One possible resolution to the controversy may lie in the well-recognized heterogeneity of TMDs,^10,11 and even heterogeneity of disorder within the myofascial subtype.¹² In other words, maybe only some subtypes of myofascial face pain respond to oral splints. Several studies have taken this approach. For example, Krogstad and colleagues¹³ compared treatment outcomes in patients with facial pain who had pain secondary to whiplash vs. pain resulting from other causes. Turk and colleagues^14–16 and Rudy and colleagues¹⁷ focused on differentiating treatment responses based on psychological characteristics of the patients.

Our own recent research^18,19 focused on the role of widespread vs. local pain in symptom presentation and long-term course of the disorder. We speculated that myofascial face pain in patients who have comorbid widespread pain indicative of fibromyalgia may respond differently to common dental treatments from patients whose pain is primarily expressed in the masticatory muscles.

Patients with fibromyalgia, a widespread pain syndrome of unknown etiology, have elevated rates of myofascial face pain,^20–22 and patients with myofascial face pain have elevated rates of fibromyalgia.^20,23,24 The significance of this overlap is unclear. Is myofascial face pain without widespread pain the same disorder as myofascial face pain occurring with widespread pain? The differentiation is not merely semantic. It is critical in that it might have implications for treatment response.

Increasingly, published work suggests that the widespread pain of fibromyalgia is due to central nervous system dysfunction.^25–27 In contrast, myofascial face pain is viewed by many dentists as resulting from bruxism or occlusal factors.^28–33 Nevertheless, each of these views has provoked considerable controversy.^34–36 The various presumed mechanisms of action of oral splints⁶ predominantly involve modification of local, rather than central, pathogenic factors. Therefore, we hypothesize that patients with localized myofascial face pain are more likely than similar patients who also have widespread pain to respond to treatment with an oral splint.

Subjects were randomly assigned to treatment with an active or a palatal-only hard acrylic splint.

To date, no published study has compared the efficacy of oral appliances in patients whose myofascial pain is expressed primarily in the facial region with that in patients who display evidence of widespread pain indicative of fibromyalgia. The purpose of this investigation, then, was to address this issue.

	METHODS

TOP ABSTRACT METHODS STATISTICAL ANALYSIS RESULTS DISCUSSION CONCLUSION REFERENCES

 
Subjects. Subjects consisted of female patients attending an orofacial pain treatment service within the Oral Medicine Clinic at the University of Medicine and Dentistry of New Jersey in Newark. In addition to recruiting new referrals to the clinic, we received referrals from dentists in the local community, who responded to a mailing announcing the study’s commencement. All referred women received a comprehensive evaluation, including examination for TMDs.

Myofascial face pain. Participating patients met criteria for the myofascial subtype of TMD according to the research diagnostic criteria, or RDC,¹ in which a facial pain complaint was associated with localized tenderness in response to palpation at three or more of 20 muscle sites. Patients meeting criteria for other TMDs, such as osteoarthritis of the temporomandibular joint, were not automatically excluded, provided that their chief complaint was pain (as opposed to clicking or difficulty opening their mouths). To deem eligible a prospective subject who had additional comorbid TMD conditions, we had to make a clinical judgment that the pain was primarily myofascial.

Only women were enrolled, given evidence that rates of myofascial face pain, especially among those seeking treatment, are strikingly higher among women.^37–39 Subjects were required to be fluent in English, although English did not have to be their first language. For other study purposes, they were required to have at least six maxillary and six mandibular posterior natural teeth that occluded. In addition, they could not currently wear orthodontic braces. Prospective subjects who had worn an oral appliance for treatment of facial pain or bruxism were ineligible, because prior experience with a similar treatment modality could affect expectations for improvement.

Sixty-eight women were enrolled, and 63 completed all phases of the investigation. Of the five who did not complete the study, two had been assigned to each of the two treatment conditions (that is, palatal vs. active splint) and withdrew. We removed one additional subject, assigned to the active splint condition, when she displayed symptoms of a thought disorder. This subject was referred for psychiatric evaluation and treatment.

The mean age of the 63 enrolled women was 33.7 years (± standard deviation, or SD, = 10.9 years). The mean (± SD) number of years of education was 14.4 (± 2.2 years), equivalent to two years of college. Forty-nine (78 percent) of the women identified their race as white. Mean (± SD) pain level in the two weeks before the start of the trial was reported as 4.5 (± 1.8), assessed on a 0 to 10 pain-intensity scale (where 0 = n o pain, 10 = worst possible pain). The mean duration of pain at the start of the trial was five years, with 30 percent of the subjects reporting a duration of one year or less and 19 percent reporting a duration of 10 years or more.

Procedures. After providing written informed consent and undergoing evaluation for eligibility, subjects were enrolled in a six-week treatment protocol in which they were randomly assigned to treatment with an active or a palatal-only hard acrylic splint. The splint was provided free of charge.

Active vs. palatal treatment conditions. Patients assigned to the active treatment condition were fitted with a maxillary, flat-plane, hard acrylic splint that covered the hard palate. Patients assigned to the palatal splint condition were fitted for a splint constructed in a manner similar to that of the flat-plane splint, except that it did not cover the occlusal surfaces or interfere with occlusion in any way. The splint was retained with clasps on maxillary teeth. Since one theory of the mechanism of action of oral splints involves cognitive awareness,⁶ in which the splint alters tongue position and oral volume, thereby increasing patients’ awareness of their oral habits, the palatal splint was designed to only partially cover the palate, leaving space for the tongue to assume its normal position. Both groups of subjects were instructed to wear the splint when sleeping at night.

During the six weeks of the study, five visits to the clinic occurred. Two weeks after the initial assessment for eligibility and after completion of a questionnaire documenting current symptoms, subjects were fitted with the splint to which they had been randomly assigned. We asked subjects to assess the extent to which they expected the appliance to result in symptom improvement. This is designated as a baseline period. In a visit two weeks later (designated as the two-week follow-up), a dentist made any needed appliance adjustment, and self-reported symptoms were recorded. Procedures at the next visit two weeks later (that is, the four-week follow-up) were identical to those at the previous visit. At the final visit (that is, the six-week follow-up), self-reported symptoms were recorded, and the RDC examination was repeated by a dentist who was blind to treatment group assignment. Participants were debriefed and paid $300.

To address ethical concerns related to non-treatment of a painful condition, we instructed subjects in both groups to add the following conservative treatments at each appointment: at baseline, soft diet; at two weeks, moist heat/massage; at four weeks, exercise; at six weeks, over-the-counter nonsteroidal anti-inflammatory medication (for example, aspirin or ibuprofen). Other than these conservative treatments, subjects were allowed to use previously prescribed medications or other treatments and were asked to maintain consistent levels throughout the study. To monitor compliance with these instructions, we noted use of cointerventions at each appointment.

RDC. Three dentists were trained to conduct orofacial examinations according to the RDC.¹ To determine diagnostic agreement and ensure quality control for examination procedures, periodic reliability examinations were conducted throughout the study. Perfect agreement was established at the diagnostic level regarding the presence or absence of a diagnosis of myofascial face pain ( = 1.0 across 27 reliability examinations; correlation of tender-point counts among examiner pairs, r = .89). To keep RDC clinicians blind to the subject’s treatment group, the clinician who saw the subject at baseline through the six-week follow-up for appliance adjustment was not the same clinician as the one who conducted the RDC examinations at baseline and at the six-week follow-up visit. In addition, subjects were instructed not to discuss their treatment or show their splint to the clinician conducting the RDC.

Psychological measures. We used two measures of psychological status. First, average mood during the previous two weeks was rated on a 10-point scale, in which 0 = best possible mood and 10 = worst possible mood. As a standard measure of psychological distress, we used the depression symptom scale from the Symptom Checklist-90, or SCL-90.⁴⁰ For each of the 13 items on this self-report scale, the respondent was asked to indicate how much she was distressed by such symptoms as "crying easily," "blaming yourself for things" and "feelings of guilt." The five-response categories ranged from "not at all" to "extremely." Scores from baseline and the six-week follow-up visit are reported here.

Widespread pain. Evidence of widespread pain was gathered from three sources. For a subject to be scored as having widespread pain, she needed to meet at least one of these three standards:

– self-report of fibromyalgia (that is, widespread muscle aches and pains lasting three or more months);

– reports of moderate or more (that is, quite a bit of or extreme) soreness of muscles on an SCL-90 question that is an item in its somatization sub-scale;

– moderate or severe extracranial pain on palpation by the pain clinician at two or more of the following locations: sternocleidomastoid, splenius capitus, trapezius; all extracranial muscles were palpated with two pounds of pressure.

Given that a comprehensive examination of tender points was not conducted using standardized procedures,⁴¹ we should note that our operational definition of widespread pain is not necessarily synonymous with a diagnosis of fibromyalgia.

Functional outcome. At baseline and the six-week follow-up visit, we asked patients to indicate which, if any, of 12 activities they were prevented or limited from doing as a result of their facial pain problem. The activities ranged from chewing, eating hard foods, eating soft foods, talking and yawning to having one’s usual appearance. Patients also were asked to specify the number of days in the last two weeks that they were kept from their usual activities because of facial pain. Finally, we asked patients to indicate how much the facial pain interfered with their daily activities, in the last two weeks, using a 10-point scale (anchored by 0 = no interference to 10 = unable to carry on any activities).

Other measures. After the splints were inserted at baseline, we asked patients to evaluate the extent to which they thought the treatment they were receiving was likely to help their symptoms. Expectations for improvement were rated on a four-point scale (1 = extremely likely to 4 = extremely unlikely). At all appointments, subjects were asked to indicate which of a variety of other facial pain treatments they had received during the previous two weeks. At these same appointments, they reported on how often they wore the splint during the previous two weeks (that is, every night, more than half the time, less than half the time, or not at all).

	STATISTICAL ANALYSIS

TOP ABSTRACT METHODS STATISTICAL ANALYSIS RESULTS DISCUSSION CONCLUSION REFERENCES

 
We used a statistical software package (Statistical Package for the Social Sciences, version 9.0, SPSS Inc.) to analyze all data. For single-time-point comparisons of demographic or clinical characteristics, independent sample t tests (for two-group analysis) or analysis of variance, or ANOVA (for four-group analysis) was used to test differences on quantitative measures. ² was used to test differences between groups for categorical measures at a single time point. To assess the overall pattern of group differences in symptoms at baseline through the six-week follow-up, we used the SPSS general linear model (repeated-measures) procedure. To assess categorical functional outcomes at the six-week follow-up appointment, while controlling for baseline functional complaints, we used logistic regression. In a test of four groups differing on both splint type and widespread pain status, group status was accounted for by three dummy codes, with the group having local pain and receiving the palatal splint forming the contrast group.

In all analyses using the "least pain in the past two weeks," the natural log of the least pain measure (plus a constant of 1) was used, as the analysis revealed that it was not normally distributed (for example, by the six-week follow-up appointment, 54 percent of the total sample of subjects indicated that their least pain was 0 on a 10-point severity scale). Similarly, pain duration was not normally distributed and was positively skewed. Therefore, where relevant, the log transformation of pain duration was used in the analyses below.

	RESULTS

TOP ABSTRACT METHODS STATISTICAL ANALYSIS RESULTS DISCUSSION CONCLUSION REFERENCES

 
Baseline. Of the 63 subjects completing the study, 32 (51 percent) had been randomly assigned to the active splint condition and 31 (49 percent) had been assigned to the palatal splint condition. To ensure that random assignment had adequately equated the groups, we compared the active vs. palatal splint groups on both demographics (age, income, education) and clinical indicators at baseline (that is, log pain duration, average pain in the past six months, worst pain in the past six months, extent to which pain interfered with daily activities in the past six months, average mood and the score on SCL-90 depression scale). Independent sample t tests revealed that treatment groups neither differed from each other nor exhibited a trend (P < .10) toward differences.

We compared subjects regarding their expectations for improvement, to ensure that any differences between treatment groups could not be attributed to differential expectations. Again, the active and palatal splint groups did not differ from one another (t = .43; P > .10). Both groups had similar expectancy scores (1.89 vs. 1.81 for the active and palatal splint groups, respectively), indicating slightly higher than moderate expectations for improvement.

The groups did not differ on use of cointerventions during the course of the study, with 81 percent of subjects in the active splint group and 84 percent of subjects in the palatal splint group reporting that they used no other treatment during the study. Compliance with treatment also was similar in both groups. By the six-week follow-up appointment, 53 percent of subjects assigned to the active splint group and 52 percent of subjects assigned to the palatal splint group reported that they wore the splint every night, as prescribed. In the active and palatal splint groups, 44 percent and 36 percent of subjects, respectively, wore the splint more than half the time, but less than all the time. Only one subject in the active splint group and three subjects in the palatal splint group reported that they wore the splint less than half the time.

Active vs. palatal splint group symptom change over time. Repeated-measures ANOVA was used to describe the overall pattern in symptom change between baseline and the six-week follow-up evaluation for the three self-reported pain severity measures. For average pain, the effect of time was statistically significant (F_3,59 = 18.52; P < .0001), indicating a general improvement in pain levels. The time-by-group interaction, testing how the active vs. palatal splint groups changed differentially over time, approached significance (F_3,59 = 2 .72;P = .052).

For worst pain, the effect of time was statistically significant (F_3,59 = 10.67; P < .0001). The time-by-group interaction, which tested how the active vs. palatal splint groups changed differentially over time, approached significance (F_3,59 = 2 .44;P = .073). For (log) least pain, the effect of time was statistically significant (F_3,59 = 7 .85;P < .0001). The time-by-group interaction, testing how the active vs. palatal splint groups changed differentially over time, was not significant for least pain (F_3,59 = 89; P > .10). Figures 1 through 3 display group differences at each time point on the three pain measures. As shown, the active vs. palatal splint groups differed at the two-week follow-up evaluation for worst pain reports and at the six-week follow-up evaluation for least pain reports. All other group differences were not significant at specific time points (P > .10).

View larger version (46K): [in this window] [in a new window]   Figure 1. Mean value of average pain ratings across four observation periods for subjects in the active vs. palatal splint groups.

View larger version (46K): [in this window] [in a new window]   Figure 2. Mean value of worst pain ratings across four observation periods for subjects in the active vs. palatal splint groups. At the two-week follow-up, P < .05 for the active vs. palatal splint groups.

View larger version (47K): [in this window] [in a new window]   Figure 3. Mean value of (log) least pain ratings across four observation periods for subjects in the active vs. palatal splint groups. At the six-week follow-up, P < .05 for the active vs. palatal splint groups.

To assess how any changes in physical symptoms might affect psychological status, we compared subjects assigned to the two treatment groups on six-week follow-up reports of average mood and the SCL-90 depression scale. The average mood score at the six-week follow-up did not differ between subjects in the active splint group (mean ± SE score, 3.44 ± 0.39) and subjects in the palatal splint group (mean ± SE score, 4.23 ± 0.43) (t = 1 .35;P > .10). In addition, subjects assigned to the active splint group (mean ± SE, 1.50 ± 0.11) did not differ significantly from subjects in the palatal splint group (mean ± SE, 1.66 ± 0.12) on the SCL-90 depression scale (t = .42; P > .10).

We next used logistic regression analysis to compare the active and palatal splint groups regarding differences in functional outcomes at the six-week follow-up evaluation, after controlling for baseline functional complaints. Thirteen functional outcomes (for example, facial pain prevented or limited the subjects from chewing, drinking, eating hard foods, yawning) were examined. For only one of these measures (that is, yawning) was there a trend for subjects who received the active splint (13 subjects [40.6 percent]) to exhibit improved functional outcome compared with subjects who received the palatal splint (20 subjects [64.5 percent]). For another measure (that is, swallowing: adjusted odds ratio = 0.05; P < .10), there was a nonsignificant trend for subjects who received the active splint (four subjects [12.5 percent]) to exhibit relatively worse functional outcome than subjects who received the palatal splint (one subject [3.2 percent]). For none of the 13 functional outcome measures were differences marked enough to meet traditional standards (P < .05) of statistical significance.

Local vs. widespread pain at baseline. To assess the role of widespread pain in moderating the effect of the splint, we divided the sample into subjects who exhibited evidence of widespread pain according to any of the three operational definitions and subjects who exhibited no evidence of widespread pain. Of the 63 subjects in the study, 30 (47.6 percent) were classified as having widespread pain, while the remaining 33 (52.4 percent) were classified as having local facial pain.

We compared subjects in the local and widespread pain groups on demographic variables (that is, age, income, education) and clinical indicators at baseline (log pain duration, average pain in the past six months, worst pain in the past six months, extent that the pain interfered with daily activities in the past six months, average mood, depression score on the SCL-90 scale and number of tender points according to the RDC). The groups differed in the number of painful muscle sites (that is, intraoral and extraoral RDC sites) on palpation, with subjects in the widespread pain group having more painful facial muscles (mean ± SE, 13.13 ± 0.83) than subjects in the local pain group (mean ± SE, 9.18 ± 0.73) (t = 3 .59;P < .001). At baseline, the groups did not differ on any measure of functional pain interference. The only other measure on which the groups differed significantly (t = 2 .30; P < .05) was duration of pain. We were surprised to find that subjects with local pain reported pain of longer duration (mean ± SE, 6.4 ± 1.0 years) than subjects with widespread pain (mean ± SE, 3.5 ± 0.79 years) (t = 2.46; P < .05 on the log-transformed pain duration measure). The four groups did not differ on expectations for improvement or on use of cointerventions (both P > .10).

Local vs. widespread pain group symptom change over time as function of treatment. We then tested for differences between the groups as a function of both splint type and presence or absence of widespread pain. Four groups were compared: an active splint group with widespread pain (n = 13), an active splint group with local pain (n = 19), a palatal splint group with widespread pain (n = 17) and a palatal splint group with local pain (n = 14).

Repeated-measures ANOVA was used to describe the overall pattern in pain severity change between baseline and the six-week follow-up evaluation for the three self-reported pain severity measures, as a function of both splint type and widespread pain status. For all measures, the effect of time was statistically significant (P < .0001). For average pain, the only time-by-group interaction that approached significance was for the active splint group with local pain (F_3,57 = 2 .06;P = .10). For worst pain, the time-by-group interaction was statistically significant for the active splint group with local pain (F_3,57 = 2 .93;P < .05). For the log of least pain, the time-by-group interaction was not significant for any group (all comparisons P > .10).

Figures 4 through 6 display differences among the four groups at each time point for each of the three pain measures. At the six-week follow-up evaluation, the active splint group with local pain differed from all other groups on both average and worst pain measures. On the least pain measure, it differed at the six-week follow-up from only the palatal splint group with widespread pain.

View larger version (55K): [in this window] [in a new window]   Figure 4. Mean value of average pain ratings across four observation periods for subjects in the active vs. palatal splint groups with local vs. widespread pain. At the two-week follow-up, P < .05 for the palatal splint group with widespread pain vs. the active splint group with local pain. At the six-week follow-up, P < .05 for the active splint group with local pain vs. all other groups.

View larger version (56K): [in this window] [in a new window]   Figure 5. Mean value of worst pain ratings across four observation periods for subjects in the active vs. palatal splint groups with local vs. widespread pain. At the two-week follow-up, P < .05 for the active splint group with local pain vs. the palatal splint group with local pain. At the six-week follow-up, P < .05 for the active splint group with local pain vs. all other groups.

View larger version (56K): [in this window] [in a new window]   Figure 6. Mean value of (log) least pain ratings across four observation periods for subjects in the active vs. palatal splint groups with local vs. widespread pain. At the four-week follow-up, P < .05 for the palatal splint group with widespread pain vs. the active splint group with local pain. At the six-week follow-up, the palatal splint group with widespread pain differed from all other groups: P < .10 vs. the palatal splint group with local pain, P < .01 vs. the active splint group with local pain and P < .10 vs. the active splint group with widespread pain.

To assess how any changes in physical symptoms might affect psychological status, subjects in the four groups were compared on the basis of six-week follow-up reports of average mood and depression on the SCL-90 scale. Post hoc comparisons showed that the average mood scores in the palatal splint group with local pain (mean ± SE score, 4.00 ± 0.65), the palatal splint group with widespread pain (mean ± SE score, 4.41 ± 0.58), the active splint group with local pain (mean ± SE score, 2.47 ± 0.37) and the active splint group with widespread pain (mean ± SE score, 4.85 ± 0.64) differed in the following ways. The active splint group with local pain had significantly lower scores (indicating better mood) than all other groups (P = .05 vs. the palatal group with local pain and P < .01 for both of the widespread pain groups), but the active splint and palatal splint groups with widespread pain did not differ from each other (P > .10).

Oral splints were of modest value for patients with myofascial face pain in the overall sample.

We found similar patterns for the SCL-90 depression scale at the six-week follow-up evaluation. Post hoc comparisons showed that the depression scale scores for the palatal splint group with local pain (mean ± SE score, 1.59 ± 0.20), the palatal splint group with widespread pain (mean ± SE score, 1.71 ± 0.16), the active splint group with local pain (mean ± SE score, 1.29 ± 0.08) and the active splint group with widespread pain (mean ± SE score, 1.80 ± 0.24) differed in the following ways. The active splint group with local pain tended to differ from the palatal splint group with widespread pain (P < .10) and differed significantly from the active splint group with widespread pain (P < .05), but did not differ from the palatal splint group with local pain (P > .10).

The table shows the percentage of subjects in the four groups who reported functional limitations at the six-week follow-up evalution. Subjects in the active splint group with local pain had lower rates of functional problems than subjects in the palatal splint group with local pain (associated here with odds ratios greater than 1.0, signifying a better functional outcome).

	DISCUSSION

TOP ABSTRACT METHODS STATISTICAL ANALYSIS RESULTS DISCUSSION CONCLUSION REFERENCES

 
The initial aim of this study was to explore the overall efficacy of a full-coverage hard acrylic splint compared with that of a palatal splint in the treatment of patients with myofascial face pain. Although trends were consistently toward better improvement among subjects who received the active splint, treatment groups differed significantly after six weeks on only one of the three self-reported pain severity measures. They did not differ on the number of tender points on palpation. The general pattern of differences was modest, with groups typically differing by one point on a 10-point pain severity scale.

We conclude that oral splints were of modest value for patients with myofascial face pain in our overall sample. While limited, the effect does not appear to be due to expectations for improvement or other factors associated with placebo effects, since our design was able to rule out these alternative explanations. These equivocal conclusions about the effectiveness of oral splints are consistent with findings in the literature in which weak effects have been detected and contradictory conclusions about efficacy have been made.²

When subjects were classified into subgroups having widespread pain or only local face pain, a potentially important pattern emerged. Even with limited statistical power engendered by our small sample, the pattern over time indicated that the active splint significantly reduced reports of worst pain for subjects with local pain but not for subjects with widespread pain. The trend was in the same direction for the measure of average pain. Among subjects with only local pain, the mean difference between the active vs. palatal splint groups was approximately 1.5 on a 10-point scale for average or worst pain severity after six weeks of treatment. In contrast, among subjects with widespread pain, the difference between the active and palatal splint groups was virtually zero for both the average and worst pain measure, although there was a nonsignificant trend for these groups to differ on a log scale of least pain.

How do we interpret the examination-based measure (that is, the number of painful masticatory muscles on palpation)? At the six-week follow-up, scores did not differ when we compared subjects receiving the active vs. palatal splint. Nor did scores differ when we subdivided the sample according to widespread pain status. In analyses not detailed above, we found that the number of painful muscles corresponded quite well (r = .44; P < .0001) to self-reported current pain severity at the six-week follow-up. The number of painful muscles on palpation, however, did not correspond to self-reported worst pain (r = .13; P > .10) or average pain (r = .17; P > .10) at the six-week follow-up.

Given these data, we can interpret the clinical meaningfulness of the RDC-based palpation measure from two opposing perspectives. One is that, because it bears relationship to only momentary or current pain, the number of painful muscles on palpation is not a stable measure of outcome and should be considered less clinically significant than other measures. The other perspective is that, as a reflection of current pain, the palpation measure is less subject to retrospective bias compared with measures of worst pain or average pain during the past two weeks. We tend to accept the first view. In earlier research, we examined and discussed the poor correspondence between pain on palpation and self-reported pain.⁴² Moreover, we have documented that myofascial face pain symptoms can vary dramatically over relatively brief periods.⁴³ Altogether, this suggests that pain on palpation may not be as sensitive to clinically meaningful change as more stable reports of average or worst pain within a short retrospective period.

Subjects in this study with widespread pain did not differ from subjects with local pain on baseline symptoms of mood or depression.

In general, our results showed that changes in mood and depressive symptoms did not vary depending on whether subjects had been assigned to the active or palatal splint groups. When the treatment groups were divided into local vs. widespread pain, differences emerged at the six-week follow-up between the local pain group that received the active splint and the other three groups regarding average mood.

It bears noting that, in contrast to our previous research with a different sample of patients with myofascial face pain, some of whom had widespread pain,¹⁹ subjects in this study with widespread pain did not differ from subjects with local pain on baseline symptoms of mood or depression. However, after six weeks of treatment, those classified as having widespread pain reported worse mood and more depressive symptoms than did those with local pain. Given that differences in the current study were seen only after treatment failure, it is reasonable to infer that the psychological symptoms noted in previous cases of widespread pain^19,44–51 are at least partially reactive⁵² to treatment failure.

The results of this study must be considered preliminary and the limitations of the investigation should be acknowledged. Because the study was not explicitly designed to investigate the role of widespread pain in moderating oral splint treatment outcomes for myofascial face pain, our assessment of widespread pain was based on multiple methods of assessment, none of which corresponds to current diagnostic standards. The proportion of subjects classified as having widespread pain in this sample of patients with myofascial face pain (48 percent) is higher than the proportion reported by most investigators, including us when we examined another sample.¹⁹ This suggests that we misclassified several subjects as having widespread pain who did not suffer from it. Future research should use standardized methods, such as the American College of Rheumatology’s diagnostic method for assessing fibromyalgia,⁴¹ which involves palpating 18 specific anatomic points with 4 kilograms of pressure.

Another limitation of this investigation is that assessment of outcome terminated after six weeks of treatment. Whether the effectiveness of a full-coverage splint among patients with localized myofascial face pain persists over longer periods should be explored in future research.

Should the pattern found in these pilot data be borne out in larger-scale investigations that use standardized methods for assessing widespread pain and that examine outcomes over long periods, important implications will surface. First, Lewis Carroll’s statement that "everybody has won, and all must have prizes"⁵³ is appropriate. In other words, people who argue for the effectiveness of oral splints and people who argue against their effectiveness are both right, in that it depends on the widespread pain characteristics of the patients. More than 30 years ago, Gordon Paul, a psychotherapy researcher, said that the most important issue is not whether a particular treatment works, but "what treatment, by whom, is most effective for this individual with that specific problem, and under which set of circumstances?"^54(p111) The presence or absence of widespread pain may help to define the specific circumstances under which oral splints should be prescribed.

Nevertheless, given the limitations of this pilot investigation, it is premature to recommend that dentists both screen all patients with myofascial face pain for widespread pain and prescribe oral splints for those with local pain. We recently recommended the former¹⁹ but await further evidence before recommending the latter.

	CONCLUSION

TOP ABSTRACT METHODS STATISTICAL ANALYSIS RESULTS DISCUSSION CONCLUSION REFERENCES

 
Our findings indicate that oral splints have a modest effect on pain severity for a heterogeneous group of patients with myofascial face pain. However, patients with myofascial face pain who have widespread pain are unlikely to receive any benefit from oral splints. In contrast, patients with myofascial face pain with only local pain are more likely to experience some pain reduction when treated with oral splints. Future research must determine whether our preliminary conclusions apply to a larger sample of patients who are assessed more comprehensively for both widespread pain and long-term treatment outcomes. Moreover, treatments that remediate the facial pain of patients with widespread pain need to be developed and tested, according to the strongest possible experimental research methods.