In what follows, we shall summarize the results of clinical trials employing oral appliances for the treatment of sleep apnea and snoring. In addition, we shall review and summarize the information regarding changes in daytime function as a result of using the appliance, clinical and dental side effects, and compliance with treatment.
Oral appliances for the treatment of sleep apnea
Table summarizes the results of individual studies using oral appliances starting with the 1982 polysomnographic study of TRD by Cartwright and Samelson [11
] until the present. The only criterion for the inclusion of a particular study into this table, and thus the only common feature of all studies listed, was the availability of at least partial results of nocturnal monitoring of respiration with and without oral appliance. Otherwise, the studies are highly variable in their design, methodology, data analysis, outcome definition and assessment and presentation of results. This makes the interpretation of individual results, and particularly any attempt to summarize all of them, very challenging. Before describing the methods of analysis and presenting the summary of the data, it is very important to keep in mind the following points.
Clinical studies with oral appliances: 1982–2005
First, Table contains the studies from the two extreme ends of the spectrum of scientific rigor. At the highest end of the spectrum there are prospective, randomized, crossover, controlled (either against placebo-appliance or another treatment modality) studies. At the lowest end of the spectrum are individual case reports. Some studies are prospective case series, but most studies are retrospective analysis of series of cases.
Second, the investigations listed in Table form a very inhomogeneous group with respect to several variables: time of follow-up study, type of polysomnography, which respiratory variables measured, presentation of results, type of oral appliance used, missing data, etc. The time interval between the diagnostic and “with appliance” polysomnography varied from a few hours to a few months. In some studies, split polysomnography (i.e. diagnostic part followed by “with appliance” part) was carried out, while in others, the two sleep studies were separated by a couple of days to several months. In some investigations, formal in-hospital polysomnography was performed, while in others, only at-home monitoring of oxygen saturation was carried out. In some of the earliest investigations, only the apnea index (AI) was measured. Later investigations reported the oxygen desaturation index (number of times per hour of sleep that oxygen saturation falls by more the 4% from the baseline—ODI4). Recent investigations all reported either the apnea–hypopnea index (AHI) or the respiratory disturbance index (RDI). Most investigations presented only the mean values, rather than the individual data, and some presented only the median values. Oral appliances used by the investigators included tongue retaining devices, soft palate lifters, and fixed and variable mandibular advancement appliances. In the vast majority of investigations, the mandibular advancement appliance was used, less than ten investigations employed the TRD, and only one investigation used the soft palate lifter. In many investigations, not all patients who had a diagnostic sleep study also had a follow-up “with appliance” study.
Clearly, the above factors have a significant effect on the analysis of the pooled data presented in Table .
The first step in analyzing the results of individual investigations is to decide on which outcome variables to analyze. Since we are interested in sleep apnea, the following four variables are an obvious choice—baseline index of respiration (we shall denote this as AHIbase), “with appliance” index of respiration (denoted as AHIappl), success rate defined as the reduction of AHIbase to a value less than the defining value for sleep apnea, and response rate defined as the reduction of AHIbase by greater than 50% while still remaining higher than the defining value for sleep apnea.
Table lists 89 distinct investigations, involving a total of 3,027 patients. Inspection of the individual results indicates the widest possible variability. Respiration is analyzed in terms of AI, ODI4, AHI, or RDI. The definition of sleep apnea was based either on AI
10 or ODI4
10. Success and response rates were defined differently in many studies; however, whenever possible we extracted the information to calculate the success and response rates according to the standard definition given above. In other words, no matter what definition of sleep apnea was employed in a particular study, i.e. whether it was defined as AI
10, or ODI
10, etc.—the success rate we calculated was based on the percent of patients in whom AHIappl
was less than 10, and the response rate was calculated as the percent of patients in whom 10
Clearly, it is not possible to carry out a meta-analysis of these studies because of the differences in study design, data collection, statistical analysis and presentation of the data. Even the simple descriptive statistics based on the pooled data must be interpreted with caution due to the methodological differences listed above. To obtain the general information about the efficacy of oral appliances, we analyzed the results of individual investigations in several different ways, as follows.
First, we rejected all case reports [24
] and all studies, which did not report the mean AI, the mean AHI, or the mean RDI [43
]. These studies reported either the total number of apneas per night, snoring only, median AHI, or ODI4, etc. This procedure left a total of 75 studies involving 2,832 patients. We used this “pooled” data to calculate the “pooled” means for each outcome variable—AHIbase
, success rate and response rate. If an individual study did not report this outcome variable—it was not used in the calculation of pooled means. For example, to calculate the AHIbase
, we only had to reject 1 study [23
] where this variable was not reported, thus leaving 74 studies with 2,816 patients. To calculate the AHIappl
, we had to reject 2 studies [23
], thus leaving 73 studies with a total of 2,724 patients. Similarly, looking at Table , we can easily see which studies had to be rejected to calculate the success and the response rates. The results are shown in Table . We note that oral appliance reduced the AHIbase
from 31 to 14. The success and response rates were 54% and 21%, respectively.
Summary of the outcome variables for studies listed in Table
A second way to analyze the data presented in Table is to select only those investigations where all four outcome variables (AHIbase, AHIappl, success rate, response rate) were reported. This left 49 studies involving 1,517 patients. It is interesting to note that the results shown in Table , are almost identical to what was found in a larger dataset.
Summary of results for complete studies
A third way to analyze the data is to select only the randomized, crossover, placebo-controlled studies. There are only five such studies—all done after the year 2000 and all using the mandibular advancement appliance. Inactive appliance was used as placebo in four studies [72
], and a pill was used as placebo in one study [92
]. In the latter study, only the mean AHIs were reported, but not the success and response rates. The results are shown in Table . We note that: (1) only patients with mild to moderate sleep apnea were studied, (2) in two out of five studies, the mean AHI was reduced by less than 50%, (3) the success and response rates differ depending on the cut-off value of the AHI (either five or ten), (4) for AHI
10 cut-off, the success rate is very similar to that of the uncontrolled trials, but the response rate is lower. The overall conclusion from these placebo-controlled crossover trials is that oral appliances significantly improved sleep apnea by reducing AHIbase
from 25 to 14 with the combined success and response rates of 64%.
Summary of results if randomized, crossover, placebo-controlled studies
One of the issues facing the effective use of oral appliances is our ability to determine the appropriate degree of protrusion necessary to resolve apnea and snoring. Until recently, the only way to do so was to advance the mandible forward as much as is tolerated by the patient, to carry out polysomnography and hope that sleep apnea is eliminated. Unlike CPAP, which may be controlled remotely without waking the patient, the design of most appliances requires removing it from patient’s mouth to change the degree of protrusion. This wakes up the patient repeatedly, resulting in a high probability that the titration study will be unsuccessful, thus wasting valuable resources. However, several “titration” protocols were described recently to eliminate this problem. Fleury et al. [93
] studied a protocol where this titration was done at home, over a period of several weeks, by advancing the mandible in 1 mm steps and recording symptoms and ODI4. The effective protrusion was defined as that which results in either resolution in symptoms or reduction in ODI4 to below ten. Using this protocol, they were able to determine the protrusion, which abolished sleep apnea in 64% of patients. Tsai et al. [94
] described a remotely controlled mandibular advancement device, which could be titrated during the night in much the same way that the CPAP is titrated. The mandible was advanced remotely during the night in 1 mm increments until respiratory events were eliminated. The success was confirmed subsequently by carrying out all night polysomnography with the oral appliance set to the effective protrusion determined during the titration study. The positive predictive value of this titration protocol was 90%. On the other hand, Kuna et al. [114
] found that the titration protocol was not predictive of the response during chronic use. In this investigation, 9 out of 21 patients with OSA achieved reduction in AHI to less than 10 during titration. However, none of them demonstrated the same beneficial response during longer use of the appliance at home with the effective protrusion determined during the titration night. These results indicate that oral appliance titration to predict the amount of mandibular advancement required to reduce AHI to less than ten is still imprecise and must be used with caution when determining the appliance settings for home use. Nevertheless, the application of titration protocols is a new and important development in this field, which may improve the success of oral appliance therapy by identifying patients who are likely to respond to this treatment.
Predicting who will respond to the oral appliance therapy is not yet possible, although there are several studies where the differences in various parameters (mainly weight and measures of airway size and collapsibility) were studied in responders and non-responders [39
]. The best correlates were always weight and oropharyngeal airway size. However, although these studies provide useful information, particularly with respect to the factors that determine airway occlusion, they do not as yet provide us with a method to predict who will respond to this therapy.
Oral appliances vs CPAP
Since CPAP remains as the “gold standard” treatment of sleep apnea, the objective success rate of any other treatment must be judged against it. There are seven randomized, crossover studies, which compared mandibular advancement appliances against CPAP. The results are summarized in Table . We note that the findings of all such studies are remarkably consistent—CPAP results in better improvement in AHI than oral appliances. In all studies except one, CPAP normalized the respiration, bringing AHI to less than ten. In one study where AHI with CPAP remained above ten, the highest pressure used was 10 cm H2O, which is probably too low. However, patients subjectively prefer oral appliances over CPAP. In five out of seven studies, the patients expressed preference for an oral appliance, in one study neither treatment was preferred and in another study CPAP was preferred.
Randomized, crossover, CPAP vs oral appliance studies
There are 2 additional investigations [87
], which are not listed in Table because of missing data regarding follow-up sleep study. Nevertheless, these investigations provide interesting information regarding the comparison of oral appliances and CPAP. Smith and Stradling [87
] attempted to determine whether oral appliances can substitute CPAP at least for 1 month. The authors found that patients achieved similar reduction in ODI4 with CPAP (from 29 to 1) as with oral appliance (from 29 to 4). However, the patients did not like their oral appliance and were refusing to use it. Out of 50 patients on CPAP who were switched to oral appliance, only 11 were still using it by end of 1 month. Most patients discontinued its use because of discomfort, side effects, or treatment failure. This study therefore favors CPAP. On the other hand, McGown et al. [102
] carried out a questionnaire survey of 126 patients treated with oral appliances. There were 41 patients who had tried both CPAP and oral appliance; 71% preferred oral appliance, 19% preferred CPAP, and 10% were unsure. This study favors oral appliance.
We conclude therefore that CPAP is more effective than oral appliances in reducing AHI, but despite this, most patients prefer oral appliances, undoubtedly because they find them to be less cumbersome than CPAP.
Oral appliances vs other treatments
Since 1988 there were several studies [21
], which compared either different types of appliances, different degrees of protrusion or different inter-incisal distance. The results are listed in Table . Columns labeled “base” and “appl” show the AHIbase
, respectively. The column labeled “comp” gives the AHI measured when other, i.e. “comparison” treatment was used. The type of oral appliance employed and the precise variables that were compared are described in the last column. Since most studies employed parallel group design, baseline AHI before comparison treatment is also shown in this column.
Studies comparing oral appliances to treatments other than CPAP
Examination of the individual investigations reveals that when oral compliances are compared to each other, either two different appliances or the same appliance with different degrees of protrusion or opening—it is cleat that the efficacy (objective and subjective) is very much dependent on the type of appliance and the degree of advancement. This further emphasizes the point that oral appliance therapy should be carried out by a dentist with expertise in this field who is familiar with different types of appliances and can select the most appropriate one for the particular patient. There is no “best” appliance. The best one is that which is comfortable to the patient and achieves the desired efficacy.
There are several studies, mainly case series, comparing oral appliances with surgical treatments. Comparisons with UPPP [59
] demonstrated the superiority of oral appliances. At 1 year follow-up, sleep apnea was resolved (AHI
10) in 78% of the oral appliances group and 51% of the UPPP group. With longer follow-up, this success rate deteriorates. At 4 years follow-up, 63% of the oral appliance group and 33% of the UPPP group continue to have AHI
10. One parallel group study of oral appliance (TRD) vs radiofrequency ablation (somnoplasty) found that both treatments significantly reduced sleep time spent with loud snoring. However, there was no difference between oral appliance and somnoplasty [21
]. A recent small case series of 4 patients (out of 43 treated with oral appliances) who elected maxillomandibular advancement surgery [121
] showed that initial AHI
50 dropped to 12 with oral appliance and to 2 after surgery.