The Efficacy of Device Designs (Mono-block or Bi-block) in Oral Appliance Therapy for Obstructive Sleep Apnea Patients: A Systematic Review and Meta-Analysis

Oral appliance (OAm) therapy has demonstrated efficacy in treating obstructive sleep apnea (OSA). The aim of this systematic review was to clarify the efficacy of device designs (Mono-block or Bi-block) in OAm therapy for OSA patients. We performed a meta-analysis using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) system. Two studies (Mono-block OAm versus Bi-block OAm) remained eligible after applying the exclusion criteria. When comparing Mono-block OAm and Bi-block OAm, Mono-block OAm significantly reduced the apnea–hypopnea index (2.92; 95% confidence interval (95%CI), 1.26 to 4.58; p = 0.0006), and patient preference for Mono-block OAm was significantly higher (2.06; 95%CI, 1.44 to 2.06; p < 0.0001). Lowest SpO2, arousal index, non-REM stage 3, sleep efficiency, Epworth Sleepiness Scale (ESS), Snoring Scale, and side effects were not significantly different between the two groups (lowest SpO2: −11.18; 95%CI, −26.90 to 4.54; p = 0.16, arousal index: 4.40; 95%CI, −6.00 to 14.80; p = 0.41, non-REM stage 3: −2.00; 95%CI, −6.00 to 14.80; p = 0.41, sleep efficiency: −1.42, 95%CI, −4.71 to 1.86; p = 0.40, ESS: 0.12; 95%CI, −1.55 to 1.79; p = 0.89, Snoring Scale: 0.55; 95%CI, −0.73 to 1.83, p = 0.55, side effects: 1.00, 95%CI, 0.62 to 1.61, p = 1.00). In this systematic review, the use of Mono-block OAm was more effective than Bi-block OAm for OSA patients.


Introduction
Obstructive sleep apnea (OSA) is a disease where the upper airway narrows or collapses repeatedly during sleep [1][2][3][4]. Anatomical factors of the upper airway and decompensation due to neuromodulation have been reported as the main causes of OSA [3,4]. In particular, obesity and micrognathia worsen the anatomical balance around the throat, increase pharyngeal obstruction, and are highly involved in the onset of OSA [5].
Oral appliance (OA) therapy is a treatment option for OSA [6]. Among the different appliances, mandibular advancement-type oral appliances (OA m ) are mainly used. Wearing an OA m will pull the lower jaw forward and cause an expansion of the upper airway and oral cavity volume, thus preventing upper airway obstruction during sleep [7]. OA m are outstanding in their simplicity, economic efficiency, and portability, and recent reports show the beneficial effects of OA m even for severe cases of OSA [8,9]. An American Academy of Sleep Medicine (AASM) and American Academy of Dental Sleep Medicine (AADSM) clinical practice guideline recommends that sleep physicians consider the prescription of oral appliances, rather than no treatment, for adult patients with OSA who are intolerant of continuous positive airway pressure (CPAP) therapy or prefer alternate therapy [10].
There are various types of OA m devices [11]. However, most fall under one of two major designs: Mono-block and Bi-block types. As for Mono-block OA m , since both the upper and lower jaw are fixed in place by the same device, jaw movement is restricted, leading to a sense of constraint during sleep. Further, there are many cases in which the anteroposterior position of the lower jaw cannot be easily adjusted. On the other hand, the Bi-block OA m has separate constructions for the upper and lower jaws and is equipped with connectors or attachments that advance the mandible. The lower jaw can be opened and moved sideways, resulting in a reduced sense of constraint and discomfort for the patient. The Bi-block OA m has a device that completely separates the upper and lower jaws, and a movable device that cannot completely separate the upper and lower jaws. The main feature of the Bi-block OA m is the ease of anteroposterior position adjustment of the lower jaw by means of a screw or connector built into the device. However, the treatment principle of expanding the upper airway by moving the lower jaw forward is the same in Mono-block OA m and Bi-block OA m .
Although there are several systematic reviews (SRs) on OA m designs, a decisive conclusion as to which design is most effective for OSA treatment has not yet been reached. Therefore, in this systematic review, we extracted research results from past comparative studies of Mono-block OA m and Bi-block OA m (or similar designs), conducted a meta-analysis, and evaluated each design according to its OSA treatment efficacy. If one type was superior to the other, we evaluated the different designs to figure out the most effective design of that type. Our meta-analysis was performed using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) system [12,13].

Materials and Methods
This SR was performed following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines [14]. The protocol for this review was registered with the international prospective register of systematic reviews (PROSPERO) with the registration number CRD42019131303.

Eligibility Criteria
For this review, we referenced studies using OA m on OSA patients aged 18 years and above. When selecting studies for this review, the following criteria regarding the OA m had to be fulfilled: (1) the device was custom-made by creating a dental impression of the lower jaw for every patient, (2) the OA m design was classified only as Mono-block and Bi-block types, and (3) titration of the OA m (including adjustment) was performed. Further, in order to comprehensively evaluate the influence of the structure of the OA m , we did not establish any restrictions regarding the layout, material, thickness, or detailed design of OA m (type of connector used in Bi-block, presence of air vents in Mono-block, etc.). Another criterion for the inclusion of studies was the evaluation of diagnostic and therapeutic OSA treatment efficacy through either polysomnography (PSG) or Out of Center Sleep Testing (OCST). We excluded studies that included patients under the age of 18 years and patients with other coexisting sleep disorders, studies in which ready-made devices were used, and studies in which results were achieved by securing the tongue in a forward position through tongue retaining devices.

Literature Search
The searches were performed using the following databases: MEDLINE, Cochrane Central Register of Controlled Trials (CENTRAL), and Igaku Chuo Zasshi (Ichushi-Web). In the literature search, there were no restrictions on the year or language published. We selected only randomized controlled trials (RCTs) in the design of the OA m , detailing the comparison of the Mono-block to the Bi-block or similar designs. Prospective and retrospective non-randomized clinical studies, case reports/case series, conferences abstracts, and reviews were excluded. A search strategy is shown in Figure 1 and the final literature search was conducted and completed on 27 April 2019. When searching, if an unpublished article such as a paper on conference proceedings was found, we investigated whether or not it was published later. Mono-block, etc.). Another criterion for the inclusion of studies was the evaluation of diagnostic and therapeutic OSA treatment efficacy through either polysomnography (PSG) or Out of Center Sleep Testing (OCST). We excluded studies that included patients under the age of 18 years and patients with other coexisting sleep disorders, studies in which ready-made devices were used, and studies in which results were achieved by securing the tongue in a forward position through tongue retaining devices.

Literature Search
The searches were performed using the following databases: MEDLINE, Cochrane Central Register of Controlled Trials (CENTRAL), and Igaku Chuo Zasshi (Ichushi-Web). In the literature search, there were no restrictions on the year or language published. We selected only randomized controlled trials (RCTs) in the design of the OAm, detailing the comparison of the Mono-block to the Bi-block or similar designs. Prospective and retrospective non-randomized clinical studies, case reports/case series, conferences abstracts, and reviews were excluded. A search strategy is shown in Figure 1 and the final literature search was conducted and completed on 27 April 2019. When searching, if an unpublished article such as a paper on conference proceedings was found, we investigated whether or not it was published later.

Study Selection Procedure
Eligible studies were selected in two phases. In the first phase, two authors independently screened the titles and abstracts. In the second phase, the full texts of all potentially eligible studies identified during the first phase were independently reviewed by two authors. During the full-text assessment, irrelevant studies were excluded based on the inclusion and exclusion criteria. When the selection differed between the two authors, we added a third author and solved the issue by discussion.

Study Selection Procedure
Eligible studies were selected in two phases. In the first phase, two authors independently screened the titles and abstracts. In the second phase, the full texts of all potentially eligible studies identified during the first phase were independently reviewed by two authors. During the full-text assessment, irrelevant studies were excluded based on the inclusion and exclusion criteria. When the selection differed between the two authors, we added a third author and solved the issue by discussion.

Data Extraction
Data were extracted independently by three authors. The important information collected was the author name, year of publication, study design, type of device, number of subjects and dropouts, mean age, body mass index (BMI), mean baseline apnea-hypopnea index (AHI), follow-up periods, and study outcome. The extracted results included values for both before and after the OA m treatment. The primary outcomes were mortality rates and cardiovascular events, and surrogate outcomes were as follows: (1) treatment effects: AHI (including the respiratory disturbance index (RDI) and respiratory event index (REI)), lowest SpO 2 , arousal Index, sleep efficiency, sleep stage (non-REM stage 3: NREM stage 3), subjective daytime sleepiness (the Epworth Sleepiness Scale (ESS)), the loudness and effects (disturbance factor) of snoring (the Snoring Scale (SS)); (2) adherence (the duration of OA m usage at night and the number of days the OA m was used in the preceding week); (3) sleep-related quality of life (SF-36 Physical Component, SF-36 Mental Component); (4) hypertension: systolic blood pressure, diastolic blood pressure; (5) side effects: temporomandibular disorders (arthralgia or myalgia), tooth pain, occlusal changes (overbite, overjet), the changes of occlusal contact, the changes of bite force; and (6) patient preference. If the standard error of the mean (SEM) was reported for outcomes, the standard deviation (SD) was also calculated from the number of subjects in the study and the reported SEM.

Data Synthesis and Statistical Analysis
Meta-analyses were conducted using Review Manager (Nordic Cochrane Centre, Cochrane Collaboration, 2014 Copenhagen, Denmark) version 5.3 software by pooling data across studies for each outcome measure. The available case analysis was applied where data were missing. When multiple studies were combined, the risk ratio (RR) and the mean difference (MD) were used. The effectiveness was evaluated based on RR or MD and its 95% confidence interval (CI) in each study. A forest plot was constructed using the RR or MD of the outcome variable in each study. All analyses were performed using the fixed-effects model with results displayed as a forest plot.

Quality Assessment of Included Studies
The evaluation of the quality of evidence in this SR was performed using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) process [12]. The GRADE classification was downgraded by one level for each of the five factors we considered, which were study limitations, inconsistency, indirectness, imprecision, and publication bias. Seven authors judged whether the five factors were present for each outcome. The risk of bias of the studies included for this review were evaluated using the Cochrane Risk of Bias tool described in the Cochrane Handbook for Systematic Reviews of Intervention. A GRADE evidence profile was created using the GRADEpro software for each clinical domain and adverse events. The following definitions of the quality of the evidence were applied [13]: high quality (further research is very unlikely to change our confidence in the estimate of the effect), moderate quality (further research is likely to have an important effect on our confidence in the estimate of the effect and may change the estimate), low quality (further research is very likely to have an important effect and is likely to change the estimate), very low quality (we are very uncertain about the estimate).

Literature Search and Selection Results
The PRISMA flow diagram is presented in Figure 2. Our selection of studies included 201 MEDLINE articles, 385 Cochrane Central Register of Controlled Trials (CENTRAL) articles, and 31 Igaku Chuo Zasshi (Ichushi-Web) articles. Excluding 158 duplicate articles, we screened the titles and abstracts of 459 articles. We then narrowed our selection down to 18 articles, of which 10 were excluded after confirming the eligibility of each articles (three duplicate articles, one non-RCT article, four articles without OA m titration and with fixed mandibular position at the time of OA m device fabrication, two cases of conference proceedings, one case in which the respective paper could not be located). Finally, seven articles were accepted (two articles of Mono-block OA m versus Bi-block OA m [15,16] and five articles of Bi-block OA m versus Bi-block OA m [17][18][19][20][21] fabrication, two cases of conference proceedings, one case in which the respective paper could not be located). Finally, seven articles were accepted (two articles of Mono-block OAm versus Bi-block OAm [15,16] and five articles of Bi-block OAm versus Bi-block OAm [17][18][19][20][21]). We were not able to find any articles covering a comparison among Mono-block OAm.
A risk of bias summary of randomized studies in Mono-block OAm versus Bi-block OAm according to Cochran's tool is shown in Figure 3. The study design adopted in two [15,16] of the selected studies was a randomized crossover study, which we regarded as having high risk of performance bias. However, in the outcome evaluation of both reports, the bias had been judged to be small and thus we evaluated it as low risk. Another potential bias was that subsidies were granted as research funds [15]. However, since these were received by the university, we evaluated the case as having low risk of bias. A risk of bias summary of randomized studies in Bi-block OAm versus Biblock OAm according to Cochran's tool is shown in Figure 4. The study design adopted in four out of five articles [17][18][19][20][21] was a randomized crossover study, which we regarded as having high risk of performance bias. Additionally, regarding attrition bias, a dropout was recorded in the middle of one study [20]. Since no ITT (intention-to-treat) analysis was conducted, we evaluated the case as having a high risk of case reduction bias. Another potential bias was that subsidies were granted as research funds [21]. However, since these were public funds, we evaluated the case as having low risk of bias.
A risk of bias summary of randomized studies in Mono-block OA m versus Bi-block OA m according to Cochran's tool is shown in Figure 3. The study design adopted in two [15,16] of the selected studies was a randomized crossover study, which we regarded as having high risk of performance bias. However, in the outcome evaluation of both reports, the bias had been judged to be small and thus we evaluated it as low risk. Another potential bias was that subsidies were granted as research funds [15]. However, since these were received by the university, we evaluated the case as having low risk of bias. A risk of bias summary of randomized studies in Bi-block OA m versus Bi-block OA m according to Cochran's tool is shown in Figure 4. The study design adopted in four out of five articles [17][18][19][20][21] was a randomized crossover study, which we regarded as having high risk of performance bias. Additionally, regarding attrition bias, a dropout was recorded in the middle of one study [20]. Since no ITT (intention-to-treat) analysis was conducted, we evaluated the case as having a high risk of case reduction bias. Another potential bias was that subsidies were granted as research funds [21]. However, since these were public funds, we evaluated the case as having low risk of bias.   There were no significant statistical differences in treatment outcomes between the two appliances. There was a statistically significant (p < 0.05) preference for a OAm design with minimal coverage of teeth and palate.
There was a trend toward greater improvement with the appliance with less acrylic resin bulk and less interocclusal contact. OA selection should favor titratable, unobtrusive designs with appropriate construction to promote acceptance and adherence to OA therapy. The results showed that a statistically significant improvement in the respiratory parameters was achieved with both appliances (p < 0.01). However, the activator was significantly more effective (p < 0.01) than the Silencor.
Both appliances reduced daytime sleepiness and snoring and improved sleep quality, and both influenced the treatment outcome.

Meta-Analysis
Most of the articles included in this SR were short-term reports and there were no reports of mortality or cardiovascular events. We extracted and evaluated surrogate outcomes from included articles.

Meta-Analysis
Most of the articles included in this SR were short-term reports and there were no reports of mortality or cardiovascular events. We extracted and evaluated surrogate outcomes from included articles.

Meta-Analysis
Most of the articles included in this SR were short-term reports and there were no reports of mortality or cardiovascular events. We extracted and evaluated surrogate outcomes from included articles.

Mono-Block OA m Versus Bi-Block OA m
The forest plots in the surrogate outcomes (AHI, lowest SpO 2 , arousal Index, NREM stage 3, sleep efficiency, ESS, Snoring Scale, side effects, patient preference) are shown in Figure 5.

Bi-Block OAm Versus Bi-Block OAm
Although the study designs adopted in all five articles [17][18][19][20][21] were randomized controlled trials, various OAm devices were utilized, such as Herbst, Thornton Adjustable Positioner (TAP), Klearway, and IST models. Additionally, the material, thickness, and detailed design of each device differed, making it difficult to generate a comprehensive interpretation of results. Therefore, for this systematic review, no meta-analysis was performed regarding these five studies. Table 3 provides the GRADE evidence profile. A funnel plot was not used to assess publication bias in this meta-analysis because the number of eligible articles was less than 10. As a consequence of risk of bias, inconsistency, indirectness, imprecision, and other considerations, the overall quality of evidence for the AHI and patient preference was rated as low. The quality of evidence for the lowest SpO2, arousal index, NREM stage 3, sleep efficiency, ESS, Snoring Scale, and side effects was rated as very low. Although the study designs adopted in all five articles [17][18][19][20][21] were randomized controlled trials, various OA m devices were utilized, such as Herbst, Thornton Adjustable Positioner (TAP), Klearway, and IST models. Additionally, the material, thickness, and detailed design of each device differed, making it difficult to generate a comprehensive interpretation of results. Therefore, for this systematic review, no meta-analysis was performed regarding these five studies. Table 3 provides the GRADE evidence profile. A funnel plot was not used to assess publication bias in this meta-analysis because the number of eligible articles was less than 10. As a consequence of risk of bias, inconsistency, indirectness, imprecision, and other considerations, the overall quality of evidence for the AHI and patient preference was rated as low. The quality of evidence for the lowest SpO 2 , arousal index, NREM stage 3, sleep efficiency, ESS, Snoring Scale, and side effects was rated as very low.

Discussion
In this SR, we investigated which designs were effective when performing OA m treatment for OSA patients. The primary outcome of successful OSA treatment is the improvement of life prognosis and the prevention of cardiovascular disease [22,23]. However, there were no reports of mortality or cardiovascular events in the included articles. The surrogate outcomes are often used, and there are many articles that have reported on surrogate outcomes. Therefore, in this systematic review, we analyzed the surrogate outcomes referred to in previous SRs [6,10,24].
Several systematic reviews on the design of OA m have been reported in the past [25][26][27]. Ahrens et al. [25,26] investigated the subjective patient outcome and polysomnographic indices. They concluded that there was no specific OA m design to improve the subjectively perceived treatment efficiency and the polysomnographic indices effectively. Serra-Torres et al. [27] showed that adjustable and custom-made OA m have a better performance than fixed and prefabricated devices, and that Mono-block OA m cause more adverse effects. However, the application of OA m used in the reports accepted for these systematic reviews did not include the titration of the lower jaw. In this systematic review, we targeted OA m for which titration (including adjustment) of the lower jaw was performed. As for the forward positioning of the lower jaw, many devices are set at 50-75% of the maximum mandibular advancement amount [28,29]. However, a standardized value has yet to be established. Pulling the lower jaw forward expands the upper airway and increases treatment efficacy [30]. However, the burden on the temporomandibular joint, masticatory muscles, and teeth increase, which can lead to an increased risk of side effects such as temporomandibular disorders or tooth movement [31]. In clinical practice, it is necessary to perform the titration of the mandibular position after its initial fixation in order to strike a balance between treatment efficacy and side effects [32]. Therefore, the results of this systematic review can be considered to have high clinical significance.
There were only two studies comparing Mono-block OA m and Bi-block OA m . Meta-analysis results showed that Mono-block OA m were more effective in terms of AHI and patient preference values. During sleep, humans perform various jaw movements such as sleep bruxism and swallowing; however, for most of the time during sleep, the mouth remains open [33]. Generally, skeletal muscles relax and muscle tonus decreases during sleep [34]. The same phenomenon is observed for masticatory muscles, and it is assumed that the opening of the mouth is a result of the relaxation of muscles responsible for keeping the mouth closed [35]. It is assumed that in some cases, the tongue and soft palate retract with the opening of the mouth, which leads to a narrowing of the respiratory tract and resulting respiratory problems. Bi-block OA m contain various adjustment mechanisms to easily adjust the position of the lower jaw. Such layouts make it possible to open the mouth more or less freely. However, this also implies an increased chance of the lower jaw moving backward and downward during sleep. That is to say, the improvement of AHI values may be poor. On the other hand, with Mono-block OA m , the upper and lower jaws are fixed in one position and mouth movement is often restricted. Therefore, as also reported by Bloch et al. [15] and Zhou et al. [16], AHI values improved at a better rate with Mono-block OA m compared to Bi-block OA m .
There was also a report claiming that AHI values decreased significantly with Bi-block OA m . It should be noted that the average BMI and AHI values differed from case to case. Ghazal et al. [20] reported a small decrease in AHI from 21.5 ± 13.5 to 11.1 ± 11.8 (48.4% decrease) after using IST devices for six months (n = 47). However, after 24 months (n = 24), AHI values significantly improved, from 18.4 ± 8.9 to 4.6 ± 5.8 (75.0% decrease). Further, after six months of use of TAP devices for control (n = 48), AHI values significantly improved from 19.8 ± 12.7 to 6.7 ± 9.1 (68.8% decrease). After 24 months (n = 21), the device remained effective and AHI values improved significantly from 19.8 ± 12.7 to 5.4 ± 5.1 (72.7% decrease). There are various types of Bi-block OA m ; however, a collective opinion as to which design is most effective has not yet been formed. This systematic review only includes studies in which titration or post-adjustment comparison was performed. Regarding the AHI results, it was inferred that a large amount of mandibular advancement was set in the Mono-block OA m with complicated adjustment. Additionally, the capability of opening the mouth granted by Bi-block OA m was also considered to be a contributing factor. Further, the ESS, Snoring Scale, and side effects meta-analysis results showed no difference between Mono-block and Bi-block OA m . Although no detailed descriptions concerning adherence were given, no differences were found. However, it should be noted that the patient preference values were higher for Mono-block OA m . According to Bloch et al. [15], 15 out of 24 subjects (62.5%) preferred Mono-block OA m , whereas one subject (6.7%) preferred the Herbst model, which is a Bi-block OA m . The following reasons for the preference of Mono-block OA m were reported: greater alleviation of OSA symptoms (n = 11), robustness and ease of installation (n = 5), and less side effects (n = 4). The subject who preferred the Herbst model stated an impression of greater alleviation of OSA symptoms. Further, according to Zhou et al. [16], seven out of 16 subjects (43.8%) preferred Mono-block OA m , whereas two subjects (12.5%) preferred the SILENT NITE model, which is a Bi-block OA m . Although detailed reasons regarding preferences for each device were not clear, it was reported that six subjects experienced equipment failure with the SILENT NITE model during the study period, which may have contributed to the high number of subjects preferring Mono-block OA m . It can be said that devices producing high patient preference values seem to satisfy not only in terms of therapeutic efficacy and side effect alleviation, but also the ease of installation and durability of the OA m . Although Mono-block OA m evoke an image of severe constraint, this sense of constraint does not necessarily seem to be reflected in the patient preference values. In recently reported cases of OSA patients with locomotive syndrome and paralysis after a cerebral infarction [36], methods that allow for an easy installation seem to be preferred. In particular, Mono-block devices utilizing soft resin can be installed with only one hand. It seems that the choice of equipment also has to be considered in the context of each individual patient.
It should be noted that our systematic review was conducted under the following limitations. Firstly, a small sample size and limited number of studies were included in the review, with many reports covering only short research periods. Secondly, we included studies with a high risk of bias due to subject blind testing or incomplete data. Follow-up research will require a revised research design and greater sample size. In particular, OA m treatment is highly symptomatic and conducted over a long period of time, which necessitates further investigations of adherence and side effects.

Conclusions
Despite frequently small sample sizes and a majority of short-term reports, current evidence shows that Mono-block OA m are more effective than Bi-block OA m for OSA patients. Further well-designed, larger trials are required to determine the benefit for patients.
Author Contributions: H.I. and D.H. contributed to the study design; data collection, interpretation, and statistical analysis; and manuscript preparation. Y.S. contributed to the data collection, interpretation, and statistical analysis; and manuscript preparation. K.S. and H.Y. contributed to the study design; data collection, interpretation, and statistical analysis; and manuscript preparation. A.F. and E.K. contributed to data interpretation and manuscript preparation.
Funding: This work was supported by grants from The Japanese Academy of Dental Sleep Medicine.