following two clinical questions. In patients with masticatory muscle pain:

  1. 1.

    Does the use of a full-coverage hard acrylic occlusal appliance (stabilization splint, Michigan splint) lead to a significant decrease of symptoms?

  2. 2.

    Is symptom improvement achieved with a stabilization splint more pronounced than the success attained with other forms of management (including placebo treatment) or no treatment?

Considering the great number of patients suffering from myofascial face pain, an answer to our questions would be of considerable clinical relevance.

Methods

To identify all relevant articles and dental congress abstracts of randomized controlled clinical trials (RCTs), in which stabilization splint therapy was used and compared to no therapy or concurrent treatments, the following information sources and search strategies were used:

Search in electronic databases

  • Ovid Medline (online database). The subject terms included in the search as well as the search strategy are listed in Table 1.

    Table 1 Search strategy and results in Ovid Medline (date of the search: 11 December 2003)
    Full size table

  • Cochrane Library (online database). The search term was “splint.”

  • ISI Web of Science (Science Citation Index Expanded) (online database). The option “Full Search” was chosen. The keywords included in the search were “bite splint,” “occlusal appliance,” “occlusal splint” and “splint.”

  • Japana Centra Revuo Medicina (CD-ROM). The search in this database was carried out in a similar way as in Medline.

    The following Japanese terms were used:

    • Randomized controlled trials: Musakui taisyo shiken, Musakui hikaku shiken

    • Clinical trials: Rinsyo shiken

    • Random allocation: Randamu waritsuke

    • Double-blind method: Nijuu mouken shiken

    • Controlled clinical trials: Hikaku rinsho shiken, Hikaku taisho rinsho shiken

    • Myofascial pain dysfunction (MPD) syndrome: Kinmaku totsuu kinoushogai shoukougun, MPD shoukougun

    • Temporomandibular joint (TMJ) dysfunction syndrome: Gaku-kansetsusho, Gaku-kinouijo, Gaku-kinoushogai, Tougai kagakushogai

    • Masticatory muscle: Soshaku-kin

    • Myofascial pain syndrome: Kinmaku-totsuu shoukougun, Kinmakutsuu shoukougun

    • Pain: Totsuu

    • Bite splint: Baito supurinto, Baito purein, Kougou kyojo

    • Occlusal appliance: Okuruuzaru apuraiansu

    • Occlusal splint: Okuruuzaru supurinto

    • Splint: Supurinto

Handsearch in selected journals

The most important, peer-reviewed journals of Austria, France, and Germany, which are currently not included in Medline, namely Actualités odonto-stomatologiques (Medline listing discontinued in 1991), Stomatologie (formerly—until 1996—Zeitschrift für Stomatologie; discontinued in Medline in 1990), and Deutsche Zahnärztliche Zeitschrift (listing in Medline discontinued in 1992), were handsearched through December 2003.

In addition, the abstracts published in the special issues of the Journal of Dental Research, which relate to the annual General Session and Exhibition of the International Association for Dental Reseach (IADR), were reviewed for the years 1990–2003.

Bibliography search of the identified publications and reviews

The references listed in the relevant articles were perused to identify additional publications pertinent to our clinical question. In addition, the reference lists of relevant review articles were checked.

Inclusion/exclusion criteria

Only trials in which patients had explicitly been diagnosed with masticatory muscle (myofascial) pain were considered. Studies in which additional diagnoses were allowed (e.g., TMJ arthralgia/osteoarthritis; disk interference disorders), were excluded from further analysis. Similarly, articles in which unspecific terminology was used to characterize the investigated patient samples (e.g., “mandibular dysfunction”, “[TMJ] pain dysfunction syndrome” or “temporomandibular disorders”) were not considered because these terms may also include TMJ-related conditions.

There were neither age restrictions of study participants nor language restrictions for inclusion. The last update of the search was made on 11 December 2003.

Assessment of reporting quality

The reporting quality of the identified articles of RCTs was assessed independently by two reviewers (J.C.T, A.H.). For this purpose, the quality score developed by Jadad et al. [46] was used. The Jadad scale consists of five items which focus on three dimensions of internal validity (randomization; double blinding; description of withdrawals and drop-outs). Uncertainties on data interpretation and discrepancies in scoring the reporting quality were resolved by discussion between the two reviewers.

Results

Altogether, 13 relevant publications, representing nine clinical studies, were identified by the two reviewers. The articles of six studies were published in English, the rest in Dutch [98], German [86], and Japanese [84]. Two studies were carried out in Sweden, two in the USA, one in Canada, one in Germany, one in Italy, one in Japan, and one in the Netherlands. Only two trials [91, 98] were conducted over an observation period of at least 1 year.

The results of the search in Ovid Medline are displayed in Table 1. Seven articles listed in this database are relevant to our question [7, 17, 28, 47, 80, 86, 98].

The search in the Japanese database yielded one hit [84]. The searches in the Cochrane Library and in ISI Web of Science identified no additional publications.

The handsearch in the special issues of the Journal of Dental Research yielded five meeting abstracts [42, 43, 44, 90, 91]. All five abstracts referred to the same prospective trial. In the following, only the two most recently published abstracts, which complement each other, are considered [44, 91].

The bibliography search of the identified publications identified no additional publication. (In one article [78], a paper from an Argentinean dental journal was cited [79] in which it was allegedly reported “that splint therapy associated with diazepam, in this order, produced more effective TMD pain relief when these therapies were applied exclusively.” In the cited article by Roldan et al. [79], however, bite splints were not mentioned.)

The major characteristics (patients, inclusion and exclusion criteria, treatment groups, number of participants, study duration, outcome variables for treatment success) and findings (improvement of signs and symptoms, authors’ conclusions, reviewers’ comments) of the publications are summarized in Tables 2 and 3. In all but one study [86], stabilization splints were fabricated in the maxilla. Masticatory muscle pain was an outcome variable in every study. In all but two trials [7, 17], functional parameters such as mandibular mobility and TMJ sounds were also considered. In three articles [28, 47, 86], Helkimo’s Clinical Dysfunction Index (CDI) [38] was used.

Table 2 Characteristics of RCTs in which stabilization appliance therapy in patients with masticatory muscle pain was investigated
Full size table
Table 3 Characteristics of studies in which stabilization appliance therapy in patients with masticatory muscle pain was investigated
Full size table

The methodology of the clinical investigations differed in several important aspects:

  1. 1.

    Recruitment of study participants: in all but two trials, study participants were restricted to patients seeking care at or referred to clinical centers specialized for the diagnosis and management of orofacial pain. In the remaining two studies, participants were exclusively [80] or partly [17] recruited by notices in local newspapers or journals.

  2. 2.

    Number of participants: the number of participants included in most studies was small. The total number of participants lay between 26 [86] and 168 individuals [91] (Table 2).

  3. 3.

    Description of randomization/blinding: the method of randomization was described only in two publications [28, 86]. Five studies [17, 28, 47, 80, 98] used a blinded design: the examiner who evaluated the treatment was blind to the type of treatment the patient received. Lack of (double-)blinding leads to an overestimation of the treatment effect [85]. (It is controversial whether or not double-blinding is possible in trials comparing active and “placebo” appliances.) The publication by Ekberg et al. [28] is the only one with an appropriate description of both randomization and blinding, and with a Jadad score of 5 (Table 4). (It has to be considered that the reported information about the randomized studies of Huggins/Truelove et al. [44, 91] as well as Sakuma et al. [84] was limited in the available meeting abstracts.)

    Table 4 Assessment of the quality of the studies by using the quality score proposed by Jadad et al. [46]
    Full size table
  4. 4.

    Appliance use: in three trials [17, 80, 86], the appliance was worn (nearly) 24 h/day. In four other studies, the splint was worn only at night [7, 28, 47, 98]. In the published meeting abstracts [44, 84, 91], no pertinent information was given.

  5. 5.

    Treatment provided in the control group(s): Michigan splints were compared with the following alternative approaches:

    • Non-occluding palatal appliance [17, 28, 80, 84]

    • Anteriorly occluding maxillary splint [86]

    • Full-covering maxillary soft appliance [44, 91]

    • Occlusal adjustment [98]

    • Physical therapy [98]

    • Body acupuncture [47]

    • Different treatments [44, 91]

    • No treatment [7, 47]

In one study [80], physical therapy and verbal support were given to each patient included in the trial.

Based on the results described in the identified publications, our two clinical questions can be answered as follows:

  1. 1.

    Management of myofascial face pain with a stabilization splint worn at night is likely to lead to a statistically significant short-term improvement when compared with no treatment [47].

  2. 2.

    Current evidence is inconclusive about the question of whether the observed improvement during and after stabilization therapy is greater than the one achieved by a non-occluding palatal appliance (i.e., a “placebo” splint ). In two recently conducted trials [28, 84], there was a statistically more significant decrease of pain and functional impairment in the group that received a stabilization splint. In two other studies of somewhat lower reporting quality and validity as compared to the trial by Ekberg et al. [28] (see Discussion), no statistically significant difference could be found between the two types of appliances [17, 80].

  3. 3.

    A stabilization appliance does not appear to yield a better clinical outcome than a soft splint [44, 91].

  4. 4.

    There is some evidence from one study with a small number of patients that with a (mandibular) stabilization appliance a statistically significant greater symptom improvement can be achieved than with an anteriorly occluding maxillary splint (“relaxing appliance”) [86].

  5. 5.

    Evidence is missing that treatment with a stabilization appliance leads to a statistically significant greater improvement of signs and symptoms than body acupuncture [47], physical therapy [98] and occlusal adjustment [98].

Discussion

The present investigation has focused on the effect of hard acrylic stabilization splints on pain located in the masticatory muscles. Our review is the first one that has looked specifically into this question. The available data indicate—at different levels of scientific evidence—that a hard acrylic stabilization splint does not yield a better clinical outcome than a non-occluding palatal appliance, a soft splint, or conservative treatment without a splint such as physical therapy and acupuncture. In the only controlled clinical study in which stabilization splints were compared with a partial-coverage anterior splint—a Sved appliance [88]—, there was a trend for greater symptom improvement in the stabilization splint group. This finding should be interpreted with caution, however, because the sample size was small [86]. In one study [98], occlusal adjustment was carried out. In a recently published systematic review, Koh and Robinson [51] came to the conclusion that there is an absence of evidence that this invasive procedure is an effective therapeutic measure. Therefore, the systematic selective adjustment of the occlusal surface of teeth is not recommended [31, 51, 93].

Within the hierarchy of scientific evidence, systematic reviews are considered to have the highest quality level. These types of publication are carried out to answer one or more focused clinical questions about a topic related to health care [50]. In systematic reviews, it is not always possible to quantitatively combine the data from the identified studies (meta-analysis) [24]. Among other reasons, this may be due to the lack of reported original data or to methodological differences applied in the trials. In that case, a qualitative rather than a quantitative systematic review is carried out [49].

Kalso et al. [49] have noted that the strength of evidence lies in the quality of controlled trials: “Systematic reviews can only be as convincing as the quality of the controlled trials allows.” The 13 pertinent publications to answer our clinical question represent nine (mostly short-term) RCTs of varying reporting quality and validity. We have included the meeting abstracts of the RCT by Huggins/Truelove et al. [44, 91] and Sakuma et al. [84], although these reports have not yet been substantiated by more detailed articles in peer-reviewed journals. Of the 13 publications, three [17, 47, 80] were considered in the qualitative systematic reviews on occlusal splints published by Forssell et al. in 1999 [31], Kreiner et al. in 2001 [52], and Al-Ani et al. in 2004 [2], respectively.

We had to exclude a great number of articles. The reasons for exclusion were:

  1. 1.

    Study participants were not randomized (e.g., [8, 33, 36, 74]). Randomization is the most important procedure to avoid selection bias [48]. In general, non-randomized studies overestimate treatment effects [9].

  2. 2.

    Patients with masticatory muscle pain and TMJ pain were included in a trial / no clear distinction was made between muscle pain and TMJ pain (e.g., [8, 15, 16, 18, 20, 32, 35, 40, 41, 55, 56, 57, 60, 70, 73, 74, 77, 92, 94, 101]). (Nonetheless, inclusion of the results gained in these studies would not have altered the conclusions of the present review.)

  3. 3.

    The effect of stabilization splints was evaluated in patients with TMJ pain (e.g., [25, 26, 27]) or TMJ disk displacement (e.g., [54, 59]).

  4. 4.

    The effect of splint therapy on the accuracy of mandibular movements of patients with masticatory muscle pain was studied (e.g., [64]).

  5. 5.

    The effect of soft occlusal splints on masticatory muscle pain was investigated [102].

  6. 6.

    Multiple simultaneous treatments were carried out (e.g., [14, 37, 39, 69, 78]).

  7. 7.

    No control group was used (e.g., [4, 8, 36, 100]). Without controls, it cannot be excluded that unspecific effects, e.g., spontaneous remission, natural course of the symptoms, regression to the mean, the placebo effect, the Hawthorne effect, biased (favorable) patient’s answers, and accompanying therapeutic measures [95], may have been responsible for observed differences among groups.

When appraising articles on clinical studies in a qualitative systematic review, vote counting (determination of the number of articles showing that an intervention works or does not work) should be avoided [49]. Instead, more weight should be given to publications of (a) high reporting quality (i.e., Jadad score 3 to 5) and (b) high validity [49].

In our review, reporting quality of RCTs was assessed with the help of the Jadad score (range: 0–5). Among the many checklists, scales, and indexes that have been suggested for the evaluation of the (reporting) quality of randomized trials, the Jadad score was developed using standard scale development techniques [46]. Since it is the only known validated scale [6], it has been widely used by clinical researchers. Nonetheless, some critical voices have been raised lately alluding to the fact that the Jadad scale gives more weight to the quality of reporting than to actual methodological quality [49]. In addition, recent reports have pointed out that the inter-rater reliability of the Jadad score may be low [6, 12]. We have striven to avoid this latter problem by independent assessments of the identified articles and discussion between the two reviewers in case of inconsistencies of the scoring results. As Kalso et al. [49] point out, writing a qualitative systematic review requires at least two authors—it “is not a lonely (wo)man’s affair.”

Six of the nine studies considered had a Jadad scale score of 3 or more. In contrast, three studies had an unacceptably low reporting quality of 1 or 2. It has been shown that studies with a Jadad score of 2 or less tend to give an overoptimistic picture of the real treatment effect than studies with a higher score (e.g., [29]).

There are indications that the reporting quality of a published article does not always correlate with the actual methodological quality of the trial [45]. Deficits in reporting about the results of RCTs have been mentioned by authors in dentistry [65, 87] as well as in medicine [1, 6, 19, 34, 45, 75]. For example, only 25.4% of the articles of RCTs published in the journal Intensive Care Medicine up to the year 2000 had a Jadad score of 3 or more [53]. In our review, there is reason to believe that the methodological quality of the studies by Huggins/Truelove [44, 91] and Sakuma et al. [84], which were awarded 1 point, respectively, is much better than the actual Jadad score suggests.

As far as the assessment of the second quality factor—high validity—is concerned, two important criteria are a sufficient baseline pain intensity [66] and an adequate number of patients in each group [67]. In the identified trials, recruiting of study participants took place by either resorting to patients seeking care at or being referred to an orofacial pain care center, or by placing announcements in local print media [17, 80]. It should be taken into consideration, however, that patients seeking TMD treatment by referrals are probably different from individuals recruited by a notice in a local newspaper or journal [28]: Rubinoff et al. [80] argued in the critical discussion of their study published in 1987 that the latter patients “may have been biased toward milder conditions that were tolerable to the patient until prompted by a media notice.” This assumption appears to be correct. When the baseline pain intensities of the three studies published in peer-reviewed journals in which a stabilization appliance were compared with a non-occluding appliance are analyzed, the following can be observed:

  1. a.

    In the trial by Rubinoff et al. [80], the majority of the study participants had low pain before the start of the study. All patients were recruited by a newspaper notice.

  2. b.

    In the investigation by Dao et al. [17], pre-treatment pain intensities were about 40 mm, which is equivalent to moderate pain [13], on a visual analog scale (VAS). Part of the participants were recruited through announcements published in local journals; the other part were referred by dentists.

  3. c.

    In the study by Ekberg et al. [28], the pain intensity prior to the start of the study was between moderate and very severe. All patients were referred for and requested treatment.

Hence, differences in the recruitment of patients may be one reason for different pre-treatment pain levels and the different results achieved in the three studies. The three trials also differ considerably with regard to the included number of patients (cf. Table 2). The result from this comparison is that two studies with an acceptable reporting quality and a moderate validity [17, 80] found no difference between a stabilization appliance and a non-occluding appliance. These results differ from the recent study by Ekberg et al. [28] with both strong evidence and a strong validity supporting the efficacy of the Michigan splint compared with a palatal appliance.

Only two trials—those conducted by the Huggins/Truelove group [44, 91] and by van der Glas et al. [98], respectively—had an observation period of at least 1 year. In three studies [28, 47, 86], Helkimo’s CDI [38] was used. However, the validity of this index has been shown to be doubtful [99].

Our search has also demonstrated the inconsistencies that exist among different authors with regard to the diagnosis “muscle pain” (Table 2). This variability is reflected in the number of diagnostic systems that have been proposed over the past decades for classifying the different subsets of TMDs (c.f. [71]). It was not before 1992 that biologically plausible classifications with specific diagnostic categories became available [22, 89]. As a result, comparisons among the studies identified in this search are difficult to make, and pooling of data is impossible. As far as the evaluation of trials carried out in the 1970s and 1980s is concerned (c.f. [8, 36, 80]), study participants diagnosed with “myofascial pain dysfunction syndrome” are not necessarily patients suffering from myalgia only. Although masticatory muscle pain is a symptom encountered in most, albeit not all of these patients, symptoms such as limited mandibular opening, TMJ sounds, and deviation on jaw opening may be present. Only one of these older articles allowed an explicit assessment of the symptom “muscular pain” [8].

Furthermore, our investigation questions the strategy of relying exclusively on Medline when looking for evidence. Of the 13 pertinent articles, seven were identified by the Ovid Medline search. Conversely, the meeting abstracts found by handsearching were not listed in any of the consulted electronic databases. Hence, as our search demonstrated, important study results may be missed if one relies on Medline as the sole information source [1, 21, 97]. Besides, limitation of the search to the English language, as it is often seen in reviews, may lead to different conclusions (English language bias) [23, 63].

Another point that can be made is that keywords are likely to change depending on the prevailing thinking in the field. For example, whereas “temporomandibular disorders” is a term that was agreed upon by most clinicians in the early 1990s [62], Medline lists articles about this topic under the medical subject heading “Temporomandibular Joint Dysfunction Syndrome.”

Conclusions

The dearth of adequate clinical studies that are available to answer our clinical question mirrors the lack of hard data in an important and frequent scenario in clinical dentistry. We found this result astonishing because full-coverage heat-cured (or self-curing) acrylic resin occlusal splints have been in use since the 1960s [76], and these devices have been recommended by many clinicians around the world for the management of patients with TMDs, including masticatory muscle pain [68]. Generally, the wealth of anecdotal reports and uncontrolled clinical observations tend to give a much more optimistic picture about the presumed effects of the stabilization splint therapy.

Due to the limited number of available studies, our clinical question can only be answered tentatively: based on the currently best available evidence it appears that most patients with masticatory muscle pain are helped by the incorporation of a stabilization splint. Nevertheless, evidence is equivocal that improvement of pain symptoms after incorporation of an intraoral appliance is caused by a specific effect of the splint [17, 28, 80]. In addition, there is not enough data about the long-term efficacy and effectiveness of these widely used therapeutic tools. It should be noted that a scarcity of prospective randomized controlled trials with high power does not discredit the concept or the applicability of EBM, because EBM is based on the best available evidence. However, by pointing out deficits in the quantity and quality of the evidence, EBM highlights the empirical nature of current management and emphasizes the need for more focused clinical research in dentistry.