Surgery for obstructive sleep apnoea in adults

Abstract

Background

Obstructive sleep apnoea/hypopnoea syndrome(OSAHS) is the periodic reduction or cessation of airflow during sleep. The syndrome is associated with loud snoring, disrupted sleep and observed apnoeas. Surgery for obstructive sleep apnoea/hypopnoea syndrome aims to alleviate symptoms of daytime sleepiness, improve quality of life, and reduce the signs of sleep apnoea recorded by polysomnography.

Objectives

The objective of this review was to assess the effects of any type of surgery for the treatment of the symptoms of obstructive sleep apnoea/hypopnoea syndrome in adults.

Search methods

We searched the Cochrane Airways Group Specialised Register and reference lists of articles. We contacted experts in the field, research dissemination bodies and other Cochrane Review Groups. Searches were current as of July 2008.

Selection criteria

Randomised trials comparing any surgical intervention for obstructive sleep apnoea/hypopnoea syndrome with other surgical or non‐surgical interventions or no intervention.

Data collection and analysis

Two reviewers assessed electronic literature search results for possibly relevant studies. Characteristics and data from studies meeting the inclusion criteria were extracted and entered into RevMan 5.

Main results

Twelve studies (709 participants) of mixed risk of bias met the inclusion criteria. Data from ten studies were eligible for assessment in the review. The participants recruited to the studies had mixed levels of AHI, but tended to suffer from moderate daytime sleepiness where this was measured. No data could be pooled. Uvulopalatopharyngoplasty (UPPP) versus conservative management (one trial): An un‐validated symptom score showed intermittent significant differences over a 12‐month follow‐up period. No differences in Polysomnography (PSG) outcomes were reported. Laser‐assisted uvulopalatoplasty (LAUP) versus conservative management/placebo (two trials): One study recruited a mixed population and separate data could not be obtained for this trial. In the other study no significant differences in Epworth scores or quality of life reported. A significant difference in favour of LAUP was reported in terms of apnoea hypopnoea index (AHI) and frequency and intensity of snoring. Palatal implants versus placebo (one trial): Symptoms and AHI were lower with palatal implants. LAUP versus bipolar radiofrequency volumetric tissue reduction (one trial): Within‐treatment group differences were significant for symptoms and AHI, but the between treatment group differences were not available to assess whether indirect inferences could be made regarding the effects of either treatment. UPPP versus oral appliance (OA) (one trial): AHI was significantly lower with OA therapy than with UPPP. No significant differences were observed in quality of life. UPPP versus lateral pharyngoplasty (lateral PP) (one trial): No significant difference in Epworth scores, but a greater reduction in AHI with lateral PP was reported. Expansion sphincter pharyngoplasty (one trial): Both interventions reduced AHI although statistical analysis on the difference between treatments was not reported. Tongue advancement (mandibular osteotomy) + PPP versus tongue suspension + PPP (one trial): There was a significant reduction in symptoms in both groups, but no significant difference between the two surgery types. Complications reported with all surgical techniques included nasal regurgitation, pain and bleeding. These did not persist in the long term. An additional study assessed the effects of four different techniques. No data were available on between group comparisons. Multilevel temperature‐controlled radiofrequency tissue ablation (TCRFTA) versus sham placebo and CPAP (one trial): There was an improvement in primary and secondary outcomes of TCRFTA over sham placebo and but no difference in symptomatic improvement when compared with CPAP. Radiofrequency assisted uvulopalatoplasty technique versus channeling technique (one trial): Dysphagia scores were lower in the channeling group immediately post‐operation, but at four months the difference was no longer significant. Snoring scores did not differ at long‐term follow‐up. The number of participants achieving AHI lower than 10 at four months was slightly higher with RAUP than channeling technique.

Authors' conclusions

There are now a small number of trials assessing different surgical techniques with inactive and active control treatments. The studies assembled in the review have failed to demonstrate consistent effects in favour of surgery and do not provide convincing evidence to support its use in sleep apnoea/hypopnoea syndrome. Short‐term outcomes are unlikely to consistently identify suitable candidates for surgery. Long‐term follow‐up of patients who undergo surgical correction of upper airway obstruction is required. This would help to determine whether surgery is a curative intervention, or whether there is a tendency for the signs and symptoms of sleep apnoea to re‐assert themselves, prompting patients to seek further treatment for sleep apnoea.

Plain language summary

Surgery for obstructive sleep apnoea/hypopnoea syndrome

Surgery for obstructive sleep apnoea/hypopnoea syndrome aims to relieve obstruction by increasing the size of the airway in the throat, bypassing the airway or removing a lesion. A limited number of trials assessing diverse surgical techniques were identified. There were inconsistent effects reported across the trials. The available evidence from these small studies does not currently support the widespread use of surgery in people with mild to moderate daytime sleepiness associated with sleep apnoea.

Background

Obstructive sleep apnoea/hypopnoea syndrome (OSAHS) is the periodic reduction (hypopnoea) or cessation (apnoea) of airflow during sleep that results from pharyngeal narrowing or collapse. The syndrome is associated with loud snoring, disrupted sleep and observed apnoeas. The condition is often brought to the attention of the health care professionals through the concern of bed partners or parents, who may witness apnoeic episodes followed by choking and periods of loud snoring. This may have led to disharmony in the relationship.

While the epidemiology of OSAHS is not completely clear, it is thought to be a relatively common problem, affecting 2‐4% of males and 1‐2% of females in middle age (approximately 30 to 69 years) (Jennum 1992; Stradling 1995a; Young 1993). Risk factors may include obesity, male gender, retrognathia, hypothyroidism, nasal obstruction and evening alcohol ingestion (ASDA 1996). In children, the epidemiology of OSAHS is poorly described with few estimates of prevalence (Boudewyns 1995). However the most common cause of paediatric OSAHS is adeno tonsillar hyperplasia (Potsic 1987).

The main symptoms of OSAHS are loud snoring, daytime sleepiness (Guilleminault 1993; Yamadera 1995), and the associated cognitive impairment (Ferguson 1995). There is also some evidence that OSAHS may be associated with road traffic accidents (Findley 1988; RCP 1993). Therefore in some countries, including the UK, heavy goods and public transport vehicle drivers may have their licenses withdrawn if they have untreated OSAHS (Taylor 1995).

Conservative treatments include weight loss, modification of the patient's sleep position, medications to relieve nasal obstruction and avoidance of evening alcohol and hypnotics (ASDA 1996), although the evidence for these is unclear. The most commonly used and recommended treatment modality is Continuous Positive Airways Pressure (CPAP) during sleep (SIGN 2003). Oral appliances are also used (Lim 2006).

Surgical treatments for OSAHS aim to relieve the obstruction by increasing the surface area of the airway, to bypass the pharyngeal airway, or to remove a specific pathological lesion. The principal interventions are briefly described by the ASDA 1996 and can be considered as:

1. Tracheostomy (which bypasses the pharyngeal airway)

2. Uvulopalatopharyngoplasty (UPPP) which increases the area of the retro palatal airway by resection of the free edge of the uvula and soft palate, and may be combined with tonsillectomy‐ a modification is laser‐assisted uvulopalatoplasty (LAUP)

3. Tonsillectomy and adenoidectomy ‐ accepted treatment of OSAHS in children

4. Inferior sagittal mandibular osteotomy and genioglossal advancement with hyoid myotomy and suspension (GAHM) aim to create an enlarged retrolingual airway

5. Laser midline glossectomy and lingualplasty also create an enlarged retrolingual airway

6. Maxillo‐mandibular osteotomy and advancement enlarges both retrolingual and retropalatal airway

7. Epiglottoplasty for selected cases of laryngomalacia

8. Removal of local specific obstructing pathological lesions

The place of surgery in the treatment of OSAHS and the relative effectiveness of different interventions is controversial. Most studies recommending a particular surgery are based on evidence from case series. Previous reviews of the surgical therapy for OSAHS are generally narrative, summarising the evidence provided by case series and uncontrolled observational studies (Banerjee 1995; Hochban 1995; Rodenstein 1992). There is one systematic literature search with a meta‐analysis of results, but this was restricted to articles published in English and found on MEDLINE, the search criteria were not clearly stated and it was based on evidence from non‐randomised studies (Sher 1996). A systematic review of randomised controlled trials would summarise the best evidence of the value of surgery in the treatment of OSAHS.

Objectives

To review the efficacy of surgery in the treatment of obstructive sleep apnoea.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials (RCTs).

Types of participants

Participants in a trial with a diagnosis of OSAHS are eligible for inclusion, using the criteria of more than five apnoeas or hypopnoeas per hour of sleep (the Apnoea / Hypopnoea Index or AHI). We will exclude children from this review as data from surgery in this population has been studied elsewhere (Lim 2006).

Types of interventions

Treatment group: any specific surgical interventions for OSAHS (see above). 
 Control group: other surgical or non‐surgical intervention, or no intervention.

Types of outcome measures

Primary outcomes

1. Epworth Sleepiness Score (ESS)/symptoms of sleepiness

2. Apnoea Hypopnoea Index (AHI)

Secondary outcomes

1. Presence or absence of oxygen desaturation

2. Average number of oxygen desaturations per hour of sleep

3. Average duration of oxygen desaturations

4. Quality of Life (using a recognised scale)

5. Complications of surgery

6. Withdrawals

7. Postoperative morbidity

8. Postoperative mortality

9. One year mortality

Search methods for identification of studies

Electronic searches

Trials were identified using the Cochrane Airways Group Specialised Register of trials, which is derived from systematic searches of bibliographic databases including the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE and CINAHL, and handsearching of respiratory journals and meeting abstracts. All records in the Specialised Register coded as 'sleep apnea' were searched using the following terms:

((surg*) and (palate* or uvula or pharyn* or maxillofacial or maxillo‐facial or upper‐airway* or "upper airway*" or nasal)) or (uvulopalatopharyngoplasty or uppp or uvpp or upp or palatoplasty or pharyngoplasty or palatopharyngoplasty or ppp or uvulopalatoplasty or laup or tracheostomy or mini‐tracheostomy or "genioglossal advancement" or "maxillomandibular advancement" or "maxillomandibular osteotomy" or "maxillary advancement" or "maxillary osteotomy" or "mandibular advancement" or "mandibular osteotomy" or "intrapalatine resection" or "tongue volume reduction" or "inferior sagittal osteotomy" or "hyoid bone suspension" or "hyoid suspension" or "hyoid myotomy" or septoplasty or polypectomy or adenoidectomy or tonsillectomy or adenotonsillectomy)

The most recent search was conducted in July 2008.

Searching other resources

The reference lists of relevant review articles, and references of all identified RCTs were checked for more possible RCTs. The NHS Centre for Reviews and Dissemination, the National Health Technology Assessment Programme, the NHS National Research Register and the Aggressive Research Intelligence Facility were contacted. Local and national sleep laboratories and experts in the fields of sleep and respiratory medicine and ENT surgery were contacted during the development of the protocol and search strategy, and to identify potentially relevant trials. Other Cochrane Review Groups were also contacted to identify citations found by handsearching of journals (such as surgical journals).

Data collection and analysis

Selection of studies

Two independent assessors assessed the titles and abstracts of articles using the criteria for inclusion. Full text reports were obtained if either or both of the reviewers believed the citation to be of possible relevance. Disagreement on inclusion of full text reports was resolved through consensus between the reviewers.

Data extraction and management

Data were extracted and checked by two reviewers (SS and TJL). Data were entered in to Review Manager software.

Data on treatment efficacy were recorded at the longest period of time post intervention reported in the studies. Sensitivity analysis based on length of study will be performed in future versions of the review.

Assessment of risk of bias in included studies

The methodological quality of eligible controlled trials was assessed by two independent assessors based on the degree of allocation concealment, and the risk of bias using criteria recommended in chapter 8 of the Cochrane Handbook. We have restricted assessment to randomisation, blinding and follow‐up of participants in the studies.

Data synthesis

No planned subgroup analyses were published in the first version of the review. Studies were pooled according to whether the surgical intervention were laser assisted (LA) or general intervention surgery. For the update, subgroup analyses were based upon low medium and high mean baseline AHIs (low: 5‐15; medium: 15‐30 and high: >30). Post hoc sensitivity analyses were to be undertaken where the amount of statistical heterogeneity in the assembled studies exceeded that expected by chance as measured by the I2 (>0%).

Results

Description of studies

Results of the search

For details on search history, please see Table 1. Searches up to July 2008 identified 16 citations. Three of these were retrieved for further scrutiny, and two met the eligibility criteria of the review (Friedman 2008; Pang 2007). Twelve trials met the inclusion criteria of the review.

1. Search history.

Search dates	References
Initial version of the review	A total of 594 articles were identified through electronic searching, 44 through correspondence and a further 28 through hand‐searching reference lists. Of these, 656 articles were excluded because of the following non‐inclusive reasons: did not focus on sleep apnoea (n=134); did not look at surgical interventions (n=328); obviously not an RCT (overviews, letters, case series etc.) (n=194). Ten articles, for which full text reports were obtained, were assessed for inclusion by both reviewers. All of these articles were excluded, five because they were descriptions of a surgical procedure, reviews or letters, four because they were not intervention studies and one was not about sleep apnoea. There was no disagreement on final inclusion/exclusion of studies.
Update ‐ July 2004	An update search was conducted in July 2004. 110 references identified; 15 retrieved for further scrutiny. Seven studies met the inclusion criteria (Lojander 1996; Tegelberg 1999; Ferguson 2003; Cahali 2003; Naya 2002; Woodson 2003, Thomas 2003). One non‐English language article is awaiting assessment. Seven studies failed to meet the inclusion criteria of the review
Update ‐ July 2005	An update search was conducted in July 2005. 20 references were identified and two studies were retrieved for additional scrutiny. Of these, one trial met the inclusion criteria (Larrosa 2004) and another pertains to an ongoing study (Tommi 2004). Both studies have a sham surgical procedure as a control arm.
Update ‐ July 2006	A total of 10 references were identified from electronic literature searches. Three of these were obtained as full‐text papers. Two studies failed to meet the entry criteria of the review, and a further study is awaiting assessment. One new trial of 150 participants met the entry criteria of the review (Atef 2005).
Update ‐ July 2007	A total of 11 references were identified from electronic literature searches. One was retrieved as full‐text and met the entry criteria of the review. This study added data from 40 participants (Bassiouny 2007).

Included studies

Data from one study (Larrosa 2004) could not be incorporated in to the review as the study included a number of non‐apneic snorers. We have not been able to obtain discreet data for those with AHI >5 in the review. Baseline characteristics of this trial are described in Characteristics of included studies.

Study design

All studies were randomised, parallel group trials. Lojander 1996 consisted of two studies, one comparing CPAP with conservative treatment, the second comparing UPPP with conservative management. We have extracted data from the surgical study.

Participants

Sample sizes

A total of 709 participants were recruited in the studies. Trial size was generally small, with samples ranging from 17 to 150.

OSAHS severity and symptoms

Overall, participants suffered from mild to severe sleep apnoea in the trials. Using baseline AHI as an indicator of severity, Ferguson 2003 and Atef 2005; Friedman 2008 recruited participants with mild and mild to moderate OSAHS. Mean AHI in Ferguson 2003 was below 20 per hour for both groups, whose symptoms were relatively mild (ESS <11 at baseline). In Friedman 2008 and Atef 2005 mean baseline AHI was between 23 and 27. ESS was 12 in Friedman 2008. Participants in Lojander 1996 had moderate‐severe OSAHS (oxygen desaturation index (ODI4) >20. Tegelberg 1999 recruited participants with mild and moderate OSAHS (AI: 5‐25), as did Bassiouny 2007 (AHI 10‐30). Cahali 2004 recruited participants with AHI greater than 10, who had refused or failed to tolerate CPAP. The participants in Woodson 2003 were moderate OSAHS sufferers, with baseline AHIs between 5 and 40. Thomas 2003 recruited participants with moderate to severe OSAHS (AHI >20), and moderate symptoms of daytime sleepiness (mean ESS: 12.7). Pang 2007 recruited participants with significant AHI and who were intolerant of CPAP therapy.

Airway obstruction inclusion criteria

Participants in Cahali 2004 and Friedman 2008 were required to have soft palates. Woodson 2003 excluded participants if they had tonsillar hypertrophy or nasal/supraglottic obstruction on examination. In Thomas 2003 entry criteria stipulated Fujita type II airway collapse and multilevel obstruction. Pang 2007 recruited participants with type I Fujita retropalatal obstruction and lateral pharyngeal wall collapse. Tegelberg 1999 excluded participants with significant nasal obstruction. Lojander 1996 considered patients for surgery if they had 50% obstruction at palatal level in the Mueller's manoeuvre, but less than 50% obstruction at the epiglottic level. Atef 2005 excluded participants with nasal or retrolingual obstruction, large obstructing tonsils and prior palatal surgery. Bassiouny 2007; Ferguson 2003 and Naya 2002 did not stipulate airway obstruction as an entry criterion.

Intervention

Type of procedure and comparator

Four studies assessed the effects of UPPP (Cahali 2004; Lojander 1996; Tegelberg 1999; Pang 2007). In addition Lojander 1996 performed mandibular osteotomy with hyoid myotomy if the patient had a narrow posterior airspace, an inferiorly positioned hyoid and a sharp sella‐nose‐mandibular angle. Lojander 1996 undertook a comparison of surgical treatment with conservative management, and Tegelberg 1999 compared a surgical intervention with an oral appliance. Cahali 2004 compared the relative effects of UPPP with lateral PP, and Pang 2007 compared expansion sphincter pharyngoplasty with UPPP. Naya 2002 conducted a randomised study between four surgical techniques (partial palatal resection (Quesada technique); UPPP (Fujita); UPPP (Simmons) and UPPP (Fairbanks)). However, no 'between group' comparisons were made as all the analyses were undertaken comparing the different surgical techniques with baseline. Ferguson 2003 and Atef 2005 assessed the effects of laser assisted uvulopalatoplasty (LAUP). However, Atef 2005 compared this procedure with bipolar radiofrequency volumetric tissue reduction (BRVTR) and Ferguson 2003 compared it with conservative management. In Atef 2005 the participants were randomised at two levels; to either surgical technique, and within each treatment group to between one and five treatment sessions. Bassiouny 2007 assessed radiofrequency assisted uvulopalatoplasty and submucosal channeling of the palate. Friedman 2008 assessed the effects of palatal implants to stiffen soft palatal tissue. Thomas 2003 compared palatopharyngoplasty combined with either tongue advancement (mandibular osteotomy) or tongue suspension. Woodson 2003 randomised participants to either temperature‐controlled radiofrequency tissue ablation (TCRFTA), CPAP, or sham TCRFTA. Active TCRFTA involved tongue (five sessions) and palate (two sessions) sites. Sham TCRFTA consisted of a maximum of three sessions of energy delivery to tongue sites with a blocking control box on the radiofrequency generator set to off.

Duration of follow‐up

Lojander 1996 assessed the effects of treatment at three and 12 months. Tegelberg 1999 assessed participants up to four years. Ferguson 2003 offered repeat surgery and assessed the effects of treatment three months after final surgical procedure. In the control group, values were recorded at 6 months after baseline. Atef 2005 collected outcome data at three and 18 months post‐operatively. Bassiouny 2007 assessed outcome postoperatively at three days, and also reported data at four months. Cahali 2004 and Pang 2007 recorded outcomes from 6 months postoperatively. End of treatment values were recorded at 16 weeks in Thomas 2003 and at 12 weeks in Friedman 2008. In Woodson 2003 the CPAP group were assessed after 8 weeks of treatment, whereas assessment in the TCRFTA and placebo groups occurred after the final treatment session. Based on the number of treatment sessions in the surgery and placebo groups the mean treatment duration was 4.5 months in the TCRFTA group and three months in the placebo group.

Outcomes

All studies measured polysomnographic (PSG) variables with the exception of Bassiouny 2007 and Thomas 2003. Atef 2005; Tegelberg 1999; Ferguson 2003; Friedman 2008; Cahali 2004; Woodson 2003; Naya 2002; Pang 2007 measured AI, AHI or ODI₄. Lojander 1996 measured ODI₄ and ODI₁₀. ESS was measured in Cahali 2004; Ferguson 2003; Friedman 2008; Woodson 2003; Thomas 2003. Lojander 1996 made additional measurement of symptoms on an in‐house visual analogue scale. Ferguson 2003 and Bassiouny 2007 reported data on severity and frequency of snoring, assessed by a bed partner or household member. Quality of life was measured in Woodson 2003 and Ferguson 2003. Withdrawal rates were recorded by all studies.

Risk of bias in included studies

Where reported, the methods of randomisation indicated that the studies were conducted appropriately. Blinding was not possible for some of the comparisons undertaken (e.g. surgery versus conservative management or oral appliance therapy). Three studies undertook a single‐blind comparison with a sham surgical procedure that was conducted with a local anaesthetic (Bassiouny 2007; Larrosa 2004; Woodson 2003). Incomplete data were analysed in all the studies, which may affect the reliability of some of the findings. The lack of information either in trial reports or following requests, means that we remain uncertain as to how study design protects the findings from bias (Figure 1).

1.

Methodological quality summary: review authors' judgements about each methodological quality item for each included study.

Effects of interventions

Surgical interventions versus conservative management or placebo

Temperature controlled radiofrequency tissue ablation (TCRFTA) versus placebo

(Woodson 2003)

AHI:No significant difference (placebo ‐1.8; TCRFTA: ‐4.5, P = 0.34).

ESS:There was a significant greater reduction in symptoms in favour of TCRFTA compared with placebo (mean difference in change: ‐1.1, P = 0.005; absolute scores at end of treatment: TCRFTA: 9.8, placebo: 10.6) .

Quality of life: There was a significantly greater improvement on the Functional Outcomes of Sleep Questionnaire (FOSQ) in favour of TCRFTA of 0.8 (P = 0.005; absolute score at end of treatment TCRFTA: 17.7; placebo: 17. 
 
 Complications: There were no significant differences between active and sham TCRFTA in terms of frequency and severity of complications post‐operatively. Haematomas occurred three times in each group. There was one ulceration in the TCRFTA group, and no infections occurred. There was no difference in post‐operative pain or swallowing difficulty in both groups. Swallowing difficulty increased in both sham and active treatment groups after 1 week before abating to baseline levels. These data are from only a limited number of participants and the true rate of complications given the incomplete numbers should not be assumed to be represented by the values in the published paper.

Laser‐assisted uvulopalatoplasty (LAUP) versus conservative management

(Ferguson 2003) 
 
 AHI: There was a significant difference in final AHI values in favour of LAUP (P = 0.04). There was a 21% reduction in mean AHI after LAUP (P=0.03).LAUP of 8 (P = 0.04). Mean AHI in the control group at the end of treatment was 22 events/h.

Symptoms: No significant difference in ESS. When symptoms were reported as dichotomous data, there was no difference between surgery and conservative management in participants who reported un refreshing sleep and excessive daytime sleepiness.

Frequency and intensity of snoring was significantly improved in LAUP group compared with baseline (intensity: LAUP: 4.8 versus control 8.5, P < 0.0001; frequency: LAUP: 5.5 versus control: 8.5, P < 0.005).

Quality of life: There was no significant difference between LAUP and control in Calgary Sleep Apnoea Quality of Life Index (SAQLI) scores. 
 
 Satisfaction: Only data were recorded for the surgery group. Eleven/21 participants were satisfied with LAUP. Thirteen/21 had little or no difficulty undergoing surgery. 
 
 Complications: Moderate/severe pain immediately after the procedure (n = 17); dysphagia (n = 4); bleeding (n = 9); nasal regurgitation (n = 5); infections (n = 4); temporary change in vocal quality (n= 1).

UPPP versus conservative management

(Lojander 1996) 
 
 Oxygen desaturation indices: No significant differences were observed between the two groups in ODI₄ and ODI₁₀ at 12 months.

Symptoms of daytime sleepiness (VAS): A significant difference was observed in favour of surgery over conservative management at 3 and 12 months (3 month: surgery: 12 (0‐17) versus control: 56 (11‐100), P<0.001; 12 month: surgery: 27 (5‐86), versus control: 72 (33‐86), P<0.01).

Complications: Reported complications were 22% in the surgery group. Complications included removal of infected material (N =2), tracheostomy (N = 1) and dysphagia (N = 2). In the surgery group one participant had a MI, and one participant had a transient ischaemic attack.

Withdrawals/loss to follow‐up:Five patients refused to come to the follow up visits. In the conservative management group one patient needed CPAP, and three underwent surgery due to worsening of symptoms. Four participants in the control group and two participants in the treatment group withdrew from the study.

Palatal implants versus placebo

(Friedman 2008)

AHI: Mean change in AHI was significantly greater when compared with placebo (8.8, P < 0.0001).

Symptoms of sleepiness: ESS scores improved more with implants than placebo (1.9, P = 0.0002).

Quality of life scores (SF‐36): Surgery led to better quality of life when compared with placebo (11.9, P <0.0001).

Withdrawals/loss to follow‐up: Two and five participants were withdrawn from the study from the implant and placebo groups respectively.

Complications: Two participants were reported to have had extrusions of implants (their group assignment was not described). In both cases reinsertion was undertaken.

Surgical interventions versus active non‐surgical interventions

UPPP versus oral appliance therapy

(Tegelberg 1999)

AHI: There was a significant difference in favour of oral appliance therapy compared with surgical treatment alone at one and four years follow up OA: 4.5 (2.6) & 7.2 (SD 2.6) versus surgery: 9.8(2.5)& 14.2 (SD 3.4), P < 0.001.

Level of obstruction was assessed by fibreoptic pharyngoscopy and different types of obstruction were documented. Regardless of treatment group success rate did not differ in the different obstruction types 
 
 Quality of Life (MSE‐P questionnaire ‐ VAS: low values indicate better health): Data were reported for one year after intervention. No significant differences were reported between surgery and OA therapy for vitality and sleep. There was a significant difference in favour of surgery reported for contentment: OAs: 33.7 versus surgery: 27.4, P < 0.05.

Complications: In the group who underwent surgical treatment, three participants had fibrotic narrowing without symptoms by four years post‐intervention. Nasopharyngeal regurgitation of fluid (8%) and dysphagia (10%) of participants were reported. In the group treated with oral appliances one patient had recurrent apthous ulcers due to an allergic reaction to the acrylic polymer, and two patients were not comfortable. Five appliances had minor defects, seven oral appliances required repair, and repeated adjustments were needed in one participant. 
 
 Withdrawals/Loss to follow‐up: More participants withdrew or were lost to follow‐up in the oral appliance group compared with the group treated with surgery (12/49 versus 3/46). This finding should be treated with caution as OA therapy is ongoing, whereas the surgical intervention in this study was one‐off. Furthermore the reasons given for withdrawal could only be related to therapy in five participants.

TCRFTA versus CPAP

(Woodson 2003) 
 
 AHI: AHI was lower in participants treated with CPAP compared with those randomised to TCRFTA (4.6 versus 16.8) representing a significant difference in the change from baseline between these two groups in favour of CPAP of 10.7 events/h (P = 0.004). 
 
 ESS: There was no significant difference between surgery and CPAP groups. Both groups improved significantly from baseline.

Quality of life (FOSQ): No significant difference. Both groups improved significantly from baseline. 
 
 Complications: The complications for the surgery group are reported above. Side effects occurred in 20/28 participants analysed on CPAP. These included nasal symptoms associated with CPAP use, inconvenience, sleep interruption, air mechanics and skin or eye symptoms. These were not serious in nature. 
 
 Withdrawals/loss to follow‐up: The attrition rate in the TCRFTA group was 7/30 compared with 4/30 in the CPAP group. Reasons for discontinuation with treatment were not reported.

Different surgical interventions compared

UPPP versus lateral PP

(Cahali 2004)

AHI: No significant difference was reported between groups in terms of absolute scores. Both interventions reduced AHI. Whilst the magnitude of the improvement was significant in the lateral PP group, mean baseline AHI was higher in this group (41.6 versus 34.6) . At the end of treatment mean AHI was 15.5 in the lateral PP group and 30 in the UPPP group.

ESS: There were significant improvements in both groups (median reductions of 11 and 10 in lateral PP and UPPP groups respectively), but there was no significant difference between these two values.

Complications:There were four reported cases of nasal reflux in both treatment groups.

Withdrawals:Two participants were lost to follow‐up from the UPPP group.

Laser‐assisted uvulopalatoplasty (LAUP) versus bipolar radiofrequency volumetric tissue reduction (BRVTR)

(Atef 2005)

AHI: Data were reported for each set of subgroups (i.e. one to five subgroups) for both treatment groups. Whilst significant improvements were reported in the subgroups of those receiving three or more treatment sessions for BRV, and in all subgroups of the LAUP group at three months, no between treatment group data were presented. The changes from baseline in the treatment groups persisted, with only the group treated with one session of LAUP showing a non‐significant change from baseline.

Study defined success (reduction of AHI by 50% and AHI <20 events/hr): Data were presented as %, but it was not clear whether the proportion of those improving represented were based on those randomised, or those participants presenting for follow‐up assessment at 18 months only. The number of BRVTR treated participants showing an improvement in AHI at 18 months were 0, 9, 55, 60 and 82.2% in the five groups respectively, and 36.3, 48.2, 60, 70 and 64.2% in the five treatment groups of the LAUP treated participants.

Symptoms: Mean VAS scores in the LAUP treated participants varied across the five subgroups assessed from 3.5 to 8.5 at 18 months. For BRVTR the mean VAS scores were between 3 and 7. No statistical tests were reported on whether these data differed between or within each treatment group.

Tongue advancement (mandibular osteotomy) + PPP versus tongue suspension + PPP

(Thomas 2003)

AHI: No data were presented on AHI

ESS: There were significant reductions in both surgical treatment groups. No between group comparison was undertaken.

Partial palatal resection (Quesada technique); UPPP (Fujita); UPPP (Simmons); UPPP (Fairbanks) compared to one another.

Naya 2002 reported a significant improvement from baseline in AHI of 25, across all treatment groups (P = 0.0001) and no difference between the four surgical techniques analysed (P = 0.085). No additional clinical outcomes were assessed.

Radiofrequency assisted uvulopalatoplasty (RAUP) versus submucosal channeling of the palate

(Bassiouny 2007)

AHI: At four months both techniques reduced AHI significantly compared with baseline. The number of participants with AHI below 10 was slightly higher in the RAUP (40% versus 25%, P = 0.05).

ESS: No assessment of sleepiness was undertaken. 
 
 Pain and dysphagia: There was no significant difference between the group means on VAS scores at four months.

Expansion sphincter pharyngoplasty (ESP) versus UPPP

(Pang 2007)

AHI: ESP led to lower AHI at 6 month polysomnography compared with UPPP, although statistical significance was tested with groups for changes from baseline (P = 0.005 for both groups).

Discussion

The initial version of this systematic review did not identify any randomised trials to support surgical interventions in the treatment of sleep apnoea. This updated version incorporates findings from 12 randomised trials of mixed quality, recruiting 709 participants with predominantly mild to moderate symptoms of sleep apnoea. Each study undertook different comparisons, precluding summary estimates of data. Individually, the findings from the studies included in the review do not support the widespread use of surgical intervention as a means of improving sleep quality over other therapeutic options currently available. The response to therapy could vary depending on the nature of airway obstruction in individual patients.

The concept of control in sleep apnoea trials has been contentious (Wright 1997). This is due to the predominantly physical nature of treatments used in management of OSAHS, including surgery. However, there has been a recent trend to conduct single blind trials in surgical trials controlled with a sham procedure (Larrosa 2004; Tommi 2004; Woodson 2003; Friedman 2008). A sham procedure is vital in establishing an estimate of efficacy whereby the effects of an active surgical intervention can be controlled. Adequate control with active therapies such as conservative management, drug treatment, CPAP or oral appliances relies on the continued acceptance of these interventions. The advantage of a sham surgical procedure is its control for the one‐off or short‐term nature of surgery. Given the implications of leaving significant populations who suffer from debilitating daytime sleepiness associated with sleep apnoea untreated in long‐term studies, the most likely studies to be conducted in severely symptomatic patients will have an active form of control. Such studies would provide useful information on the relative effectiveness of surgery.

The severity of sleep apnoea is conventionally measured by polysomnography, incorporating measurements of the frequency of AHI. It is widely acknowledged that the disease is defined in terms of both physiological markers such as AHI, and subjective markers such as symptoms and heath‐related quality of life (SIGN 2003). Although the correlation between changes in AHI and symptoms remains weak (Weaver 2005), reducing AHI and reducing the burden of sleep apnoea in terms of symptoms and quality of life are primary goals of therapy (SIGN 2003).

When compared with sham or usual care, surgery did improve some important outcomes, but this effect was not consistent. For example, there was an effect in favour of TCRFTA over a sham procedure in Epworth symptom scores and SAQLI questionnaire (Woodson 2003), but LAUP did not reduce these scores compared with conservative management (Ferguson 2003). Conversely, AHI reduction favoured LAUP in Ferguson 2003, but was not significantly different between TCRFTA and placebo in Woodson 2003. The control regimens differed in these trials, although it could be that in Woodson 2003, blinding was not adequately concealed, with a limit on the number of treatment sessions, and the absence of localised stimulation at the treatment sites.

The symptoms of those participating in the trials were mild to moderate. It is not unreasonable to suppose that participants exhibiting greater signs of daytime somnolence at baseline stand to perceive greater benefit from treatment. With the exception of Woodson 2003; Friedman 2008 and Lojander 1996, there was a lack of an impact on symptoms. This may be at least partly explained by the moderate baseline symptoms of the populations studied. Patients with severe OSAHS have been excluded from three of the studies assembled (Ferguson 2003; Tegelberg 1999; Woodson 2003). Trials in more overtly symptomatic participants might assist in determining the strength of evidence in favour of surgery in patients who have otherwise been shown to benefit from CPAP therapy (Giles 2006), but who may not persist with treatment (Haniffa 2004). Thomas 2003 and Cahali 2004 recruited participants with relatively high AHI, although these participants had moderate baseline symptoms. Within‐group comparisons of change in symptom scores in actively treated participants were statistically significant. However, the absence of adequate control means that they provide evidence on the issue of relative efficacy.

Screening failures were reported in one study, and this indicated a very highly selected sample of participants with just under a fifth of those screened meeting the eligibility criteria of the study (Woodson 2003). This could undermine the generalisability of the study findings of this trial, but may equally be a reflection of the necessary rigour required when selecting patients who are most likely benefit from removal of tissue or mandibular manipulation. Ferguson 2003 failed to show a symptomatic improvement in unselected participants given LAUP over conservative management. End of treatment values for AHI suggested that surgery may not have been sufficiently effective in improving sleep quality, so that feelings of residual daytime sleepiness were not altered.

Unlike other forms of management of OSAHS which involve adherence with a device or drug regimen, surgery offers a 'one off' or 'short course' treatment which could alleviate signs and symptoms of OSAHS, forestalling future requirement for therapy and attendant difficulties with their poor acceptance (Haniffa 2004). There is a paucity of data on the long term follow‐up of surgically treated participants, which would otherwise help to establish whether those who opt for surgery require further intervention, and what the complication rate is. There is a need to ensure that patients are not merely 'symptom free', but that they are not at risk from long term cardiovascular disease and events where OSAHS is under‐treated (Mooe 2001).

Data from long‐term (greater than 12 month) follow‐up in this review are drawn from Tegelberg 1999, Lojander 1996 and Atef 2005. In Tegelberg 1999, the mean difference in favour oral appliance therapy in AHI after one year increased at four years. This could be a reflection of the continued acceptance of the device in the more censored population participating in the study at four years in the OA group, where a higher withdrawal rate occurred. The 'contentment' domain of a quality of life questionnaire used in this study favoured surgery. Whilst improvement in symptoms were reported by Lojander 1996, this was on an un validated scale. The data from the Atef 2005 study are limited by the lack of between treatment group assessment, and the unclear numbers of participants completing follow‐up.

A large retrospective cohort study has identified a significant effect of surgery over CPAP on >1 year mortality in 20,826 ex‐servicemen with OSAHS treated with either of these options in North America (Weaver 2004). It should be noted that imbalances between the groups existed in terms of sample size, age and co‐morbidity, which could bias the findings against CPAP. The absence of data on acceptance and continued usage of CPAP therapy, along with an absence of polysomnographic, symptom, adverse events and quality of life data for patients undergoing both treatment options, mean that the study does not provide definitive evidence on long‐term effects of either of these interventions.

Concerns regarding the interference of surgery on CPAP effectiveness have been raised (Mortimore 1996). In non‐randomised evidence to date, this appeared to increase requisite therapeutic pressure levels in 29% patients continuing with CPAP after UPPP surgery (Masdon 2004). Studies which assess CPAP, oral appliance and drug therapy tend to exclude patients who have undergone prior surgical treatment, presumably on the assumption that they will not stand to benefit from 'additional' treatment to the same degree as those who are naive to treatment. However, it is imperative that studies should remain relevant to both those who present for the first time, and those who present because other forms of treatment have failed. Future research is needed to determine the extent that CPAP effectiveness is compromised in patients who have undergone removal of palatal tissue.

Authors' conclusions

Implications for practice.

The available evidence does not currently support the widespread use of surgical interventions in the management of unselected patients with obstructive sleep apnoea. Given the proven efficacy of CPAP in patients with moderate and severe symptoms and significant sleep disordered breathing, surgery cannot be recommended as a front line therapy, ahead of positive airways pressure in these patients. Long‐term follow up comparing surgery with oral appliances suggests that the initial effect of surgery on airway obstruction lessened over time. This observation requires confirmation. The appeal of an apparently curative intervention for sleep apnoea should be balanced against the uncertainty surrounding its safety, continued effectiveness and inconsistent impact on subjective and objective markers of disease in patients with OSAHS.

Implications for research.

Surgical treatment has not been shown to be consistently effective compared with different forms of active and inactive control. However, although surgery may not be appropriate in all patients with OSAHS, there may be subgroups of patients who could benefit from this type of therapy, and surgical trials should continue to report the nature of airway obstruction in their participants, and how this relates to symptomatic response. Randomised controlled trials are required to help establish whether there are differences between surgical techniques in terms of ESS and other important markers of OSAHS. Long‐term comparison with the current first line of care for sleep apnoea (CPAP) would help to confirm whether CPAP is superior in terms of symptoms and ongoing management in the light of the possible moderation of surgical success over time, and its uncertain long‐term safety profile. Additional randomised studies are required to assess the impact of UPPP on CPAP effectiveness/adherence. Cardiovascular outcomes such as blood pressure should also be included in future studies. The advent of more tolerable forms of positive airway pressure delivered via automatic titration should also be compared with surgical procedures. Long term follow up of patients (possibly from prospective cohorts) would help to elucidate the continued interface between patients and sleep services, as the frequency of requirement for additional treatment is hard to establish from the randomised evidence to date. Further research should be undertaken to identify and standardise techniques to identify the site of airway obstruction. Localisation of the site of airway obstruction may allow more specific selection of surgical procedures.

What's new

Date	Event	Description
14 January 2013	Amended	Abstract and plain language summary edited following feedback received on plain language summary (not submitted via Cochrane Library feedback mechanism).

History

Protocol first published: Issue 1, 1997
 Review first published: Issue 2, 1998

Date	Event	Description
1 September 2008	Amended	Converted to new review format.
1 July 2008	New search has been performed	Literature search re‐run; two new studies included. The conclusions did not change for this review.
8 July 2005	New citation required and conclusions have changed	Substantive amendment

Data and analyses

Comparison 1. TCRFTA versus sham TCRFTA.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Change from baseline in AHI	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2 Change from baseline in Epworth sleepiness score	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3 Change from baseline in quality of life (Functional Outcome of Sleep Questionnaire)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
4 Pain @ 1 week (VAS score)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5 Pain @ 3 weeks (VAS score)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

1.1. Analysis.

Comparison 1 TCRFTA versus sham TCRFTA, Outcome 1 Change from baseline in AHI.

1.2. Analysis.

Comparison 1 TCRFTA versus sham TCRFTA, Outcome 2 Change from baseline in Epworth sleepiness score.

1.3. Analysis.

Comparison 1 TCRFTA versus sham TCRFTA, Outcome 3 Change from baseline in quality of life (Functional Outcome of Sleep Questionnaire).

1.4. Analysis.

Comparison 1 TCRFTA versus sham TCRFTA, Outcome 4 Pain @ 1 week (VAS score).

1.5. Analysis.

Comparison 1 TCRFTA versus sham TCRFTA, Outcome 5 Pain @ 3 weeks (VAS score).

Comparison 2. Laser assisted UP versus conservative management.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 AHI	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2 Epworth sleepiness score	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3 Quality of life (sleep apnoea quality of life index)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
4 Snoring intensity (VAS)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5 Snoring frequency score (VAS)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
6 Withdrawals	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
7 Dysphagia	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
8 Infection	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
9 Bleeding (mild‐severe)	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
10 Pain	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
11 Nasal regurgitation	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected

2.1. Analysis.

Comparison 2 Laser assisted UP versus conservative management, Outcome 1 AHI.

2.2. Analysis.

Comparison 2 Laser assisted UP versus conservative management, Outcome 2 Epworth sleepiness score.

2.3. Analysis.

Comparison 2 Laser assisted UP versus conservative management, Outcome 3 Quality of life (sleep apnoea quality of life index).

2.4. Analysis.

Comparison 2 Laser assisted UP versus conservative management, Outcome 4 Snoring intensity (VAS).

2.5. Analysis.

Comparison 2 Laser assisted UP versus conservative management, Outcome 5 Snoring frequency score (VAS).

2.6. Analysis.

Comparison 2 Laser assisted UP versus conservative management, Outcome 6 Withdrawals.

2.7. Analysis.

Comparison 2 Laser assisted UP versus conservative management, Outcome 7 Dysphagia.

2.8. Analysis.

Comparison 2 Laser assisted UP versus conservative management, Outcome 8 Infection.

2.9. Analysis.

Comparison 2 Laser assisted UP versus conservative management, Outcome 9 Bleeding (mild‐severe).

2.10. Analysis.

Comparison 2 Laser assisted UP versus conservative management, Outcome 10 Pain.

2.11. Analysis.

Comparison 2 Laser assisted UP versus conservative management, Outcome 11 Nasal regurgitation.

Comparison 3. UPPP versus conservative management.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Excessive daytime sleepiness (VAS)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2 Falling asleep when not in bed	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
3 ODI4	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
4 ODI10	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5 Withdrawals	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
6 Dysphagia	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected

3.1. Analysis.

Comparison 3 UPPP versus conservative management, Outcome 1 Excessive daytime sleepiness (VAS).

3.2. Analysis.

Comparison 3 UPPP versus conservative management, Outcome 2 Falling asleep when not in bed.

3.3. Analysis.

Comparison 3 UPPP versus conservative management, Outcome 3 ODI4.

3.4. Analysis.

Comparison 3 UPPP versus conservative management, Outcome 4 ODI10.

3.5. Analysis.

Comparison 3 UPPP versus conservative management, Outcome 5 Withdrawals.

3.6. Analysis.

Comparison 3 UPPP versus conservative management, Outcome 6 Dysphagia.

Comparison 4. Palatal implants versus placebo.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Mean change in AHI	1		Mean Difference (Fixed, 95% CI)	Totals not selected
2 Mean change in Epworth sleepiness scores	1		Mean Difference (Fixed, 95% CI)	Totals not selected
3 Change from baseline in quality of life (SF‐36)	1		Mean Difference (Fixed, 95% CI)	Totals not selected
4 Loss to follow‐up	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected

4.1. Analysis.

Comparison 4 Palatal implants versus placebo, Outcome 1 Mean change in AHI.

4.2. Analysis.

Comparison 4 Palatal implants versus placebo, Outcome 2 Mean change in Epworth sleepiness scores.

4.3. Analysis.

Comparison 4 Palatal implants versus placebo, Outcome 3 Change from baseline in quality of life (SF‐36).

4.4. Analysis.

Comparison 4 Palatal implants versus placebo, Outcome 4 Loss to follow‐up.

Comparison 5. UPPP versus oral appliance therapy.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Apnoea Hypopnea Index	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
1.1 6 months	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
1.2 1 year	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
1.3 4 years	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
2 Quality of life: Vitality	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3 Quality of life: Contentment	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
4 Quality of life: Sleep	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5 Oxygen desaturation index	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5.1 6 months	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.2 1 year	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.3 4 years	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
6 Withdrawals	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
7 Dysphagia	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
8 Nasopharyngeal regurgitation	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
9 Repeated adjustment	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected

5.1. Analysis.

Comparison 5 UPPP versus oral appliance therapy, Outcome 1 Apnoea Hypopnea Index.

5.2. Analysis.

Comparison 5 UPPP versus oral appliance therapy, Outcome 2 Quality of life: Vitality.

5.3. Analysis.

Comparison 5 UPPP versus oral appliance therapy, Outcome 3 Quality of life: Contentment.

5.4. Analysis.

Comparison 5 UPPP versus oral appliance therapy, Outcome 4 Quality of life: Sleep.

5.5. Analysis.

Comparison 5 UPPP versus oral appliance therapy, Outcome 5 Oxygen desaturation index.

5.6. Analysis.

Comparison 5 UPPP versus oral appliance therapy, Outcome 6 Withdrawals.

5.7. Analysis.

Comparison 5 UPPP versus oral appliance therapy, Outcome 7 Dysphagia.

5.8. Analysis.

Comparison 5 UPPP versus oral appliance therapy, Outcome 8 Nasopharyngeal regurgitation.

5.9. Analysis.

Comparison 5 UPPP versus oral appliance therapy, Outcome 9 Repeated adjustment.

Comparison 6. TCRFTA versus CPAP.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Mean change from baseline in Apnoea Hypopnea Index	1		Events/hr (Fixed, 95% CI)	Totals not selected
2 Mean change from baseline in Epworth Sleepiness Score	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3 Quality of life (Functional Outcomes of Sleep Questionnaire)	1		QoL (Fixed, 95% CI)	Totals not selected
4 Withdrawals/loss to follow‐up	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected

6.1. Analysis.

Comparison 6 TCRFTA versus CPAP, Outcome 1 Mean change from baseline in Apnoea Hypopnea Index.

6.2. Analysis.

Comparison 6 TCRFTA versus CPAP, Outcome 2 Mean change from baseline in Epworth Sleepiness Score.

6.3. Analysis.

Comparison 6 TCRFTA versus CPAP, Outcome 3 Quality of life (Functional Outcomes of Sleep Questionnaire).

6.4. Analysis.

Comparison 6 TCRFTA versus CPAP, Outcome 4 Withdrawals/loss to follow‐up.

Comparison 7. UPPP versus lateral PP.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 AHI	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2 Change in AHI from baseline	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3 Nasal regurgitation	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
4 Withdrawals	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected

7.1. Analysis.

Comparison 7 UPPP versus lateral PP, Outcome 1 AHI.

7.2. Analysis.

Comparison 7 UPPP versus lateral PP, Outcome 2 Change in AHI from baseline.

7.3. Analysis.

Comparison 7 UPPP versus lateral PP, Outcome 3 Nasal regurgitation.

7.4. Analysis.

Comparison 7 UPPP versus lateral PP, Outcome 4 Withdrawals.

Comparison 8. Tongue advancement + palatopharyngoplasty versus tongue suspension + palatopharyngoplasty.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Epworth sleepiness scores	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

8.1. Analysis.

Comparison 8 Tongue advancement + palatopharyngoplasty versus tongue suspension + palatopharyngoplasty, Outcome 1 Epworth sleepiness scores.

Comparison 9. RAUP versus submucosal channeling technique.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Reduction in AHI below 10	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2 Pain score	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3 Dysphagia	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

9.1. Analysis.

Comparison 9 RAUP versus submucosal channeling technique, Outcome 1 Reduction in AHI below 10.

9.2. Analysis.

Comparison 9 RAUP versus submucosal channeling technique, Outcome 2 Pain score.

9.3. Analysis.

Comparison 9 RAUP versus submucosal channeling technique, Outcome 3 Dysphagia.

Comparison 10. Expansion sphincter pharyngoplasty (ESP) versus UPPP.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 AHI	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

10.1. Analysis.

Comparison 10 Expansion sphincter pharyngoplasty (ESP) versus UPPP, Outcome 1 AHI.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Atef 2005.

Methods	Randomised, parallel group trial. Method of randomisation not reported; participants also randomised within each treatment group to between 1‐5 sessions of treatment. Blinding not reported. Loss to follow‐up: LAUP: 13; BRVTR: 11)
Participants	N = 150. No data were provided on gender. Mean age: 47. Mean AHI for treatment groups at baseline: 25‐27. Participants suffered from mild to moderate OSA. Inclusion criteria: AHI between 5‐30; OSA obstruction mainly at the palate site; refusal of other treatment options (CPAP, conservative management (including weight‐loss) and oral appliances) Exclusion criteria: BMI >30kg/m2; simple snoring, OSA due to nasal or retrolingual obstruction; large obstructing tonsils; prior palate surgery.
Interventions	LAUP versus BRVTR. Follow‐up conducted at 3 and 18 months post‐operatively. Participants randomly assigned within each treatment group to between one and five treatment sessions.
Outcomes	AHI; symptoms (measured on visual analogue scale e); withdrawal
Notes	Jadad score: 2
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information presented to ascertain risk of bias in generating randomisation sequence. Participants randomly assigned within each treatment group to between one and five treatment sessions.
Allocation concealment (selection bias)	Unclear risk	Insufficient information presented to ascertain risk of bias of investigators being aware of order of treatment group assignment
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Information not available
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Information not available

Bassiouny 2007.

Methods	Randomised, parallel group trial. Method of randomisation not reported. Single‐blind study.
Participants	N = 40 AHI: Unclear. BMI: 28 Inclusion criteria: Snoring due to palatal redundancy and/or excessive lengthening; AHI 10 ‐ 30; O2 saturation >85%. Exclusion criteria: Snoring/OSA due causes other than palatal redundancy and/or excessive lengthening (such as tonsillar hypertrophy, tongue base hypertrophy or lateral pharyngeal walls redundancy); Simple snoring (AHI < 10 events per hour); Severe OSA (AHI > 30 events per hour); Morbid obesity, body mass index >30.
Interventions	RAUP versus RCT. Follow‐up conducted over 4 months.
Outcomes	Pain; snoring score; dysphagia
Notes	Jadad score: 1
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information presented to ascertain risk of bias in generating randomisation sequence.
Allocation concealment (selection bias)	Unclear risk	Insufficient information presented to ascertain risk of bias of investigators being aware of order of treatment group assignment
Blinding (performance bias and detection bias) All outcomes	High risk	Single‐blind study.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Information not available

Cahali 2004.

Methods	Randomised, parallel group trial. Method of randomisation not reported (stratified for severity of AHI); blinding not undertaken.
Participants	N = 29. No data were provided on gender or mean age. Two withdrawals from UPPP group. Mean BMI: Lateral PP: 29.3; UPPP: 30.1; median ESS: 14 Inclusion criteria: low lying soft palate associated with fibreoptic pharyngoscopic finding of narrowing/collapse in retropalatal region without narrowing in hypopharynx (at rest and during Muller maneuver); >18 years; AHI >/= 10; failure to tolerate CPAP; failure to achieve significant weight loss in overweight participants; all participants chosen on basis of subjective finding of bulky lateral oropharyngeal tissues Exclusion criteria: >130kg in weight; morbid obesity; hypothyroidism (uncontrolled or present <12 m); gross maxillary/mandible deformities.
Interventions	UPPP versus lateral PP. Follow‐up conducted at least 6 months after operation (range: 7‐12 months)
Outcomes	AHI; ESS; neck circumference; BMI; Time taken to return to normal nutrition; QoL
Notes	Jadad score: 2
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information presented to ascertain risk of bias in generating randomisation sequence. Stratified for severity of AHI.
Allocation concealment (selection bias)	Unclear risk	Insufficient information presented to ascertain risk of bias of investigators being aware of order of treatment group assignment
Blinding (performance bias and detection bias) All outcomes	High risk	Open label assessment
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Information not available

Ferguson 2003.

Methods	Randomised parallel group trial. Method of randomisation: Random numbers table. Blinding to treatment was not attempted.
Participants	N = 46. 35M; LAUP: 21; No treatment: 24 (one withdrawal after randomisation). Mean age: 44.6 (SD 8.1); BMI 36 (SD 4.5) AHI: LAUP: 18.6 (SD 4.3); Control 16.1 (SD 4); ESS: LAUP: 10.7 (SD 3.7); Control: 10 (SD 5.2) One participant in each group had previously had CPAP. Inclusion criteria: AHI: 10‐25; complaint of 'loud snoring'
Interventions	Laser‐assisted UPP in an outpatient setting. Participants in the control group had no treatment. LAUP was repeated in the treatment group until either snoring was reduced, no further tissue could be removed, or surgery was refused. Surgery was repeated at 1‐2 month intervals. Control group were offered surgery after 6 months
Outcomes	AHI; ESS; SAQLI; Snoring intensity/frequency (assessed by bed partner); Side effects and complications
Notes	Jadad score: 3
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random numbers table
Allocation concealment (selection bias)	Unclear risk	Insufficient information presented to ascertain risk of bias of investigators being aware of order of treatment group assignment
Blinding (performance bias and detection bias) All outcomes	High risk	Surgery compared with usual care.
Incomplete outcome data (attrition bias) All outcomes	Low risk	One withdrawal after randomisation (low attrition rate).

Friedman 2008.

Methods	Randomised parallel group trial.
Participants	N = 62. 33M. Palatal implants: 31; placebo: 31. BMI: 29; AHI: 22 Inclusion criteria: History of OSAHS and/or symptoms of OSAHS (mainly snoring & excessive daytime sleepiness); Friedman tongue position (FTP) I, II, or III; diagnosis of mild or moderate OSAHS (apnea/hypopnea index (AHI) 5 and 40); a soft palate 2 cm, but less than 3.5 cm; BMI 32. Exclusion criteria: Clinical and physiological presentation of severe OSAHS (ESS 20, frequent choking and gasping during sleep; OSAHS (AHI 40)); unwilling to be randomly assigned to placebo; FTP IV; tonsil size 3 or 4; classified stage IV of Friedman staging system.
Interventions	Surgically inserted palatal implants versus sham procedure (active and placebo implants were indistinguishable). Outcomes assessed 12 weeks post‐operatively.
Outcomes	AHI; AI; Min SaO2; ESS; Quality of life (SF‐36); loss to follow‐up; complications
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	'Block randomization (blocks of 6) was performed by the manufacturer...'
Allocation concealment (selection bias)	Low risk	'...patients were assigned to groups before the procedure by sequential sealed envelopes with device lot numbers to be used for each patient. The lot numbers that corresponded to group assignments were not revealed until after the study was completed.'
Blinding (performance bias and detection bias) All outcomes	Low risk	'The palatal implant insertion tools provided by the manufacturer for the placebo control group did not include the palatal implants, but they were in all other aspects identical to the implant insertion tools used in the treatment group receiving the implant. Group assignment associated with insertion tools was not distinguishable by study participants and investigators because the implants are deployed from within the hollow needle of a delivery tool.'
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	'An intent‐to‐treat analysis looking at objective surgical success was performed on all 62 patients who received initial study treatment (31 patients in each group). The seven patients who were lost to follow‐up were considered treatment failures and analyzed accordingly.' Data on AHI & ESS presented for remaining participants.

Larrosa 2004.

Methods	Randomised parallel group study. Method of randomisation: random numbers table. Single‐blind comparison.
Participants	N = 28 adults with mild OSA and non‐apneic snorers. Three withdrawals after randomisation. AHI Surgery: 13.6; control: 17; ESS: Surgery: 10.4; control: 10.9. Inclusion criteria: AHI </=30; age: 30‐60 years Exclusion criteria: Snoring not due to palatal flutter (e.g. tongue base collapse/collapse of other soft tissues)
Interventions	Laser‐assisted UPP versus sham procedure. Assessment at 3 months.
Outcomes	AHI; ESS; QoL (SF‐36); snoring indices.
Notes	Mixed population study. No data have been entered in the review. Request for additional data made 080705.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random numbers table
Allocation concealment (selection bias)	Unclear risk	Insufficient information presented to ascertain risk of bias of investigators being aware of order of treatment group assignment
Blinding (performance bias and detection bias) All outcomes	High risk	Single‐blind study
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Information not available

Lojander 1996.

Methods	Randomised parallel group rial. Not blinded. Participants randomised to either CPAP versus conservative management or to surgery versus conservative management. Method of randomisation not reported
Participants	N = 32 adults with moderate to severe OSAS. Diagnosis confirmed by sleep study. Age range 27‐65, BMI: 24‐41, ODI4 in CPAP: 10‐96. ODI4 in surgery group: 20‐72. Participants with more than 50% obstruction at palatal level in the Mueller's manoeuvre, and those with or without obstruction at the epiglottic level were considered appropriate for surgery. Inclusion criteria: confirmed diagnosis of OSAS, periodic breathing pattern in both static charge sensitive bed and thermistor channels. Participants with BMI more than 40 were excluded. Patients COPD/asthma, other serious concomitant illnesses, and participants where somnolence would cause risk or incapacity to work were excluded.
Interventions	Two groups which compared either CPAP with conservative management (not specified) or surgery (UVPP) with conservative management. Study duration: 1 year.
Outcomes	Number of nocturnal oxygen desaturation events (4% or more per hour), daytime somnolence, side effects, withdrawals.
Notes	Participants were assessed by a team of specialists (including physicians and surgeons). Jadad score: 2
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information presented to ascertain risk of bias in generating randomisation sequence.
Allocation concealment (selection bias)	Unclear risk	Insufficient information presented to ascertain risk of bias of investigators being aware of order of treatment group assignment
Blinding (performance bias and detection bias) All outcomes	High risk	Open label assessment
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Information not available

Naya 2002.

Methods	Randomised parallel group trial. Method of randomisation not reported. Blinding not specified.
Participants	N = 75 adults with ASDA defined sleep apnoea (88% M). Four treatment groups: PPR (20 participants); UPPP (Fuj): 18; UPPP (Sim): 17; UPPP (Fair): 20. Mean age: 47.7 (SD 11.2). Mean Epworth score: 12.94 (SEM 0.7). Inclusion criteria: American Sleep Disorder Association defined OSA
Interventions	Four different surgical techniques: PPR; UPPP (Fuj); UPPP (Sim); UPPP (Fair). Study duration: 12 weeks.
Outcomes	AHI; complications
Notes	No analyses undertaken between groups. Jadad score 1
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information presented to ascertain risk of bias in generating randomisation sequence.
Allocation concealment (selection bias)	Unclear risk	Insufficient information presented to ascertain risk of bias of investigators being aware of order of treatment group assignment.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Information not available
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Information not available

Pang 2007.

Methods	Randomised parallel group trial
Participants	N = 45 (UPPP: 22; ESP: 23) AHI: 42 Inclusion criteria: People with small tonsils (tonsil size 1 and 2), body mass index (BMI) less than 30 kg/m2, and Friedman clinical stage II and III who cannot tolerate nasal continuous positive airway pressure (CPAP) therapy or for whom CPAP therapy failed.
Interventions	Two surgical techniques were compared: 1. Expansion sphincter pharyngoplasty (ESP) 2. Uvulopalatopharyngoplasty (UPPP) Mean follow‐up 6.5 months
Outcomes	AHI; complications
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information presented to ascertain risk of bias in generating randomisation sequence.
Allocation concealment (selection bias)	Unclear risk	Insufficient information presented to ascertain risk of bias of investigators being aware of order of treatment group assignment
Blinding (performance bias and detection bias) All outcomes	High risk	Information not available (assumed not since different procedures necessitated surgeon being aware what group participant assigned to)
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Informatio not available

Tegelberg 1999.

Methods	Randomised, parallel group study. Comparison of surgery versus oral appliance.
Participants	95 male participants were recruited. Age: 20‐65, baseline AHI: 15.7‐23.3. Inclusion criteria: AHI between 5 and 25, age between 20 and 65. Exclusion criteria: Mental illness, drug misuse, significant nasal obstruction, insufficient teeth, pronounced dental malocclusion, severe cardiovascular disease, neurological disease, respiratory disease. At 4 year follow‐up, OA group: N = 32, UPPP group: N = 40.
Interventions	Participants were randomised to either oral appliance or surgical intervention (uvulopalatopharyngoplasty). Participants randomised to receive UPP were followed up at regular intervals. Study duration: 1 year.
Outcomes	AHI, AI, oxygen desaturation index, snoring index, clinical dysfunction score, QOL scores
Notes	Jadad score: 2
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information presented to ascertain risk of bias in generating randomisation sequence.
Allocation concealment (selection bias)	Unclear risk	Insufficient information presented to ascertain risk of bias of investigators being aware of order of treatment group assignment
Blinding (performance bias and detection bias) All outcomes	High risk	Open label assessment
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Information not available

Thomas 2003.

Methods	Randomised controlled trial. Method of randomisation: computer‐generated randomisation sequence. Blinding: not reported
Participants	N = 17 adults. Tongue suspension group (TS): 50.8 (SD 16.1); Tongue advancement group (TA): AHI: TS: 46 (SD 22); TA: 37.4 (SD 15.1); BMI: TS: 30.9 (6.2); TA: 28.7 (SD 2.7) Inclusion criteria: moderate to severe sleep disordered breathing; Fujita type II airway collapse; failure of conservative therapy (CPAP, weight loss etc); BMI <40 Exclusion criteria: prior surgical treatment; contraindications for surgical treatment
Interventions	Tongue advancement (mandibular osteotomy) + PPP versus tongue suspension + PPP. Study duration: 16 weeks
Outcomes	Epworth Sleepiness score; Airway collapse (Müller maneuver); snoring
Notes	Jadad score: 3
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated randomisation sequence
Allocation concealment (selection bias)	Unclear risk	Insufficient information presented to ascertain risk of bias of investigators being aware of order of treatment group assignment
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Information not available
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Information not available

Woodson 2003.

Methods	Randomised, single‐blinded, parallel group trial. Method of randomisation: computer generated random numbers table (allocation concealed using sealed envelopes). Withdrawals reported. Data not analysed for 8 participants who did not complete study.)
Participants	N = 90 (CPAP: 30 and TCRFTA: 30; placebo: 30); mean age: Placebo: 46; TCRFTA: 49.4; CPAP: 51.7; BMI: Placebo: 28.5; TCRFTA: 27.7; CPAP: 29.1; AHI: Placebo: 15.4; TCRFTA: 21.3; CPAP: 19.8; ESS: Placebo: 11.6; TCRFTA: 11.9; CPAP: 12.6; FOSQ: Placebo: 16.8; TCRFTA: 16.5; CPAP: 16 Inclusion criteria: mild to moderate OSA (AHI 10 to 30; age 18‐65; self‐reported daytime sleepiness; BMI </=34; no prior surgical or CPAP treatment Exclusion criteria: co‐existing significant sleep disorder; tonsillar hypertrophy; nasal supraglottic obstruction on examination; ASA class IV/V; claustrophobia; latex allergy; pregnancy; major depression; drug/alcohol abuse; history of an accident secondary to sleepiness; participation in another study
Interventions	TCRFTA (radiofrequency energy delivered to create non‐overlapping lesions in two/three tongue sites, occurring at 4 week intervals. Data recorded after kast treatment session. Palate sessions also included) versus sham TRCFTA. Both active and sham surgical interventions were compared with nasal CPAP. CPAP outcome data reported at 8 weeks.
Outcomes	Number of surgery sessions; CPAP machine usage; FOSQ; ESS; AHI; CPAP treatment pressure; pain/side effects; reaction times
Notes	Jadad score: 3
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Computer generated random numbers table.
Allocation concealment (selection bias)	Low risk	Allocation concealed using sealed envelopes.
Blinding (performance bias and detection bias) All outcomes	High risk	Single blind assessment (surgery versus sham surgery); open label assessment with CPAP.
Incomplete outcome data (attrition bias) All outcomes	High risk	Data not analysed for 8 participants who did not complete study

AHI: Apnoea hypopnea index; AI: Apnoea index; UPP: Uvulopalatoplasty; UPPP: Uvulopalatopharyngoplasty; QoL: Quality of life; Lateral PP: Lateral pharyngoplasty;OSA: Obstructive Sleep Apnoea; SDB: Sleep Disordered Breathing; ESS: Epworth Sleepiness Score; MAD: Mandibular Advancement Device; CPAP: Continuous positive airways pressure; TCRFTA: Temperature‐contreolled radiofrequency tissue ablation; PPR: Partial palatal resection; UPPP (Fuj): Uvulopalatopharyngoplasty according to Fujita; UPPP (Sim): Uvulopalatopharyngoplasty according to Simmons; UPPP (Fair): Uvulopalatopharyngoplasty according to Fairbanks; BRVTR: Bipolar radiofrequency volumetric tissue reduction; RAUP: Radioassisted uvulopalatopharyngoplasty; RCT: Radiofrequency channeling technique

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Abe 1996	No suitable comparison group
Achena 1991	No suitable comparison group
Afzelius 1981	Case series
Anand 1991	No suitable comparison group
Anon 1996	No suitable comparison group
ASDA 1996	Review
Aspinall 2001	Trial comparing anaesthetic agents, and their effect on pain after adeno‐tonsillectomy
Aubert‐Tulkens 1987	Not a trial
Balcerzak 2005	Unclear method of allocation leading to a highly imbalanced distribution of the number of participants between surgery and control
Banerjee 1995	No suitable comparison group
Bear 1980	Case series
Bernecker 1993	No suitable comparison group
Blakley 1986	No suitable comparison group
Borowiecki 1985	Case series
Boudewyns 1995	No suitable comparison group
Bower 1995	No suitable comparison group
Breaux 1995	No suitable comparison group
Brock 1994	Review
Brodsky 1987	Not a trial
Brouilette 1984	No suitable comparison group
Burbach 1993	Not a trial
Burstein 1995	No suitable comparison group
Byrd 1996	No suitable comparison group
Callanan 1994	No suitable comparison group
Calverley 1995	Review
Carenfelt 1993	No suitable comparison group
Cartwright 1988	Not randomised
Castiglione 1992	No suitable comparison group
Chabolle 1988	No suitable comparison group
Chabolle 1995	No suitable comparison group
Chabolle 1995a	No suitable comparison group
Chaska 1995	No suitable comparison group
Chetty 1994	No suitable comparison group
Chua 1994	No suitable comparison group
Cimimo 1995	Not a trial
Cimimo 1995a	No suitable comparison group
Clauser 1996	No suitable comparison group
Clerk 1994	Not surgery
Coleman 1991	Not a trial
Crampette 1992	Review
Croft 1990	No suitable comparison group
Cvetnic 1995	No suitable comparison group
Dickson 1987	Case series
Dierks 1990	Case series
Djupesland 1992	Case series
Doghramji 1995	No suitable comparison group
Donnelly 1994	No suitable comparison group
Douglas 1994a	No suitable comparison group
Douglas 1994b	Review
Douglas 1995	No suitable comparison group
Ejnell 1992	No suitable comparison group
ElSheikh 1996	No suitable comparison group
Engleman 1994	Not surgery
Fietze 1993	No suitable comparison group
Findley 1989	Review
Finkelstein 1995	No suitable comparison group
Finkelstein 1995a	No suitable comparison group
Fraser 1994	Review
Freezer 1995	No suitable comparison group
Freezer 1996	No suitable comparison group
Friberg 1995	No suitable comparison group
Fried 1995	No suitable comparison group
Frith 1985	No suitable comparison group
Fujita 1981	Case series
Fujita 1985	case series
Fujita 1991	Case series
Fujita 1991a	Case series
Furlow 1986	No suitable comparison group
Gislason 1988	Case series
Grunstein 1995	Not surgery
Gryczynska 1995	No suitable comparison group
Guileminault 1986	No suitable comparison group
Guilleminault 1983	Case series
Haraldsson 1990	Review
Haraldsson 1995	No suitable comparison group
Haraldsson 1995a	No suitable comparison group
Harmon 1986	Case series
He 1988	No suitable comparison group
Hester 1995	No suitable comparison group
Hochban 1994	Case series
Hochban 1995	Not a trial
Hochban 1995a	No suitable comparison group
Hochban 1996	No suitable comparison group
Hoffman 1995	No suitable comparison group
Hoffstein 1996	Not a trial
Hoijer 1992	No suitable comparison group
Hudgel 1991	Case series
Hudgel 1996	Review
Hultcrantz 1999	Randomised study assessing the effects of post‐operative pain relief.
Ishizuka 1996	No suitable comparison group
Jakubikova 1996	No suitable comparison group
Janson 1994	No suitable comparison group
Johnson 1994	Case series
Kamami 1994	No suitable comparison group
Kamami 1994a	No suitable comparison group
Kanami 1994	No suitable comparison group
Katsantonis 1988	Review
Kawano 1996	Case series
Keenan 1994	Case series
Keilmann 1995	No suitable comparison group
Khan 1995	Not surgery
Kimmelman 1985	Not a trial
Klonoff 1987	No suitable comparison group
Kobayashi 1996	Case report
Kosho 1995	No suitable comparison group
Kotecha 2007	Study primarily aimed at reducing snoring in people +/‐ OSA
Krespi 1994	Case series
Krieger 1988	Review
L'Estrange 1995	No suitable comparison group
Lacahanas 2005	Study assessed immediate post‐operative effects in a mixed population (including people with OSA). Not specifically designed to address efficacy.
Larsson 1994	Case series
Leiter 1996	No suitable comparison group
Liistro 1995	No suitable comparison group
Loui 1994	Case series
Lu 1995	No suitable comparison group
Lysdahl 2002	Non‐randomised prospective study.
MacCallum 2000	Study of trachestomies, non‐randomised, with only 1 sleep apnoea patient
Maisel 1992	No suitable comparison group
Marcus 1990	No suitable comparison group
Marcus 1996	No suitable comparison group
Marrone 1995	No suitable comparison group
Matsuda 1995	Case report
Mayer 1995	Case report
McGuirt 1995	Case series
Mickelson 1996	No suitable comparison group
Miljeteig 1992	Case series
Min 1995	Case series
Minarikova 1995	Case series
Mintz 1995	Not a trial
Miyazaki 1989	Case series
Mochizuki 1996	Case series
Mochizuki 1996a	Case series
Mondain 1994	Not a trial
Morin 1994	Review
Murakami 2002	Retrospective analysis.
Nakai 1995	Case series
Nimkarn 1995	No suitable comparison group
Nishimura 1996	Case series
Noseda 1996	Not surgery
Ohta 1993	Not a trial
Paaske 1994	Case report
Pack 1994	Review
Partinen 1988	Case series
Partinen 1990	Case series
Partinen 1996	Review
Penek 1995	Review
Peng 1995	Case series
Peter 1995	Review
Piper 1995	Case series
Polo 1994	Review
Pories 1995	Not OSA
Potsic 1987	Case series
Prinz 1995	Review
Pulido 1995	Case series
Quinn 1995	Case series
Rapoport 1986	Case series
Reimao 1985	Case series
Reimao 1986	Case series
Riley 1989	Case series
Riley 1990	Case series
Riley 1990a	Not randomised
Riley 1994	Case series
Riley 1995	Review
Rodenstein 1992	Review
Rodenstein 1995	Not a trial
Rodrigues 2002	Not randomised
Rothschild 1995	Case series
Sangal 1992	Not surgery
Schechtman 1995	Review
SchultzCoulon 1996	Case series
Schwartz 1992	Case series
Sdralis 1996	Case series
Semple 1993	Review
Series 1992	Case series
Sher 1985	No suitable comparison group
Sher 1995	Review
Sher 1996	Not a trial
Sher et al 1996	Not a trial
Simmons 1984	Case series
Simonds 1994	Not a trial
Skatvedt 1995	Case series
Stradling 1990	Review
Stradling 1990a	Not a trial
Stradling 1995	Review
Strollo 1996	Not a trial
Suen 1995	No suitable comparison group
Sugerman 1986	Case series
Sullivan 1985	Review
Tangugsorn 1995	Not surgery
Tangugsorn 1995a	Not surgery
Terris 1996	Case series
Terris 2002	RCT assessing the effects of two surgical techniques. This study was excluded due to low baseline AHI scores in many of the participants. The SDs indicated that a significant proportion of the participants would not be considered OSA sufferers (combined mean: 6.1, SD 5.18).
Thalhofer 1995	Case series
Thorpy 1994	Review
Troell 2000	Patient preference comparison of effect of several types of surgery on post‐operative pain in snoring or sleep‐disordered breathing, including sleep apnoea
Truy 1995	Not sleep apnoea
Verse 2002	Before and after study
Wakeing 1996	Case series
Waldhorn 1990	Not surgery
Walker 1989	Case series
Walker 1995	Case series
Weiss 1995	Case series
Wetmore 1986	Case series
Wiltfang 1996	Case series
Woodson 1994	Case series
Woodson 1996	No suitable comparison group
Wright 1995	Review
Yamadera 1994	Case series
Yamadera 1995	No suitable comparison group
Yoder 1995	Review
Zohar 1993	Not randomised
Zorick 1983	Case series
Zorick 1989	Not randomised
Zorick 1990	Case series

Contributions of authors

SS: Review update 2005 ‐ study assessment, data extraction, data entry and write‐up 
 SB: Initial author of the review; protocol initiation, study assessment, write‐up 
 TJL: Review update 2005 ‐ study assessment, data extraction, data entry, analysis and write‐up 
 FD: Interpretation and write‐up 
 JL: Review update 2005 ‐ interpretation and write‐up

Sources of support

Internal sources

• North Staffordshire Health Authority, UK.

External sources

• Aggressive Research Intelligence Facility (ARIF), UK.

Declarations of interest

None known.

Edited (no change to conclusions)