Acoustic pharyngometry ‐ A new method to facilitate oral appliance therapy

Abstract Background There is lack of reliable and accurate methods to predict treatment outcomes of oral appliance (OA) treatment. Acoustic pharyngometry (AP) is a non‐invasive technique to evaluate the volume and minimal cross‐sectional area of the upper airway, which may prove useful to locate the optimal position of OAs. Objective This retrospective study aimed to evaluate the effect of applying AP to OA treatment of patients with obstructive sleep apnoea (OSA). Methods All patients (n = 244) treated with OAs following an AP protocol at two dental clinics between 2013 and 2018 were invited to participate. A total of 129 patients accepted the invitation, and 120 patients (75 men, 45 women) were included in the analyses. Mean baseline age, BMI and apnoea hypopnea index (AHI) were 59.1 ± 0.9 years, 27.8 ± 0.4 and 21.9 ± 1.1, respectively. Mean follow‐up time was 318 ± 24 days. Results AHI at follow‐up was 6.4 ± 0.7, resulting in a treatment success rate of 86.7% (≥50% reduction of baseline AHI). The number of failures (<50% reduction of baseline AHI) did not differ significantly among patients with mild, moderate and severe OSA. 87.6% of the patients reported OA usage every night, and 95.5% reported > 5 hours usage per night, when worn. Conclusion The AP protocol applied seems to contribute to the excellent effect of OA treatment in this study. Further research on the application of AP in OA treatment is necessary in order to clarify its possible beneficial contribution to improving OA therapy.

OSA. 3 However, a more recent study based on updated diagnostic criteria on a similarly aged Swiss population suggests even higher prevalence rates. 4 Continuous positive airway pressure (CPAP) is the gold standard in OSA treatment, defined by criteria set by the American Association of Sleep Medicine (AASM), 1 but the adherence to CPAP is generally poor, and reported as an average of 65.9% in a large meta-analysis. 5 Consequently, development of effective second-line treatment options, such as oral appliance (OA) therapy, is crucial. In recent studies, excellent objective compliance with OA therapy has been reported. 6,7 Evidence suggests that OA treatment may be comparable to CPAP in improving health outcome, even in more severe OSA, as the greater usage of OA may compensate for its inferior efficacy compared to CPAP. 8,9 Objectively measured efficacy with OA treatment, that is reductions in apnoea hypopnea index (AHI) and oxygen desaturation index (ODI), will hardly ever reach that of CPAP. It would therefore be important to develop techniques in order to improve the efficacy of OA treatment.
OA titration/mandibular positioning is in many cases a 'trial and error' procedure, where the objective effect of each titration has to be assessed through an overnight polysomnographic (PSG) or an ambulatory polygraphic (PG) examination. 10 Aside from complex and costly approaches, there is a lack of reliable and accurate methods to predict treatment outcome of OA therapy. 11 Maximum comfortable protrusion is often referred to as the gold standard for titration of OAs, and clinicians often resort to increased mandibular protrusion when there is a lack of effect of OA treatment. However, a systematic review investigating the effectiveness of different mandibular advancements of OAs found that titration of the mandible above 50% of maximal protrusion did not significantly influence the success rate of the treatment, as measured by the reduction of AHI. 12 A study investigating the minimal protrusion needed for sufficient effect of OA therapy concluded that a great proportion of the study sample showed sufficient AHI reduction with no protrusion at all. 13 Historically, clinicians have relied on subjective symptoms as a measure of the efficacy of OA treatment. However, the relationship between an objective measure of the severity and subjective symptoms of OSA, such as daytime sleepiness, may be overestimated. 14 A reduced effect of OA treatment often results in that the clinician increases the protrusion of the mandible by titrating the appliance, which may increase the risk of side effects, 15 and which in turn could adversely affect the adherence to treatment.
The role of the vertical dimension in OA treatment remains unclear. Studies have shown that the vertical dimension of the appliance does not significantly impact the efficacy of the treatment, [16][17][18] while another study found that increased vertical dimension in maximal comfortable protrusion might cause increased pharyngeal collapsibility during sleep. 19 Nevertheless, the accepted gold standard for vertical dimension in OA treatment is to keep the vertical dimension as low as possible. Hence, the need for individualised mandibular advancement and vertical dimension of OA in order to optimise comfort and treatment efficacy seems obvious, and would be in accordance with recent consensus within the field of dental sleep medicine. 20

| Acoustic pharyngometry
Acoustic pharyngometry (AP) is a non-invasive technique that uses sound wave reflections to supply information about the crosssectional area and volume of the upper airway, from the lips to the epiglottis. 21,22 Information is graphically and instantly displayed as a curve representing the cross-sectional area (y-axis) along the first 25 cm of the upper airway (x-axis) ( Figure 1, Appendix 1). The total volume of the first 25 cm of the upper airway is depicted as the area under the curve for the sum of the cross-sectional areas, while the F I G U R E 1 Comparison between baseline OSA diagnosis mild (n = 38), moderate (n = 59) and severe OSA (n = 23), according to success criteria applied at follow-up after oral appliance treatment. Patients were only classified as one of the success criteria, being the lowest criteria possible; that is, if they achieved AHI < 10 and > 50% reduction of baseline AHI at followup, the treatment would be classified as success criterion 2 and not success criterion 3 [Colour figure can be viewed at wileyonlinelibrary.com] minimal cross-sectional area indicates where the upper airway is at its narrowest point. Airway volume measurements obtained with AP have been validated against computed tomography and magnetic resonance imaging, with acceptable accuracy. 23,24 The reproducibility of the AP registration technique has also been investigated. 25,26 A standard operating protocol for acoustic pharyngometry has been proposed for normal breathing, and the repeatability of the protocol has also been validated. 22,25 The collapsibility, or propensity for collapse of the upper airway (which is quantified as pressure differences in the upper airway during inspiration under negative generated pressure), and the minimal cross-sectional area have been proposed to be predictors of the presence of moderate/severe OSA. 27,28 Volume changes of the pharyngeal cavity, as measured with AP, during positive and negative pressure indicate upper airway collapsibility, with significantly greater differences shown in OSA patients. 29 Independent of age, gender and neck size, objective anatomical assessment with AP discriminates well between patients with mild versus moderate to severe OSA in a clinical setting. 30 In addition, it has been shown that the pharynx of OSA patients shows greater tendency to collapse (i.e. collapsibility) when exhaling slowly from total lung capacity to residual volume (modified Mueller manoeuvre) compared with non-OSA patients. 31,32 The collapsibility generated by the modified Mueller manoeuvre has previously been shown to be reproducible with acceptable intra-class correlation (ICC). 33 In regard to the foregoing, by using AP it should be possible to locate the position of the mandible that provides the lowest pharyngeal collapsibility and the largest minimal cross-sectional area in the upper airway during a modified Mueller manoeuvre. This information could be used to determine the optimal position of the mandible and the OA.
The aim of this study was to evaluate the use of AP in OA treatment in patients with mild, moderate and severe OSA.

| Patients and diagnostic procedures
All patients (n = 244) referred to two dental clinics (125 to Tannhelsesenteret Lørenskog and 119 to Tannhelsesenteret Sogndal) for OA treatment between 2013 and 2018 were invited to participate in the study. The patients were treated according to AP protocol described in detail in Appendix 1. The treatment was performed by one of two calibrated dentists. The calibration comprised training sessions before the introduction of the AP methodology in the clinic. The calibration between the two dentists encompassed how to locate the optimal position of the OA based on the AP measures. A total of 129 patients (men = 80, women = 49) accepted the invitation and were included in the study. The baseline diagnosis of OSA was performed by ear, nose and throat specialists at referring hospitals, using one of two PG type IV devices (Embletta® or Nox T3®). The criteria for mild OSA were AHI 5 -<15, for moderate OSA AHI 15 -<30, and for severe OSA AHI ≥ 30. 1 The objective effect of the OA treatment was investigated through home sleep testing with a PG type III device (Apnealink Plus TM or Apnealink Air TM , ResMed Germany Inc), with scoring rules in accordance with the 2007 AASM guidelines. 34 Two patients were excluded from the study because they were diagnosed using other scoring rules. A total of 7 patients were excluded from the study because they did not have an OSA diagnosis (AHI < 5 at baseline) and were treated with OAs on the indication of social snoring. The distribution of baseline diagnoses of the 120 patients finally included was the following: 38 patients mild OSA (32%), 59 moderate OSA (49%) and 23 with severe OSA (19%). A total of 111 patients (94%) had previously tried CPAP but were non-compliant, 5 (4%) had never tried CPAP, and 2 (2%) of the patients did manage CPAP treatment but were referred for OA treatment for other reasons. Data describing previous CPAP compliance were missing for 2 patients.

| Treatment protocol
The operating protocol for AP measures and locating the acoustic optimal position (AOP) of the mandible for the positioning of the OAs is described in detail in Appendix 1. Jaw registration was conducted in the individually determined position deemed optimal by AOP for all patients. Subsequent to delivery and adjustment of the OA, patients were instructed to use their OA throughout every night. The treatment was at minimum controlled after 3 weeks and 3 months. However, the majority of the patients were followed up for a longer time due to different follow-up protocol required by referring hospitals in the early treatment years. Some patients were controlled more frequently if their OA needed adjustment to improve the effect of the treatment, or to handle compliance issues. and the PG recording showed adequate objective effect on AHI (AHI < 5), the OA was not adjusted from the baseline AOP, regardless of the new AP registration. Two different OAs were used in this study, the Narval CC TM appliance (n = 88, 73%) and the Somnodent® Fusion (n = 32, 27%). The latter was used primarily in cases where the Narval appliance was not indicated due to retention issues or in cases where patients did not want the Narval appliance. None of the included patients used elastic bands on their OAs.

| Success criteria
The success at the final PG follow-up of the OA treatment was divided into three categories, according to Gjerde et al 34 Patients were only classified according to one of the success criteria, being the lowest criteria possible; that is, if they achieved AHI < 10 and > 50% reduction of baseline AHI at follow-up, the treatment would be classified as success criterion 2 and not success criterion 3.

| Statistical analyses
All statistical analyses were performed with Stata version 16.0 (College Station). Logistic regression analysis was performed in order to investigate associations with the strictest treatment success criterion 1 (AHI < 5) as the dependent variable. An identical logistic regression analysis was performed applying success criterion 3 (≥50% reduction of baseline AHI) as the dependent variable. Selection of independent variables in the unadjusted regression model included all those that were deemed to hypothetically influence the outcome of the dependent variable. Unadjusted logistic regression analysis (crude estimate) was first performed. Subsequently, an adjusted logistic regression analysis was performed including the independent variables exhibiting P < .1 in unadjusted analysis, applying a stepwise forward conditional method. Adjusted odds ratios with P < .05 were considered statistically significant in the adjusted model.
Unadjusted linear regression analysis was performed to investigate associations between side effects and subjective reported compliance. P < .05 was considered statistically significant in the linear regression. Kruskal-Wallis test was used to analyse differences between baseline AHI in the three success groups and the failures.
The Kruskal-Wallis test was supplemented with a Mann-Whitney U test to analyse significant differences using Bonferroni correction as Post hoc test, that is P < .017 and P < .008 for 3 and 4 groups, respectively, was considered statistically significant.
For the missing data analysis, linear regression analysis was performed to investigate differences in age, baseline BMI and AHI between the included respondents and the non-respondents in this study. Pearson's chi-squared test was performed to investigate differences in gender between respondents and non-respondents. P < .05 was considered statistically significant.
All data are presented as mean ± standard error of the mean.

| Ethics
The study was approved by the regional ethics committee of Western Norway, and informed written consent was obtained from all participants (protocol no: 2018/2121). The study was also approved by the Health and social services representative of the University of Bergen (protocol no: 2018/2121).

| RE SULTS
A total of 120 patients (men = 75, women = 45) were included in the follow-up evaluation. At baseline, 16 patients (13.9%) were smokers, 28 (24.4%) had hypertension, 8 (7.0%) had cardiovascular disease and 3 (2.6%) diabetes. Table 1 provides further baseline patient characteristics. The overall success rate based on criteria 1 to 3 was 86.7%, and 81.7% based only on success criteria 1 to 2 ( Figure 1). For all patients, mean AHI at follow-up was 6.4 ± 0.7, a reduction from a mean AHI of 21.9 ± 1.1 at baseline, which represents a mean percentage AHI reduction of 66.7% ± 3.8. Follow-up AHI in the three OSA baseline diagnoses of mild, moderate and severe, were 5.2, 5.6 and 10.3, respectively. When comparing the follow-up AHI in the three OSA groups, the severe group had significantly higher AHI than the mild group (P = .01), but not so compared to the moderate group (P = .09) (Mann-Whitney U test). There were no significant differences among the three baseline diagnoses and the success criteria recorded at follow-up (Kruskal-Wallis, P = .33).
As regards self-reported compliance at follow-up, 87.6% of the patients reported usage of their OA every night, and 95.5% re-

| Acoustic pharyngometry results
Descriptive data for AP measurements are shown in Table 1

| Missing data analysis
Of the total 244 patients, 120 out of the 129 respondents were included in the final analysis, while 115 (non-respondents) did not give informed consent and were not included. Missing data analysis was performed between the respondents and non-respondents regarding age, gender, baseline BMI and AHI. There were no significant differences between respondents and non-respondents with regard to mean baseline AHI (21.9 vs 21.7) and mean BMI (

| D ISCUSS I ON
This retrospective study of 120 OSA patients treated with OAs following an acoustic pharyngometry protocol showed an overall success rate of 86.7% using success criterion 1-3 (≥50% reduction of AHI from baseline). When analysing patients with moderate (n = 59) and severe (n = 23) OSA separately, the success rates using criterion 1-3 were 88.1% and 87.0%, respectively, which indicates that the success rate for this population was not affected by OSA severity (Figure 1).
The result of OA treatment following the AP protocol applied in this study seems to be superior compared to studies following the standard protocol for OA treatment using similar success criteria (≥50% reduction of AHI from baseline), where success rates of 77% and 69% were achieved for patients with moderate and severe OSA respectively in one study, and a success rate of 58% for the entire population of mild to severe OSA patients in another. 35,36 A more recent study describing outcomes of standard protocol OA treatment in severe OSA patients presented 60% success following success criteria 3. 37 The patients included who were treated with an effective OA and not a sham device (n = 55) had comparable baseline AHI, BMI, occurrence of hypertension and distribution of men/woman as the patients with severe OSA (n = 23) in this study. However, the difference in number of included patients should be noted.
The missing data analysis showed that the included patients were representative for the population in terms of baseline AHI and BMI, while age and gender differed significantly between the respondents and non-respondents; that is, the non-respondents had an overrepresentation of men and younger individuals. Younger age has previously been identified as a predictor for response to OA treatment (≥50% reduction of baseline AHI). 38 This finding is deemed to be a positive bias, indicating that the treatment effect for the entire group might be better than that reported for the patients in our study group. Further, female gender has been shown to predict treatment success in one report, 39 while it did not in another study. 38 Based on these reports, it is difficult to draw any conclusion as to whether the overrepresentation of women in the study group constituted a bias or not.

TA B L E 3
Logistic regression model for the odds of achieving success criterion 1 after OA treatment (AHI < 5), with success criterion 1 as dependent variable for the included 120 patients in the study. Dependent variable dichotomised as 0 = AHI ≥ 5, 1 = AHI < 5 The AP protocol aims for individualisation of OA treatment.
Standard protocol OA treatment suggests 'maximal comfortable protrusion', which often results in 50%-80% of maximal protrusive range as the target OA position for optimal treatment effect. 40,41 A systematic review investigating clinical practice in OA titration found that all included studies describing percentage of maximal protrusion reported a mean protrusion exceeding 85%. 10  In a recent systematic review and meta-analysis, several phenotypes predicting response to OA treatment were identified. 43 In our study, low BMI and absence of hypertension could predict treatment success in analogy to these findings. In addition, we found that increased total vertical elevation introduced by the OA could significantly predict treatment success. The total increase of vertical interincisal dimension of the OAs ranged from 4 to 12 mm, with a mean of 5.6 mm. A recent study identified patients with larger overbite as significantly more likely to achieve complete response of OA treatment, hypothesising that increased vertical dimensions for these patients normalises the intermaxillary space for the tongue in the oral cavity. 44 However, a general conclusion should not be The collapsibility of the upper airway, displayed as the reduction in total volume from the baseline to the collapse registration (modified Mueller manoeuvre) measured with AP, correlated significantly with OSA severity at baseline, which seems to be logical as patients with severe OSA probably have more collapsible upper airways than those with less severe disease. This has also been demonstrated in previous studies with AP. 31,32 In our study, larger minimum cross-sectional area and total volume maintained in the AOP, and larger AP total volume during collapse, showed indications of being predictors of reaching success criterion 1 (AHI < 5) with OA treatment in the unadjusted model. In the adjusted model, however, only total volume maintained during collapse registration at baseline remained a significant predictor for reaching success criterion 1. The latter finding indicates that a larger total volume maintained during collapse registration, indicating a less collapsible airway, increased the odds for reaching success criterion 1. Although it should be noted the odds ratio was only 1.07 for this prediction, so the clinical benefit of this should be interpreted with caution. Nevertheless, based on the above findings, the applicability of AP in OA treatment deserves further attention and more in-depth studies, which are currently underway in our research group.
The occurrence of reported side effects was rather significant in this study population, especially for dry mouth (43.9%). However, the reported side effects did not exceed that found in other studies. 45,46 When reporting side effects, patients were asked if they had experienced any of the listed side effects in association with OA treatment. Hence, some patients might have reported the side effects commonly occurring in the initial phase of OA treatment which usually disappear after a relatively short time. Further, other than pain/tension in the masticatory muscles, there was no significant relationship between side effects and reported usage of the OA, indicating that the side effects did not affect treatment compliance in this population.

| Limitations
The uncertainty of the reproducibility and validity of AP measures is a limitation of the technique. The results of the treatment indicate that AP improves OA treatment compared to standard OA treatment, when performed by two calibrated and experienced clinicians.
The current study was, however, not performed with a randomised design including a control group, and direct comparison with other studies is not possible due the differences in case selection, etc. The clinicians performing the AP and the positioning of OAs in this study were highly skilled in using the technique, and the results cannot directly be extrapolated in the hands of lesser experienced dentists.
This will require further investigations. There is also an uncertainty regarding the reproducibility of the modified Mueller manoeuvre, and the fact that the measures are performed awake and in a seated position. The rationale for using the modified Mueller manoeuvre when demonstrating collapsibility of the upper airway is that this technique was previously shown to demonstrate a reduction in pharyngeal cross-sectional area in OSA patients, which was not present in non-OSA subjects. 31,32 As the subjects in above-mentioned studies were seated and awake during AP measures, this was applied in our protocol as well (Appendix 1). The reproducibility of total lung capacity and residual volume has previously been shown to have ac-

| CON CLUS ION
The AP protocol applied in this study seems to provide a very high OA treatment success rate. Further research on the application of AP in OA treatment is needed in order to clarify its likely beneficial contribution to improving OA therapy.

ACK N OWLED G M ENTS
This study was funded by grants from the University of Bergen, Norway. The authors gratefully acknowledge Stein Atle Lie,

Biostatistician, Department of Clinical Dentistry, University of
Bergen, for his support in the statistical analyses. The authors also acknowledge Øyvind Goksøyr for his support in the statistical analyses. We also thank the employees of Tannhelsesenteret Lørenskog and Tannhelsesenteret Sogndal for their contribution in the data collection, and in that regard especially Henrik Silderen Halvorsen.

CO N FLI C T O F I NTE R E S T
We declare no conflict of interest.

AUTH O R CO NTR I B UTI O N S
All authors contributed to concept and study design. Authors PO and ULO contributed to data collection. All authors contributed to interpretation of results and critical revision of the manuscript, and provided final approval before submission.

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/joor.13134.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data can be made available upon request.

Acoustic pharyngometry -A new method to facilitate oral appliance therapy
All patients treated with oral appliances (OAs) at the two dental clinics at Tannhelsesenteret Lørenskog and Tannhelsesenteret   Sogndal between 2013 and 2018 were treated according to the following protocol, referred to as an acoustic pharyngometry (AP) protocol. This protocol is based upon the standard operating protocol for acoustic pharyngometry previously described by Kamal. 25 General information about acoustic pharyngometry examination

F I G U R E A 4
Jaw registration is performed in the AOP using the bite jig that provides the corresponding horizontal and vertical position, which is 4 mm interincisal distance and 2 mm protrusion from edge to edge in this example. A bite fork and handle are fitted into opposite ends of the bite jig, providing stability for the jaw registration material, which is added in a controlled manner while the patient is positioned in the AOP [Colour figure can be viewed at wileyonlinelibrary.com]