Effects and Side-Effects of Surgery for Snoring and Obstructive Sleep Apnea – A Systematic Review

Abstract

Study Objectives:

Many patients undergo surgery for snoring and sleep apnea, although the efficacy and safety of such procedures have not been clearly established. Our aim was systematically to review studies of the efficacy and adverse effects of surgery for snoring and obstructive sleep apnea.

Design:

Systematic review.

Measurements:

PubMed and Cochrane databases were searched in September 2007. Randomized controlled trials of surgery vs. sham surgery or conservative treatment in adults, with daytime sleepiness, quality of life, apnea-hypopnea index, and snoring as outcomes were included. Observational studies were also reviewed to assess adverse effects. Evidence of effect required at least two studies of medium and high quality reporting the same result.

Results:

Four studies of benefits and 45 studies of adverse effects were included. There was no significant effect on daytime sleepiness and quality of life after laser-assisted uvulopalatoplasty and radiofrequency ablation. The apnea-hypopnea index and snoring was reduced in one trial after laser-assisted uvulopalatoplasty but not in another trial. Subjective snoring was reduced in one trial after radiofrequency ablation. No trial investigating the effect of any other surgical modality met the inclusion criteria. Persistent side-effects occurred after uvulopalatopharyngoplasty and uvulopalatoplasty in about half the patients and difficulty in swallowing, globus sensation and voice changes were especially common.

Conclusions:

Only a small number of randomized controlled trials with a limited number of patients assessing some surgical modalities for snoring or sleep apnea are available. These studies do not provide any evidence of effect from laser-assisted uvulopalatoplasty or radiofrequency ablation on daytime sleepiness, apnea reduction, quality of life or snoring. We call for research of randomized, controlled trials of surgery other than uvulopalatopharyngoplasty and uvulopalatoplasty, as they are related to a high risk of long-term side-effects, especially difficulty swallowing.

Citation:

Franklin KA; Anttila H; Axelsson S; Gislason T; Maasilta P; Myhre KI; Rehnqvist N. Effects and side-effects of surgery for snoring and obstructive sleep apnea – a systematic review. SLEEP 2009;32(1):27–36.

SNORING IS A SIGN OF INCREASED UPPER AIRWAY RESISTANCE, AND OBSTRUCTIVE SLEEP APNEA IS CHARACTERIZED BY RECURRENT EPISODES OF UPPER airway obstruction during sleep. Excessive daytime sleepiness is common among heavy snorers and among patients with obstructive sleep apnea syndrome.^1–3 A number of recent prospective studies report than patients with obstructive sleep apnea run an increased risk of stroke and early death.^4–8

Surgical treatment for snoring and obstructive sleep apnea aims to increase the upper airway cross-sectional area, remove obstructive tissues, such as enlarged tonsils, or bypass the upper airway. Uvulopalatopharyngoplasty became popular shortly after its introduction in 1981.⁹ The uvula, the distal part of the soft palate and the tonsils are removed, with the aim of enlarging the oropharyngeal airspace and thereby reducing sleep apneas, snoring, and daytime sleepiness. Laser-assisted uvulopalatoplasty and temperature-controlled radiofrequency tissue ablation were subsequently introduced, and these surgical modalities could be performed under local anesthesia. Other surgical modalities include tracheostomy, inferior sagittal mandibular osteotomy and genioglossus advancement with hyoid myotomy and suspension, laser midline glossectomy and lingual plasty, maxillo-mandibular osteotomy and advancement, expansion sphincter pharyngoplasty, palatal implants, nasal surgery, epiglottoplasty for selected cases of laryngomalacia, and so on. The efficacy of surgical modalities has, however, been questioned in systematic reports, and laser-assisted uvulopalatopharyngoplasty is not recommended for the treatment of sleep disordered breathing.^10–14 There is a lack of systematic reviews of adverse effects, as many subjects still undergo surgery because of snoring and sleep apnea.

The present aim was systematically to review studies of the efficacy and adverse effects of surgery for snoring and obstructive sleep apnea in adults.

METHODS

Search Strategy

PubMed and the Cochrane Controlled Trials Register were searched on 1 September 2007; (“Sleep apnea syndromes” [Mesh] OR “Snoring” [Mesh]) AND (“Surgery” [Mesh] OR “Surgery” [Subheading] or Radiofrequency) Limits: English, Randomized Controlled Trial, Meta-Analysis.

PubMed was searched on 1 September 2007 for side-effects; sleep apnea syndromes/surgery [Mesh] OR snoring/surgery [Mesh] AND (adverse effects [Mesh Subheading] OR complications [Mesh subheading] OR complication [Text Word] OR complications [Text Word]) NOT (comment [Publication Type] OR editorial [Publication Type] OR news [Publication Type]) Limits: English.

Reference lists of all the identified articles were searched for additional studies.

Inclusion Criteria for the Selection of Articles and Study end Points

Two reviewers independently screened abstracts according to inclusion criteria. For all possibly relevant articles, full reports were requested. Randomized, controlled studies of medium and high quality comparing surgery for snoring or obstructive sleep apnea with sham operations or conservative treatment were included. The primary outcomes were daytime sleepiness measured using the Epworth Sleepiness Scale, Multiple sleep latency test or Maintenance of wakefulness test and Quality of life. Secondary outcomes were the apnea-hypopnea index and severity of snoring.

The review of adverse effects included randomized, controlled studies and observational studies of medium and high quality reporting complications and side-effects of surgery other than post-operative pain. All studies reporting life-threatening side effects and deaths in relation to surgery were included.

Quality Assessment and Evidence Grading

A modified JADAD ranking scale from 0–5 was used to assess the quality of randomized, controlled trials.^13,15 We used questions on single blinding instead of double blinding, with the following questions: Was the trial described as randomized? Was the allocation concealed? Were patients blinded to treatment alternative? Were investigators blinded to treatment alternative? Was there a description of withdrawals? A score of 0–2 was rated as poor quality, 3 as medium quality, and 4–5 as high quality.

The following scores were used for observational studies reporting side-effects. High quality required a prospective study; before and after data for adverse effects; defined groups of patients and a detailed description of methods and adverse effects; loss to follow-up of less than 30%. Medium quality required a prospective design or consecutive patients; a summary description of patients and methods; a detailed description of adverse effects; loss to follow-up of less than 30%.

Strong evidence required at least two studies of high quality reporting the same result. Moderate evidence required at least one study of high quality and two of medium quality, while limited evidence required at least two studies of medium quality.

Data Abstraction

The data were independently abstracted by two reviewers and the authors were contacted in the event of questions. The number of participants, baseline characteristics, type of surgery, time of follow-up, design, benefits and adverse effects were extracted.

Statistical Analysis

Means and 95% confidence interval (CI) for change in active treated patients compared with control patients was calculated from mean of change (after - before) and standard deviation of change (SD). Differences of mean change between active treatment and control treatment were presented with 95% confidence interval (95% CI). A significant effect was regarded when 0 was not included in the confidence interval. For meta-analysis, we used Cochrane Review Manager software (version 5.0; Cochrane Library Software, Oxford, England). Means and SD for change in outcome was entered in the analysis. The weighted mean difference was used for comparisons of change in difference between active treatment and control. A P-value of < 0.05 was considered significant. Weighted means for adverse effects were calculated as = Σ n_i/n * p_i. Where n_i = number of treated patients, n = sum of all treated patients and p_i = proportion of patients with an adverse effect.

RESULTS

Efficacy

There were 64 hits in the search, and 15 potentially relevant articles were read. Five randomized controlled trials comparing surgery with sham surgery or conservative treatment were identified (Figure 1). One trial comparing uvulopalatopharyngoplasty and conservative treatment was excluded from further analysis due to low quality and a modified JADAD score of 2.¹⁶ Four parallel, randomized, controlled trials of high quality met the inclusion criteria, 2 studies investigated the effect of laser-assisted uvulopalatoplasty and 2 the effect of radiofrequency ablation (Table 1).^17–20

Figure 1.

Flowchart of the process of study selection.

Table 1.

Description of Included Studies of Medium and High Quality

Source	Location	Study design	Sample size*	Operation	Outcomes	Quality
Ferguson et al., 200317	Canada	RCT, parallel	46 (45)	LAUP	ESS, AHI, SAQLI, Snoring, Side-effects	High
Woodson et al., 200318	USA	RCT, parallel	60 (52)	TCRAFTA	ESS, AHI, SF36, FOSQ, Side-effects	High
Larossa et al., 200419	Spain	RCT, parallel	28 (25)	LAUP	ESS, AHI, SF36, Snoring, Complications	High
Stuck et al., 200520	Germany	RCT, parallel	26 (23)	TCRAFTA	ESS, Snoring, Side- effects	High
Studies of adverse effects
Esclamado et al., 198923	USA	Consecutive patients	132	UPPP	Perioperative complications	Medium
Harmon et al., 198924	USA	Consecutive patients	132	UPPP	Peri- and postoperative complications	Medium
Walker et al., 199625	USA	Consecutive patients	275	LAUP	Postoperative complications	Medium
Riley et al., 199726	USA	Prospective	182 (182)	Different operations	Postoperative complications	Medium
Mickelson et al., 199827	USA	Consecutive patients	347	UPPP	Postoperative complications	Medium
Terris et al., 199828	USA	Consecutive patients	109 (109)	UPPP	Peri- and postoperative complications	Medium
Hultcrantz et al., 199929	Sweden	Consecutive patients	55	UPP	Long-term side-effects	Medium
Levring-Jäghagen et al., 199930	Sweden	Consecutive patients	76 (68)	UPP	Persistent dysphagia	Medium
Powell et al., 199931	USA	Prospective	20 (18)	TCRAFTA tongue	Postoperative complications, swallowing function	Medium
Remacle et al., 199932	Belgium	Prospective	89	LAUP UPPP	Complications	Medium
Boudewyns et al., 200033	Belgium	Prospective	103 (44)	TCRAFTA soft palate	Postoperative complications	Medium
Coleman et al., 200034	USA	Prospective	12	TCRAFTA soft palate	Postoperative complications	Medium
Grontved et al., 200035	Denmark	Consecutive patients	69	UPPP	Complications	Medium
Hagert et al., 200036	Sweden	Consecutive patients	457 (415)	UPPP LAUP	Long-term side-effects	Medium
Hukins et al., 200037	Australia	Prospective	20	TCRAFTA soft palate	Postoperative complications	Medium
Li et al., 200038	USA	Prospective	22	TCRAFTA soft palate	Long term side-effects	Medium
Osman et al., 200039	UK	Prospective	47 (47)	UPPP LAUP	Postoperative complications	Medium
Bäck et al.. 200140	Finland	Prospective	21 (16)	TCRAFTA soft palate	Postoperative complications	Medium
Brown et al., 200141	Canada	Prospective	12	TCRAFTA soft palate	Postoperative complications	Medium
Pazos et al., 200142	USA	Consecutive patients	30 (30)	TCRAFTA soft palate, tongue	Postoperative complications	Medium
Sher et al., 200143	USA	Prospective	112 (105)	TCRAFTA soft palate	Postoperative complications	Medium
Terris et al., 200145	USA	Prospective	23	TCRAFTA soft palate	Postoperative complications	Medium
Woodson et al., 200146	USA	Prospective	73 (69)	TCRAFTA tongue	Postoperative complications	Medium
Haraldsson et al., 200247	Sweden	Prospective	16	TCRAFTA soft palate	Speech evaluation	Medium
Lysdahl et al., 200248	Sweden	Consecutive patients	150 (121)	UPPP LAUP	Long-term side-effects	Medium
Stuck et al., 200244	Germany	Prospective	20 (18)	TCRAFTA soft palate, tongue	Postoperative complications	Medium
Berger et al., 200349	Israel	Prospective	49	UPPP LAUP	Early and long-term side-effects	Medium
Rombaux et al., 200350	Belgium	Prospective	49	UPPP LAUP TCRAFTA soft palate	Postoperative complications	Medium
Said et al., 200351	USA	Consecutive patients	50 (39)	TCRAFTA soft palate	Postoperative complications	Medium
Jäghagen et al., 200422	Sweden	Prospective	46 (42)	UPPP UPP	Swallowing function	High
Kezirian et al., 200421	USA	Prospective	3130	UPPP	Postoperative serious side-effects	High
Kim et al., 200552	South Korea	Consecutive patients	90	UPPP	Postoperative complications	Medium
Li et al.,54	China	Consecutive patients	108	UPPP	Taste disturbances	Medium
Pavelec et al., 200653	Czech Republic	Prospective	63	LAUP	Postoperative complications	Medium

Abbreviations: RCT, randomized, controlled trial; ESS, Epworth sleepiness scale; SF36 Short Form 36; AHI, Apnea-hypopnea index; FOSQ, functional outcome of sleep questionnaire; SAQLI, Calgary Sleep Apnea Quality of Life Index; LAUP, laser-assisted uvulopalatoplasty; UPP, uvulopalatopharyngoplasty; UPPP, uvulopalatopharyngoplasty; TCRAFTA, temperature-controlled radiofrequency tissue ablation.

^*Numbers in parentheses denote the number of patients analyzed after withdrawals.

A summary of the effects on outcome for the difference in change is presented in Table 2. These studies show no significant effect on daytime sleepiness measured with the Epworth sleepiness scale after laser-assisted uvulopalatoplasty and temperature-controlled, radiofrequency tissue ablation (Figure 2). The apnea-hypopnea index and snoring was reduced in one trial after laser-assisted uvulopalatoplasty¹⁷ but not in another trial.¹⁹ Subjective snoring was reduced, but not the apnea-hypopnea index in one trial after radiofrequency ablation.¹⁸ There was no effect on quality of life in any trial. No trial investigating the effect of any other surgical modality met the inclusion criteria.

Table 2.

Effects of Surgery Compared to Placebo or No Treatment

	LAUP Ferguson et al., 200317	LAUP Larossa et al., 200419	TCRAFTA Woodson et al., 200318	TCRAFTA Stuck et al., 200520
ESS	−2.2 (−15, 9.3)	−0.4 (−13, 12)	−1.2 (−3.1, 0.8)	−0.6 (−1.2, 2.3)
FOSQ			−0.9 (−0.1, 1.9)
SAQLI	0.2 (−0.6, 1.0)
SF-36 Physical		1.4 (−38, 41)	−1.0 (−5.1, 3.1)
SF-36 Mental		−2 (−49, 45)	2.5 (−1.4, 6.4)
AHI	−10.5, P = 0.04	7 (−76, 90)	−2.7 (−9.9, 4.5)
Snoring intensity	−4.0 (−5.5, −2.5)*	−1.1 (−12, 10)
Snoring frequency	−3.6 (−1.3, −5.9)*	−106 (−8753, 8541)
Snoring subjective		−0.5 (−3.8, 2.8)		−2.5 (−4.6, −0.4)*

^*Significant as 0 was not included in the confidence interval.

Abbreviations: LAUP, laser-assisted uvulopalatoplasty; TCRAFTA, temperature-controlled radiofrequency tissue ablation; ESS, Epworth sleepiness scale; NS, No significant change. SF36 Short Form 36; FOSQ, functional outcome of sleep questionnaire; SAQLI, Calgary Sleep Apnea Quality of Life Index; AHI, Apnea-hypopnea index

Figure 2.

Effect of surgery on daytime sleepiness using the Epworth sleepiness scale.

Laser-Assisted Uvulopalatoplasty

Ferguson et al. randomized 46 snoring patients with an apnea-hypopnea index of between 10 and 27 to laser-assisted uvulopalatoplasty or no treatment and followed them for 7 months.¹⁷ There was no significant between-group difference in the Epworth Sleepiness Scale or the Calgary Sleep Apnea Quality of Life Index. The mean apnea-hypopnea index was reduced from 19 to 15 in the actively treated patients and increased from 16 to 23 in non-treated patients, P = 0.04 for the between-group difference in change. Snoring intensity estimated by a bedroom partner on a VAS scale from 0–10 was reduced by surgery from a mean of 9.2 to 4.8 and by no treatment from 8.9 to 8.5, P < 0.0001 for the between-group difference in change. Snoring frequency was reduced from a mean of 9.4 to 5.5 in surgically treated patients and from 8.8 to 8.5 in non-treated patients, P < 0.005.

Larossa et al. randomized 28 snoring patients with an apnea-hypopnea index < 30 to laser-assisted uvulopalatoplasty or sham surgery and followed them for 3 months.¹⁹ They found no significant difference in change in any of the outcomes, i.e. the Epworth Sleepiness Scale, Short-Form 36 Physical component summary, Short-Form 36 Mental component summary, Subjective snoring intensity, Snoring index or Decibels of snoring (Table 2).

Temperature-Controlled Radiofrequency Tissue Ablation

Woodson et al. performed the largest study and randomized 60 patients with daytime sleepiness, an apnea-hypopnea index of 5 to 40 and a body-mass index < 35 kg/m² to radiofrequency ablation or sham surgery.¹⁸ They report that radiofrequency ablation, compared with sham surgery, improved Functional Outcomes of Sleep Questionnaire with a mean of 0.9 (95% CI −0.1, 1.9) P = 0.04 and reduced the apnea index with a mean of −4.8 (−9.3, 0.4) P = 0.02. These P-values were, however, significant only in one sided t-test but not in 2-sided nonparametric test, and 0 was included in confidence interval (Table 2). They found no significant difference in change for the Epworth Sleepiness Scale, Short-Form 36 Physical component summary, Short-Form 36 Mental component summary or the apnea-hypopnea index.

Stuck et al. randomized 26 non-sleepy snorers with an apnea-hypopnea index < 15, a body-mass index < 35 kg/m² and a mean Epworth sleepiness scale of 8.2 (1.4) to temperature-controlled radiofrequency ablation or sham surgery.²⁰ There was no between-group difference in change in the Epworth Sleepiness Scale. The authors report that snoring estimated by a bedroom partner on a VAS scale from 0–10 was reduced by surgery compared with sham surgery from a mean of 8.1 to 5.2 and by sham surgery from 8.4 to 8.0, P = 0.045 for the difference in change.

Adverse Effects

There were 632 hits in the search and 102 potentially relevant articles were read. Seventy-nine trials met the inclusion criteria. Five of the included studies were of high quality,^{17,18,20–22} and 33 were of medium quality (Table 1).^19,23–54. Forty-one studies were of low quality and they were excluded from further analysis, except for 7 studies reporting life-threatening side-effects and deaths^55–95 (Figure 1).

Uvulopalatopharyngoplasty

Uvulopalatopharyngoplasty was followed by severe complications in the peri- and post-operative period, including death, bleedings and respiratory compromise, in 0% to 16% of patients, with higher frequencies in studies published during the 1980s and lower complication frequencies in studies in recent years.^{21,23,24,26–28},^{39,50,52,58,60,77,94}

A total of 30 cases of death were reported from six studies published between 1989 and 2004 (Table 3). One high-quality study comprising 3,130 operations reported peri- and post-operative death in 0.2% (95% CI 0.1 to 0.4) and serious complications other than death in 1.5% (95% CI 1.1 to1.9).²¹ Respiratory compromise, bleedings, intubation difficulties, infections, and cardiac arrest were the main causes of death. Persistent side-effects occurred in a mean (range) of 58% (42%–62%) (limited evidence; Table 4).^35,36 Difficulty swallowing, including nasal regurgitation, occurred in 31% (13% to 36%) of operated patients (moderate evidence), voice changes in 13% (7% to 14%) (limited evidence), and taste disturbances in 5% (1% to 7%) (limited evidence).^{22,35,36,48,54} Other reported side-effects were globus sensation, smell disturbances, and single cases of velopharyngeal insufficiency.^36,55

Table 3.

Deaths in the Peri- and Postoperative Period

Author	Operation	Death case numbers	Source population numbers	Reasons for death	Quality
Esclamado 198923	UPPP	1	135	Intubation problems	Medium
USA
Harmon 198924	UPPP	2	132	Bleeding (n = 1)	Medium
USA				Pulmonary embolism (n = 1)
Fairbanks 199057	UPPP	16		Bleeding (n = 1)	Low
USA				Loss of airway (n = 12)
				Unknown (n = 3)
Carenfelt 199358	UPPP	3	9000	Cardiac arrest (n = 2)	Low
Sweden	LAUP	1	2900	Bleeding (n = 1)
				Infection (n = 1)
Haavisto 199460	UPPP	1	101	Breathing difficulties and asystole (n = 1)	Low
Finland
Lee 199762	UPPP or	6		Respiratory arrest (n = 1)	Low
UK	LAUP			Cardiac arrest (n = 1)
				Unknown (n = 4)
Kezirian 200421	UPPP	7	3130	Unknown (n = 4)	High
USA				Respiratory problems (n = 2)
				Cardiac arrest (n = 1)

Abbreviations: UPPP, uvulopalatopharyngoplasty; LAUP, laser-assisted uvulopalatoplasty.

Table 4.

Side-Effects After Uvulopalatopharyngoplasty

	Author	Frequency	Weighted mean	Evidence grade
Persistent side-effects	Hagert 200036	62%	58%	Limited
	Grontved 200035	42%
Difficulty swallowing	Hagert 200036	35%	31%	Moderate
	Grontved 200035	13%
	Lysdahl 200248	36%
	Jäghagen 200422	29%
Voice changes	Hagert 200036	14%	13%	Limited
	Grontved 200035	7%
Taste disturbances	Hagert 200036	7%	5%	Limited
	Li 200654	1%

Uvulopalatoplasty

Uvulopalatoplasty performed with a scalpel or laser was followed by peri- and post-operative complications in up to 5%, including post-operative bleedings and infections, with one report of death from septicemia (Tables 3).^{17,19,25,39,50,58} Persistent side-effects were reported in a mean (range) of 59% (48% to 62%) (limited evidence; Table 5).^29,36,49 Difficulty swallowing including nasal regurgitation was reported in 27% (19% to 31%) (strong evidence) and globus sensation in 27% (16% to 36%) (limited evidence).^{17,22,29,30,32,36,48,53} Other reported side-effects included voice changes, smell disturbances, taste disturbances, and single cases of velopharyngeal stenosis.^36,49

Table 5.

Side-Effects After Uvulopalatoplasty

	Author	Frequency	Weighted mean	Evidence grade
Persistent side-effects	Hultcranz 199929	56%	59%	Limited
	Hagert 200036	62%
	Berger 200349	48%
Difficulty swallowing	Levring-Jäghagen 199930	20%	27%	Strong
	Hultcranz 199929	19%
	Hagert 200036	31%
	Lysdahl 200248	27%
	Ferguson 200317	19%
	Jäghagen 200422	29%
Globus sensation in throat	Hultcranz 199929	19%	27%	Limited
	Hagert 200036	36%
	Pavelec 200653	16%

Temperature-Controlled Radiofrequency Tissue Ablation

There were no reported changes in speech or swallowing one to two months after surgery.^18,20,38,47 Side-effects of temperature-controlled, radiofrequency tissue ablation of the soft palate included hemorrhage, infections and velopharyngeal fistula in single cases.^33,50 Radiofrequency ablation of the base of the tongue was followed by severe infections or tongue abscesses in 0% to 8% of patients (moderate evidence).^{18,31,42,44,46} Single cases of mouth floor edema, severe tongue swelling, and a case of pseudo-aneurysm of the lingual artery and heavy bleeding 14 days after surgery have also been reported.^42,44,46,92

Other Surgical Modalities

No studies of any other surgical modality meeting the inclusion criteria for adverse effects were found.

DISCUSSION

Randomized, controlled trials do not support any benefit from surgery in the form of laser-assisted uvulopalatoplasty and temperature-controlled radiofrequency tissue ablation when it comes to daytime sleepiness or quality of life. Studies do not provide any evidence of effect on apnea-hypopnea index or snoring as there is a need of at least two trials reporting effect for evidence grading. One study reported a reduction in the apnea-hypopnea index and snoring after laser-assisted uvulopalatoplasty, while no such effects were found in another trial. A single study reported a reduction in snoring after radiofrequency tissue ablation. No evidence of effect was established for any other surgical modality, as no randomized, controlled trial meeting the inclusion criteria was found for any other surgical modality. Persistent side-effects were reported in a majority of the patients undergoing uvulopalatopharyngoplasty and uvulopalatoplasty, especially in the form of difficulty swallowing, globus sensation in the throat, and voice changes.

Only four controlled trials randomizing a total of 141 patients to either laser-assisted uvulopalatopharyngoplasty and sham/no treatment, or temperature-controlled radiofrequency tissue ablation and sham surgery met the present inclusion criteria. From this small sample of trials, it is not possible to exclude any effect of surgery. However, the many reports of swallowing difficulties after uvulopalatopharyngoplasty and uvulopalatoplasty suggest that other surgical modalities than these two should be further tested. Tonsillectomy is one suggested modality as large tonsils are associated with sleep apnea.⁹⁶ One disadvantage of surgery, especially more invasive surgery, is the problem of performing placebo or sham surgery in the control group. Instead of placebo, we suggest that control patients should be randomized to delayed surgery. In subsequent systematic reviews we suggest that other quality criteria than pure JADAD should be used in quality assessments. We used a modified JADAD focusing on the quality of single blinding instead of double blinding. If patients and surgeons cannot be blinded, one can always try to blind the assessors.

To date, the best study of peri- and post-operative mortality was published by Kezirian et al. in 2004. They reported a mortality of 0.2% among all Veterans Affairs inpatient uvulopalatopharyngoplasty surgery from 1991 to 2001 and serious peri- and post-operative complications in 1.5% of patients.²¹ However, without evidence of effects of the surgical procedures, it is difficult to accept mortality related to surgery. Two cohort studies followed long-term mortality after uvulopalatopharyngoplasty and they report low mortality in the long term for these patients compared with untreated patients and patients on CPAP.^97,98 These studies were, however, not randomized controlled trials, and surgically treated patients were younger and had less comorbidity than those treated with CPAP.

This systematic review included only randomized, controlled trials of medium and high quality regarding the efficacy of surgery. The lack of evidence of benefits of surgery is in agreement with a Cochrane review¹¹ and recent reviews by Elshaug et al.^12,14 Only a few randomized controlled trials were, however, available, even though surgery for obstructive sleep apnea was introduced more than 25 years ago. To evaluate adverse effects, we included observational studies of medium and high quality in accordance with the GRADE working group.⁹⁹ When attempting to estimate the risk of side-effects, sources other than randomized trials have to be searched, since randomized controlled trials are seldom designed to evaluate the risk of side-effects. Observational studies may provide better evidence of rare adverse effects and well-designed case series in particular may provide high-quality evidence of the complication rates for surgery.⁹⁹ These are the main reasons why different inclusion criteria were used for studies of efficacy and studies of side-effects. With the present design, it was found that the evidence of harm was greater than the evidence of benefits in relation to surgery on the soft palate involving the removal of the uvula, i.e., uvulopalatopharyngoplasty and uvulopalatoplasty.

In conclusion, only a small number of randomized controlled trials with a limited number of patients assessing some surgical modalities for snoring or sleep apnea are available. These studies do not provide any evidence of effect from laser-assisted uvulopalatoplasty or radiofrequency ablation on daytime sleepiness, apnea reduction, quality of life or snoring. We call for research of randomized controlled trials of surgery other than uvulopalatopharyngoplasty and uvulopalatoplasty, as they are related to a high risk of long-term side-effects, especially difficulty swallowing.

DISCLOSURE STATEMENT

This was not an industry supported study. The authors have indicated no financial conflicts of interest.