Protocol

This study complied with the Preferred Reporting Item for Systematic Review and Meta-Analyses (PRISMA) statement.⁹ The protocol was not registered.

Eligibility Criteria

The following selection criteria were applied for the review:

• Participants were aged >18 years and managed for snoring and/or OSA.

• The intervention was at least 5-year use of any OA.

• Outcome measures were dentoalveolar changes to OJ, OB, upper incisor inclination (Angle from upper incisor to cranial base [U1–SN]) and lower incisor inclination (Lower incisor [L1]–mandibular plane [MP]). Secondary outcomes were sagittal (Sella to Nasion to A point angle [SNA], Sella to Nasion to B point angle [SNB], A point to Nasion to B point angle [ANB]) and vertical (Mandibular plane to anterior cranial base angle [SN-GoGn]) skeletal changes as measured on lateral cephalograms.

• The study designs were randomized controlled trials (RCTs) cohort and before–after studies.

• Exclusion criteria were animal studies, studies not written in English, patients aged younger than 18 years, other reasons for treatment, tongue-retaining devices or combination therapy (eg, continuous positive airway pressure [CPAP] or surgery), less than the average follow-up of at least 5 years, and irrelevant outcomes (eg, efficacy or comparison of other treatment modalities).

Information Sources, Search Strategy, and Study Selection

The following five electronic databases were searched: MEDLINE (PubMed), MEDLINE (Ovid), Scopus, CINAHL, and Informit. Language restrictions were not applied. Unpublished literature was searched electronically by using ClinicalTrials.gov (www.clinicaltrials.gov) and the National Research Register (www.controlled-trials.com) with the terms “obstructive sleep apnea” and “oral appliance.” The date of the last search was January 20, 2021. Authors were contacted to identify unpublished or ongoing clinical trials and to clarify data as required. Reference lists of the included studies were screened for relevant studies.

Assessment of inclusion in the review, assessment of the risk of bias, and data extraction were performed independently and duplicated by two investigators (Drs Karadeniz and Lee) who were not blinded to the authors or the research results. Disagreements were resolved by discussion and consultation with a third author (Dr Flores-Mir).

A data extraction form was developed to record study design, participants, initial diagnosis, appliance used, number of years, and outcome data of interest.

Risk of Bias/Quality Assessment in Individual Studies

Studies were assessed for quality using different tools according to the classification of studies. The Cochrane Risk of Bias 2 tool was used to assess RCTs. The Non-Randomized Studies of Interventions (ROBINS-I) tool¹⁰ was used to assess cohort studies and uncontrolled before–after studies. The Risk-of-Bias Visualization¹¹ web application was used to produce figures to display the risk-of-bias assessments.

Summary Measures and Approach to Synthesis

The primary outcome was to determine the dentoskeletal malocclusion traits of OA therapy, and the secondary outcome was to identify additional malocclusion traits during the management of snoring/OSA in adults.

Risk of Bias Across Studies

The risk of bias across studies was applied for OB and OJ by using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) Pro software (GRADEpro GDT; available from gradepro.org¹²).

Meta-analysis

Statistical heterogeneity was assessed by inspecting a graphic display of the estimated treatment effects in conjunction with 95% confidence intervals (CIs). Heterogeneity was assessed using the I² statistic. Conventionally, I² values of 25%, 50%, and 75% indicate low, moderate, and high heterogeneity, respectively. All analyses were conducted using a random-effects model.

Study Selection and Characteristics

Database searching initially identified 1717 studies (Figure 1). After removing duplicates, screening abstracts, and assessing these studies for eligibility, 12^4,13–23 studies were included for qualitative synthesis. Of these 12 studies, one was an RCT,⁴ three were prospective cohorts,^15,18,19 three were retrospective cohorts,^20–22 and five uncontrolled before–after studies^{13,14,16,17,23} (Table 1).

View Figure

• Download as Powerpoint
• Download Figure

Table 1. Study Design; Number of Patients; Age; Sex; Diagnosis; BMI; Appliance Type; Titration Protocol; Mandibular Advancement; Years Used; Compliance; Method of Measuring Outcomes; and Changes to SNA, SNB, SN–GoGN, U1−SN°, L1−MP°, OJ, and OB^a,b

View Fullscreen

Table 1. Extended

View Fullscreen

Risk of Bias Within Studies

Overall, four of the cohort studies had a high risk of bias and two had a low risk, four of the before–after studies had a high risk and one had a low risk, and the RCT had a low risk of bias (Figure 2).

View Figure

• Download as Powerpoint
• Download Figure

Of the seven specified biases in the ROBINS-I quality assessment, confounding bias was moderate for most of the studies^{14,16–18,20–23} because of the inability of studies to control confounding variables such as the quality of remaining dentition or amount of periodontal support. Two studies had a serious risk of selection bias as participants with mesial occlusions were omitted.¹⁸ Three studies^20–22 did not indicate whether there was a consecutive inclusion of participants. Bias attributed to deviations from intended interventions related to patients crossing over between interventions (different appliances) resulted in serious bias.^18,20–22 Serious missing data bias was identified because the proportion for participants with Class III was significantly less than Class I and Class II Divisions 1 and 2.^20–22 Five studies^18–22 did not report whether assessors were blinded to the intervention received. In the end, four analyzed studies had a serious risk of bias, and two studies had a low risk of bias.

Results of Meta-analysis

A meta-analysis was conducted with only three studies^16,20,22 that used the same OA (Klearway; Space Maintainers Ltd, Vancouver, Canada) to exclude biomechanical differences. The pooled mean difference values and forest plots for OB and OJ were computed using the Revman version 5.3 (The Cochrane Collaboration) software. All other data were found not to be adequate as a result of the high methodological heterogeneity among the studies compared (eg, studies used either a different appliance or measured variables differently). The mean difference value of all included studies was calculated as posttreatment scores − pretreatment scores, with negative values indicating a decrease in the outcome of interest.

Of the three studies included in the meta-analysis, Chen et al.²⁰ showed the highest difference in OB and OJ from before to after treatment, which tended to be progressive. When considering the three studies (two studies with a mean follow-up of 7 years and one study of 11 years), the average decrease for OB was −1.94 mm (95% CI, −2.14 to −1.74; Figure 3A) and for OJ was −1.43 mm (95% CI, −1.66 to −1.20; Figure 3B). If only the two studies with an average follow-up close to 7 years were considered, in those the average decrease for OB was −2.09 mm (95% CI, −2.35 to −1.84; Figure 3C) and for OJ was −1.49 mm (95% CI, −1.77 to −1.21; Figure 3D).

View Figure

• Download as Powerpoint
• Download Figure

Risk of Bias Across Studies

The GRADE assessment revealed a very low quality of evidence across outcomes (Table 2).

Table 2. Risk of Bias Assessment Across Studies (GRADE)

View Fullscreen

Decreased OJ and OB

All of the studies investigated reported a significant reduction in incisor OJ and OB. The least amount of change was found with soft elastomeric nonadjustable monoblocs¹⁸ and small mandibular advancements of <6 mm compared with hard acrylic monoblocs. In addition, larger vertical displacement of the mandible (11 mm) was related to smaller OJ reduction. Long-term OB and OJ changes were close to 2 mm for OB and 1.5 mm for OJ. Comparing between 7 years and 11 years of follow-up (two studies with a mean follow-up of 7 years and one study of 11 years) did not convey clinically relevant differences (0.15 mm more for OB and 0.06 mm for OJ).

Dentoalveolar Changes According to the Initial Malocclusion

Patients with a large OJ at baseline normalized their dental occlusion during long-term treatment with OAs. Favorable dentoalveolar changes occurred in patients with Class II Division 1^15,17,18,22 or Class II Division 2 malocclusion,²² as forces from OAs caused proclination of mandibular incisors and retroclination of maxillary incisors. Patients with Class II Division 1 malocclusions with the most significant baseline OJ experienced the greatest benefit from OJ reduction.^17,18 Almeida et al.²² suggested that deeper OB correlated with a greater reduction in OB, and the greater the initial OJ, the greater the decrease in OJ. However, OAs may aggravate a preexisting Class I or Class III malocclusion.^15,17,22

Angular Changes in Incisor Position

Maxillary incisor retroclination (U1–SN) and mandibular incisor proclination (L1–MP) findings were consistent in all studies.^13–15,21 This suggests that changes in incisor inclination may contribute to OJ and OB reduction. However, cephalometric measurements showed a high standard deviation attributed to a wide range of individual responses to tooth movement. Changes in incisor angulation were probably attributed to anteriorly directed forces exerted by OAs on the mandibular teeth and posteriorly directed forces exerted on the maxillary teeth.

Progression of OA-Generated Malocclusion

Some authors suggested that changes in OJ and OB progressed over time.^4,16,18 Dental changes were irreversible and progressive with a rate of retroclination for upper incisors of −0.5°/year for nearly up to 2 decades of treatment.¹⁴ Proclination of the mandibular incisors was also progressive; however, the rate of proclination seemed to decline with prolonged follow-up and appeared to stop after approximately 19 years of treatment.

Skeletal Changes

The position of the maxilla relative to the cranial base (SNA°) did not change significantly over time. Almeida et al. found a significant difference in one of the study subgroups corresponding to the baseline Class I molar relationship, where a significant decrease of −0.8° was found.²¹ Two^14,15 of four studies found that the mandible in relation to the anterior cranial base (SNB°) significantly decreased in a range between −0.6° to −0.7°. A slight posterior rotation of the mandible was confirmed by an increase in MP angle (SN–GoGn) in all three studies^13,14,21 that investigated this variable. This may have been attributed to incisor interferences as the OJ was reduced.

Additional OA-Generated Malocclusion Traits

A Class III molar relationship with a mesial shift of the lower molars repositioned anteriorly in relation to the upper molars was a consistent observation.^4,17,20,23 However, the angle classification was reported only in a few studies.^4,19,20

Anterior crossbite ranged from 44%¹⁴ to 62%.¹⁶

Anterior open bite tendency was observed but not quantified.⁴ The decreased OB could be beneficial for patients who presented initially with deep OB, but for patients with normal or shallow baseline OB, reducing the OB might cause an open bite in the long run.²⁰

A posterior open bite of ≥2 mm was noted in 38% of patients.¹⁸ Pliska et al.¹⁶ observed a posterior open bite in 51% of the sample, whereas premolars in edge-to-edge or open-bite relationships were found in 14% and molars in 11%.²² Others recorded posterior open bite occurrence as “yes/no” visually on dental casts.¹⁹

Some studies^16,20,22 showed significant increases in the number of interproximal open spaces in the maxilla and the mandible, expansion of the lower arch measured by increases in mandibular intercanine and intermolar distances, and a decrease in mandibular crowding. These mandibular changes were associated with lower incisor proclination, possibly causing an enlargement of the mandibular arch length.²² In contrast, Marklund found incisor crowding.¹⁷

OA Selection

Several commercial appliances, including the Klearway^16,20 and Thornton Adjustable Positioner (TAP) appliances⁴ (TAP type 1; Airway Management Inc., Dallas, Tex) were used. However, a combination of appliances was commonly reported, for example, used Klearway but started with a different appliance^21,22; others used Klearway or Somnodent (SomnoMed, Plano, TX)^13,14 or a combination of monobloc, Somnodent, and Narval (ResMed, Didcot, UK.).¹⁷ Similarly, nonadjustable hard acrylic monoblocs were continuously replaced with soft elastomeric nonadjustable devices and/or adjustable OAs at the 16-year follow-up.²³ Others used a nonadjustable monobloc appliance^15,19 made of heat-cured methyl methacrylate resin or soft elastomeric and hard acrylic monobloc.¹⁸

The amount of vertical opening of the appliance was not consistently reported.^4,19

Adherence and Dentoalveolar Adverse Effects

Self-reported adherence to OA wear was the predominant method used in most studies.¹⁹ Highly compliant patients experienced the greatest reduction in OJ (−3.5 mm) and OB (−2.9mm), with an average treatment use of 7 days/week, 7.8 hours/night; however, those patients, on average, did not grade the dental adverse effects as negative or disturbing.⁴ Likewise, the statistically significant changes in occlusion found by Chen et al.²⁰ did not result in treatment suspension or cessation.

Smaller reductions in OJ, OB, and the posterior open bite were more common in infrequent users than frequent users (≥50% of nights/week for 5 years).¹⁸

Mandibular Advancement Protocols

Studies using Klearway^14,16,21,22 reported more than two-thirds of the maximum protrusion (6–10 mm),²¹ whereas the TAP⁴ was set at 50% of maximum protrusion. Others used 75%^15,19 of the maximum range of protrusion and at least 5 mm of protrusion. Later adjustments were made if needed until symptoms abated or until further protrusion of the mandible resulted in discomfort; however, additional mandibular advancement measurements have not been reported. On the other hand, <6 mm advancements of the mandible with monoblocs resulted in small OJ reductions after 5 years.¹⁸

Limitations

Most studies did not differentiate management of snoring from diagnosed OSA and included patients who were able to adapt to changes in their tooth positions or those who benefited from the changes. The use of objective adherence monitoring systems would be beneficial instead of self-reported questionnaires to reduce bias.

It would be reasonable to evaluate risks associated with OA-generated malocclusions versus their effectiveness/efficacy in the long-term treatment of OSA and possibly consider treatment discontinuation. When there is no superior treatment option based on current knowledge, the principle of autonomy dictates that patients have a right to participate in decision making.²⁴

Still controversial, though, mouth breathing remains an environmental factor in the development of malocclusion. With the lack of reporting on the mode of respiration, it is impossible to discuss differences in the influence of mouth breathing in the development of OA-generated malocclusion in adults. The natural movement of teeth in an older population is expected as an increase in mandibular anterior crowding continues throughout middle age, that is, beyond 60 years of age.²⁵ However, OB and OJ remain constant from 20 to 55 years of age.^25,26

Finally, although a few meta-analyses were possible in the big picture, they included studies with moderate to high risk of bias. In this regard, it could be argued that not conducting meta-analyses was also a reasonable option.

Overall, the body of evidence ranged from very low to low quality as assessed with GRADEpro.