gms | German Medical Science

After the introduction of the classical Laser-assisted Uvulopalatoplasty (LAUOP) by Kamami in 1990 [1] and the Uvulopalatopharyngoplasty (UPPP) by Ikematsu in 1963 and its modification by Fujita in 1984 [2], both techniques became increasingly widespread in the field of sleep medicine. Although both treatment strategies presented a fundamental expansion of the therapeutic spectrum and introduced ENT surgeons into sleep medicine, both techniques were merely modifications of well-established surgical procedures in the oropharynx. The introduction of temperature-controlled radiofrequency surgery of the soft palate for primary snoring in 1998 [3] and of the tongue base for obstructive sleep apnea in 1999 [4] by Powell et al. awakened the interest of the technical industry. Given the high prevalence of these phenomena (according to current literature, 7 to 9% of the middle-aged male population is affected by obstructive sleep apnea and 20 to 30% by primary snoring [5], [6], [7], [8]), a huge potential market opened up literally over night. Although most of those early promises in terms of clinical success and economic profits could not be kept, the lively interest of our industrial partners in this field is reflected in their impressive presence on congresses and meetings. Moreover, the activities and marketing strategies of often newly founded companies explains at least in part the evolving public interest in this field. Nowadays, interviews, advertisements and articles on snoring and sleep apnea may be found daily in newspapers and magazines.

This development is also reflected in the increasing number of scientific publications in the field. While only about 50 articles on surgical treatment of sleep disordered breathing were published in 1990, a MedLine survey for 2004 reveals over 200 corresponding articles. Last but not least, scientific sessions on sleep medicine have been major attractions during recent national and international congresses and meetings in the area of Otolaryngology.

Nevertheless, sleep medicine is part of an interdisciplinary field, and scepticism is still prevalent in associated disciplines such as pneumology and neurology. The increasing number of publications does not hold for high ranked interdisciplinary journals such as "Sleep" or "The Journal of Sleep Research". To date, only 8 publications on surgical treatment of sleep disordered breathing can be found in these journals. Although it has been assumed otherwise, this is not due to the suppression of surgical opinions or developments by conservative disciplines but mostly a result of limitations in the design and methodology of the majority of those "surgical" publications. A recent guideline of the German Society of Sleep Medicine therefore comes to the conclusion that the majority of surgical treatment alternatives can not be recommended in the treatment of sleep disordered breathing [9].

Unfortunately, surgical procedures have often been used without the necessary precautions and without carefully respecting indications and contraindications. This has led to a further increase in the scepticism mentioned above. As a result, instead of a potentially fruitful cooperation of surgical and conservative disciplines in the field of sleep medicine, these concerns have led to polarization and competition, attitudes that do not seem appropriate in regard to the complexity of the disorder we all want to treat.

In the last 20 years, numerous articles have been published on the classical surgical treatment options such as LAUP and UPPP. Nevertheless, it has to be said that for the majority of the remaining surgical techniques that are currently under use, sufficient scientific data is lacking. The limited number of existing publications is based on retrospective studies or prospective case series; furthermore, the number of subjects included in these trials is usually quite small and inhomogeneous. Furthermore, serious limitations in trial design often complicate the analysis of the data: different surgical techniques are often combined or added to the standard treatment (mostly UPPP). As a result, several surgical treatments have as yet not been evaluated as a single approach. The greatest concern, however, is raised by the lack of controlled clinical trials and long term follow up.

The vast majority of the current literature addressing surgical treatment of sleep disordered breathing consists of non-controlled clinical trials. Randomised controlled clinical trials are an exception. According to a current systematic review on surgical treatment of sleep disordered breathing as published by the Cochrane Collaboration in 2005, only 8 publications met the criteria of a controlled clinical trial [10].

Asking for controlled clinical trials is not just harping on principles. Especially in the field of sleep disordered breathing significant confounders have to be kept in mind that can only be addressed with a controlled study design. Factors such as the variability in overnight recording conditions lead to the significant and well-established night-to-night variability in polysomnographical recordings. The severity of the reported nightly breathing disorders depends on a multiplicity of factors such as sleep stages, body position and recorded sleep time - factors that hardly can be controlled or standardised in a clinical trial. This explains the high variability in overnight recording seen even without external intervention. With regard to objective respiratory parameters such as the apnea-hypopnea-index (AHI), variations of up to 20% are seen regularly from night to night [11]. Only a limited number of trials have attempted to address this, at least in part, by increasing the number of pre- and post-interventional recordings.

The variability mentioned above has been demonstrated for two consecutive nocturnal recordings. In clinical trials evaluating surgical treatments, several weeks or months usually lie between the pre- and postoperative measurements. This leads to an even higher variability of the sleep recordings regardless of surgical intervention. In this context, a highly interesting paper has been published by Ferguson et al. [12]; it will be discussed in detail in this review. In their publication, they presented sets of measurement before and after treatment (LAUP) as well as the corresponding data of an untreated control group. Figure 1 [Fig. 1] clearly demonstrates the impressive variations of the data that has to be kept in mind.

The variability discussed refers to objective, polysomnographic parameters only. Subjective parameters such as daytime sleepiness or quality of life are influenced by a multitude of additional factors; without a control group, such subjective variables are even harder to assess and virtually impossible to interpret.

Nocturnal breathing disorders present a dynamic group of diseases undergoing a constant change over time which is not always easy to predict. In general however, increasing respiratory parameters indicate a deterioration over time in the majority of cases even without an accompanying increase in body weight. Pendlebury et al. were able to demonstrate a significant increase in the AHI over a period of only 17 months in a group of sleep apnea patients [13]. Affected patients expect beneficial effects of therapeutic interventions to last for a reasonable period of time. While for continuous positive airway pressure (CPAP) long-term data demonstrates that beneficial treatment effects last for a long period of time, long-term efficacy has been demonstrated for only a minority of surgical treatments.

For obstructive sleep apnea, long term data has been presented at least for UPPP. For obstructive sleep apnea (and for primary snoring), follow-up periods of up to 10 years have been published [14], [15]. In a recent meta-analysis by Verse et al. based on studies with short and long term follow up, a reduction of the initial success rates from 66 to 50% could be demonstrated when UPPP is used to treat obstructive sleep apnea [16]. Even for the most common combined approach in the USA - genioglossus advancement together with hyoid suspension - a deterioration of success rates from 78 to 65% has been seen over time [17].

The assessment of therapeutic success in primary snoring is especially problematic, and the results of different studies are generally difficult to compare. Based on subjective data, medium or long term success rates have been published for LAUP and radiofrequency surgery of the soft palate only (although these data do not exceed a period of three years). In accordance with the data presented for obstructive sleep apnea, a deterioration (or a relapse) of snoring was reported over time at least for a subgroup of patients [18], [19], [20].

For all other surgical treatment options such as the uvulopalatal flap, soft palate implants, radiofrequency surgery of the tongue base, hyoid suspension or genioglossus advancement, data concerning postoperative success are limited to follow-up periods of 1 year. The only exception is maxillo-mandibular advancement surgery, where treatment effects seem to be basically stable even over a longer period of time [21].

Since its introduction into modern medical science in 1996 by Sackett [22], the idea of an "evidence based medicine" has become more and more popular and has fundamentally changed our approach to medical research. The evaluation of the efficacy of medical treatment is now no longer based on experience or personal opinion alone; clinical studies and systematic analysis of scientific data is now considered to be something like the gold standard. Specific criteria have been defined to categorise scientific literature into distinct levels of evidence, the placebo-controlled randomised clinical trial being at the top of a hierarchic system.

Well designed controlled trials are vital in medical research; nevertheless, they are not always feasible, especially when it comes to surgical treatment. For surgical interventions, a placebo group is possible in selected cases only, and randomisation or blinding of study participants or investigators can not always be performed or raises ethical concerns. For the most part, this is commonsensical, but it often leads to a serious misunderstanding. The idea of evidence based medicine is not restricted to the standard of randomised controlled clinical trials; rather, it wants to provide the highest possible evidence in every case [22]. A lack of controlled clinical trials is not equivalent to a lack of efficacy. On the other hand, methodological or ethical concerns with controlled clinical trials in surgery can not absolve us from critically evaluating how to incorporate ideas and methods of evidence based medicine in every study planned. The fact that a randomised controlled clinical trial may not be performed in one case or the other does not mean that controlled clinical trials can not be performed at all. Well designed clinical trials with high or even highest levels of evidence are also possible in the field of surgery for sleep disordered breathing, as the following examples will demonstrate.

The standard approach for a controlled clinical trial is to provide a non-treated control group. To diagnose but not to treat a distinct subgroup of patients with a condition requiring medical care may raise ethical concerns. To address this, the potential risk that non-treated patients may have to face needs to be discussed. This ultimately leads to questions about the potential risks of non-treated patients with sleep disordered breathing. For primary snoring, which is basically defined by socially disruptive snoring without an associated disruption of sleep and the absence of an increased cardiovascular risk, this question is easy to answer and ethical concerns hardly exist. In cases of obstructive sleep apnea however, daytime symptoms and especially daytime sleepiness need to be kept in mind. Furthermore, the increased cardiovascular risk of these patients requires attention [23]. This risk however is basically a long-term concern, as long as no manifest cardiovascular diseases are present that require immediate treatment. We therefore believe that postponing the onset of any kind of treatment due to an enrolment into a controlled clinical trial by several weeks or a limited number of months does not lead to a relevant danger of these patients. This potential delay furthermore needs to be discussed with regard to the delay that occurs in all sleep apnea patients these days due the increasing shortage in sleep laboratory facilities - a shortage that has been voluntarily created by political guidelines and health care restrictions.

Taking all this into account, there are no relevant concerns in performing controlled clinical trials even in sleep apnea patients, as long as patients with an excessive daytime sleepiness or patients with present cardiovascular diseases are excluded. Patients with mild to moderate sleep apnea appear to be most suitable in this context.

That controlled clinical trials following these recommendations are indeed feasible is demonstrated by a publication of Ferguson et al. from 2003 regarding LAUP [12].

In this study, 45 patients with mild to moderate obstructive sleep apnea were included (AHI10-27) and randomly assigned to two study groups. One study group received repeated LAUP, the second one served as a non-treated control group. In both groups, a control polysomnography was performed 6 months after the enrolment into the trial. While the AHI was reduced in the actively treated group from 18.6±4.3 to 14.7±7.5, which was statistically significant, an increase in AHI was seen in the control group from 16.1±4,0 to 22.7±15.2. LAUP appeared to be superior to a non-treatment, although it remains unclear why a direct comparison on the basis of normalised values has not been performed in this study.

Although the changes seen in AHI due to LAUP were regarded as clinically not significant and LAUP is not recommended for obstructive sleep apnea [24], this study gives us valuable information concerning the high variability in respiratory parameters in a non-treated group.

Ethical concerns with controlled clinical trials using non-treated control groups have already been discussed. Those who want to avoid these discussions or who want to be on the "safe side" may prefer controlled clinical trials comparing the treatment that requires clinical testing with the standard therapeutic approach. This would furthermore be in line with the 5^th revision of the Declaration of Helsinki. This revision, addressing pharmacological therapy in the first line, generally speaks against clinical trials using placebo, but advocates clinical trials using standard treatment in the control groups, to minimize potential hazards for the study subjects [25].

In obstructive sleep apnea, treatment with CPAP can serve as the gold standard to be compared with; intraoral devices may also be used for the control groups in selected cases, as for both treatment options sufficient evidence regarding treatment efficacy is available. As early as 2003, Woodson et al. provided a clinical trial comparing radiofrequency surgery of the soft palate and the tongue base with CPAP in a randomised controlled clinical trial [26].

In this trial, again patients with mild to moderate obstructive sleep apnea were included and randomised to the treatment groups (30 subjects in each arm). While patients undergoing radiofrequency surgery received a total number of 5 treatment sessions for the tongue base and 2 treatment session for the soft palate, the patients in the control group underwent adjustment of CPAP therapy in three unmonitored nights using an automatic CPAP device (the effective pressure was set according to the analysis of the automatic CPAP devices). Primary objectives in this study were reaction time and associated quality of life assessed with standard questionnaires (FOSQ, SNORE25). In both study groups a significant improvement in these parameters was detected; furthermore, no significant differences were seen between radiofrequency and CPAP therapy.

The study has obvious methodological limitations. Most important, no respiratory parameters were assessed. This seems particularly problematic as no objective assessment of treatment efficacy was provided in the CPAP group and CPAP adjustment was only based on self adjusting automatic CPAP devices used under unmonitored conditions. Nevertheless, this was the first study to demonstrate on a high level of evidence that surgical treatment may lead to equivalent effect as standard treatment at least in regard to subjective parameters.

Woodson and his co-workers even went one step further in their trial, as they used a third arm receiving "placebo-CPAP" consisting in an ineffective treatment pressure. This kind of "treatment" is often referred to as "sham-CPAP".

Regardless of the limitations discussed, this study demonstrates possible ways of how to overcome the lack of evidence in surgical trials in the field of sleep disordered breathing and shows us potential solutions for how to perform well designed trials leading to high levels of evidence.

A non-treated control group may be sufficient to assess objective changes under study conditions in a controlled clinical trial. Nevertheless, non-treated control groups are not suitable if subjective parameters need to be assessed, where significant placebo effects have to be expected. In these cases, placebo-controlled trials are necessary. As discussed above, a sham-CPAP treatment may serve as a placebo control in studies investigating nocturnal ventilation therapy. For surgical treatment alternatives, placebo controls are difficult to establish, but not an impossibility, as a recent trial from our department demonstrates, which was published in "Sleep" in 2005 [27].

Aim of this study was to assess the efficacy of radiofrequency surgery of the soft palate in the treatment of snoring. Especially in the treatment of snoring, surgical success is usually based on subjective assessment in terms of questionnaires, which makes a placebo control particularly important. For placebo, an ineffective soft palate radiofrequency "surgery" was performed by inserting the device needle under local anaesthesia without the application of radiofrequency energy. Twenty-six patients were enrolled and randomised to the two treatment groups receiving two treatment sessions. Surgery was performed by a co-investigator; the investigator himself was not informed about the status of his patients (surgery/control), fulfilling the conditions of an observer blind study.

The results of the trial are summarised in Figure 2 [Fig. 2].

The postoperative snoring scores in the actively treated group clearly demonstrates that radiofrequency surgery alone leads to a moderate improvement but not to a complete remission of snoring, and the reduction in snoring that was achieved in this study can hardly be regarded as sufficient. Radiofrequency surgery of the soft palate should be indicated with care. Nevertheless, the differences between the two groups with regard to normalised values demonstrates that radiofrequeny surgery is statistically significantly better than placebo. Interestingly, this is the first study ever to demonstrate on the highest level of evidence that a surgical treatment strategy used for primary snoring is significantly better than placebo.

Placebo-controlled trials using surgical treatment strategies for sleep disordered breathing may be suitable in those cases in which only minimal postoperative pain may be expected and where no visible anatomic changes occur. The study discussed therefore can serve as a model of how to test other minimally invasive treatment options such as injection snoreplasty [28] or the recently presented soft palate implants [29]. Furthermore, three armed trials may be performed comparing surgical treatment such as LAUP with radiofrequency surgery of the soft palate and placebo radiofrequency surgery. Multiple options may arise from this in the future.

Obstructive sleep apnea is a complex disorder affecting multiple organ systems. For the patients themselves, the subjective complaints are of foremost interest. This is what makes them see a doctor and is what they primarily seek to be relieved of. Patients´ major complaints are not necessarily reflected by their doctors' concerns. This is particularly true for primary snoring, being socially disruptive and thereby affecting quality of life, but not being a relevant hazard to health per definition.

Doctors primarily pay attention to the number and frequency of nocturnal respiratory events, that is, the apnea-hypopnea-index (AHI). Hardly any publication dealing with obstructive sleep apnea can be found that is not related to AHI in one or another way. Nevertheless, the AHI is first of all a surrogate parameter, in the same way as blood pressure or blood sugar serves in general medicine. Reducing the cardiovascular risk should be the focus of the therapeutic event, and not just the reduction of a respiratory parameter.

The AHI is an index based on the total sleep time, at least as long it is recorded with the help of a polysomnography. Undoubtedly it is a respiratory index of major importance, but it is certainly not sufficient to describe obstructive sleep apnea. An AHI of 15, exclusively consisting of hyponeas and minor desaturations, may not require any therapeutic intervention at all when it is presented by an otherwise healthy young patient. On the other hand, the same AHI, being accompanied by massive desaturations, may be of major relevance in a patient with severe cardiovascular morbidity, or when it is deteriorating an existing heart failure and therefore may require nocturnal ventilation therapy with oxygen supply.

In epidemiological trials, an increased cardiovascular risk can be detected from an AHI of 5 on, although an AHI of 5 is generally not sufficient for a diagnosis of obstructive sleep apnea or not regarded as an indication for nocturnal ventilation therapy [23]. In such trials, other associated risk factors accompanying obstructive sleep apnea serve as confounders and make it difficult to interpret these results. Interestingly, the adjusted risk for coronary heart disease starts at an AHI of below 10, but reaches a plateau and does not increase further even with higher AHI [23]. Numerous questions regarding obstructive sleep apnea as a independent risk factor are still not answered, as much as numerous questions regarding potentially beneficial effects of ventilation therapy need further investigation.

The beneficial effects of any therapeutic intervention in obstructive sleep apnea however have to be proven not only by significant reduction in surrogate parameters, but most importantly by a reduction in cardiovascular risk. For nocturnal CPAP therapy these kind of studies have been conducted [30], [31]. Doherty et al. could demonstrate in their trial that nocturnal ventilation therapy in obstructive sleep apnea can reduce the number of deaths due to cardiovascular events in patients with pre-existing cardiovascular diseases [30]; nevertheless, they could not demonstrate a significant reduction in new onsets of hypertension or cardiovascular diseases in general. Interestingly, in this study patients with CPAP incompliance served as the control group, leading to a systemic selection bias in this data.

For surgical treatment strategies, the data available is scarce. Furthermore, the few studies existing have substantial methodological flaws [32], [33], [34]. Nevertheless, no significant differences has been seen in these trials between conservative (CPAP) or surgical (UPPP) treatment with regard to cardiovascular morbidity.

Instead of a competition between surgical and conservative treatment in reducing simple indices of respiratory events, it should be time to assess potential long term benefits of surgical treatment alternatives with regard to cardiovascular morbidity. Undoubtedly, such studies are difficult to undertake - nevertheless, this should not be misused as an excuse for not performing them. In internal medicine, such trials have been routinely applied and performed for decades. Nobody would accept the evaluation of a new antihypertensive drug simply by reported effects on blood pressure without showing data on its effects on cardiovascular morbidity and mortality.

Excessive daytime sleepiness is a common but not obligatory symptom of obstructive sleep apnea. The extent of the associated daytime sleepiness shows a correlation with objective measurements such as the AHI [35], but does not directly depend on it in the individual case. Often, daytime sleepiness as a symptom requires adequate treatment. In mild sleep apnea daytime sleepiness itself may be the primary target of therapeutic intervention. In these cases, the effects on daytime sleepiness should receive more attention when evaluating different treatment options.

An improvement in daytime sleepiness can regularly be found even in those cases in which relevant effects on objective measurements could not be detected [36]. It seems obvious that a placebo effect may be the reason for this improvement in a relevant number of cases. This - again - leads to the conclusion that well designed studies need to be initiated when assessing daytime sleepiness. In the controlled trial of Woodson et al. [26] discussed above, surgical treatment consisting of combined radiofrequency surgery was equivalent to ventilation therapy in relation to the effects on daytime sleepiness.

Since its introduction by Sullivan in 1981 [37], nocturnal ventilation therapy with continuous positive airway pressure has been regarded as the gold standard in the treatment of obstructive sleep apnea. The fundamental technical improvements lead to smaller and more comfortable devices and to a constant reduction in acoustic emission. Together with the development of different types of facial masks and air humidifiers this has increased patients´ acceptance significantly. Alternative treatment modalities have to compete with this standard treatment today and in the future. Nevertheless, despite its undoubted advantages, the limitations of nocturnal ventilation therapy need to be addressed when comparing the different treatment options. The factors "primary acceptance," "compliance," and "regular use" will be discussed in the following.

Data concerning the compliance with CPAP therapy are given in numerous publications [38], [39], [40], [41], [42]. Nevertheless, for various reasons the compliance rates differ substantially between 46 [39] and 85% [41]. First of all, the compliance rates have been assessed with different groups of patients in relation to the severity of the disease and daytime sleepiness, both affecting compliance with CPAP therapy [40], [41]. Patients with excessive daytime sleepiness prior to therapy and with a relevant improvement under ventilation therapy are more likely to continue nocturnal ventilation [41]. This leads to the fact that data concerning compliance with CPAP therapy will be different for severely or only marginally affected patients.

Furthermore, patient care and coaching is essential. If patients are discharged the morning after nocturnal polysomnography without a detailed explanation of the findings, and the results and consequences of the nightly measurements are given in a brief letter only, followed by a package with a CPAP device and a standard mask, the expected compliance will be low (it should be kept in mind that a scenario like this is already a reality with several insurance companies in the U.S.) [43]. Better compliance rates may be achieved in those cases where detailed explanations and demonstrations are given for the individual case and where persons to turn to are provided for the patients. It has already been demonstrated in clinical trials that a standardised instruction with a video tape has beneficial effects on compliance rates [44], [45].

A further increase of patient compliance significantly exceeding the best ones reported may hardly be achieved even with further technical improvement. Even the additional support with air humidifiers or with auto-titrating CPAP devices has not improved compliance in recent reviews [44], [46], [47].

In this context, the definition of "regular use" defining compliance should be discussed in detail. The easiest but least reliable way of assessing compliance is to ask patients whether they use their CPAP regularly. In recent publications the data therefore was based on built-in counters of running time of the new generation CPAP devices. But even with more accurate measurements, the results of the studies should be interpreted with care. "Regular use" or "compliance" are defined differently. While some authors regard a use of 3.5 hours per night as regular use [41], other set the limit to 5 hours per night [48] or more than 4 hours per night on more than 5 days a week [39], [49]. Regardless of the fact that, without evidence-based trials to refer to, the average person sleeps 7 days a week, data is given concerning the average hours of sleep per night in the adult population, which lies around 7 hours per night [50], [51]. The reasons for not using this information for a reasonable definition of regular use remains unclear.

Data for compliance rates can only be provided for those patients who primarily accept CPAP and where nocturnal ventilation is started at all. No compliance data can be provided for those refusing the use of CPAP. Only scarce information is given concerning primary acceptance rates in the literature [52]. Nevertheless, it has been reported that 8 to 15% of the patients refuse further use of CPAP after the first night [42], [53]. Primary acceptance may furthermore vary with regard to the spectrum of treatment alternatives that are offered to the patients. When CPAP is offered exclusively, primary acceptance rates will naturally be higher.

We are all used to adjusting nocturnal ventilation therapy under laboratory conditions and to set the CPAP device on an adequate pressure the next morning according to the nightly measurements. In this way, the treatment is defined as being effective based on the nocturnal findings in the sleep lab. In my opinion, the definition is questionable. A patient may be treated effectively in this single night under those artificial nightly conditions, but whether the patient is being treated effectively in the long run can hardly be assessed in this way.

Similar problems were faced in the treatment of diabetes for decades. Patients were set on a treatment scheme with insulin, e.g., under hospital conditions and after a period of modifications and adjustments an effective treatment scheme was defined. That this scheme may not be effective under practical, day to day conditions was and is obvious, and regular check ups and adjustments by the general practitioner were required. This problem was partially solved by the discovery of the HbA1c in the 1970s [54]. Since its introduction a laboratory parameter has been available to assess long term efficacy of the different treatment schemes for diabetes. A comparable tool for sleep disordered breathing would be extremely helpful for the monitoring of patients with obstructive sleep apnea and would give us an idea about the long term efficacy of the treatments available under practical conditions and not only under one experimental night in the sleep lab. Parameters of oxidative stress may offer such options in the future.

The problems discussed in regard to compliance and regular use make it difficult to compare the efficacy of different treatment strategies. In an exciting publication of Maurer et al. it was attempted to overcome these problems with a novel and innovative approach [55].

In this retrospective study based on a large study sample of over 250 patient the efficacy of nocturnal CPAP therapy was compared to surgical intervention. The efficacy of the treatments selected was assessed 6 months after surgery or CPAP adjustment with one night of polysomnography. The respiratory parameters in the CPAP group as measured with control-polysomnography were adjusted for the individual compliance data according to the built in counters of the CPAP devices in regard to their running time.

These adjustments were performed under the assumption that the mean number of respiratory events per hour reflects the severity of the disease (in this way, the AHI has been defined as the events per night divided by the total sleep time in hours). If a patient presents with a baseline AHI of 40 which could be reduced by CPAP to a level of 10 but he only is using his CPAP during the first half of the night, a mean AHI of 25 would be calculated for this patient.

Whether this concept of assessing treatment efficacy for CPAP reflects the pathophysiology of obstructive sleep apnea remains unclear - but so does the entire concept of assessing a mean AHI over night. It should at least be kept in mind that a cluster of apneas with severe desaturation in selected periods at night leading to a low mean AHI (as it may be found in positional or REM-depending OSA) may be more disrupting or severely affecting the cardiovascular system as an identical AHI based on a random distribution of events over the entire night. A patient under "effective" CPAP treatment terminating its use in the morning hours as regularly reported by the patient) will suffer from the initial severity of his disease with the full range of adverse effects from this moment onwards. A patient who underwent only partly effective surgical intervention may suffer from residual disease over the entire night. Even if the mean (adjusted) AHI is comparable in both patients, the equivalence of both treatment effects remains questionable. An advantage for surgical therapy may result from this concept.

This conclusion is obviously highly speculative and not supported by study results. Nevertheless, it could explain the striking subjective improvement of patients without accompanying objective improvement which is often seen in surgically treated patients.

Surgical treatment strategies for sleep disordered breathing remain controversial. Although they are receiving increasing attention by the patients and are often regarded as an alternative for nocturnal CPAP treatment they can hardly be found in national or international guidelines. Those doctors who primarily treat sleep disorders conservatively have every right to point out the limitations in the data available and the methodological concerns connected with the vast majority of the studies published. With regard to the principles of evidence based medicine allocating the use of studies with the highest available evidence we should increase our efforts to provide well designed studies with the best level of evidence possible. Even if placebo controlled studies are only feasible in selected cases, recent publications demonstrate that well designed controlled trials are possible and can successfully be performed. Those centres focusing on surgical treatment need to compare their results with those of standard treatment with regard to long term efficacy in controlled clinical trials.

On the other hand, the conservative treatment still has its limitations. The problems with primary acceptance, long term compliance and the monitoring of the treatment effects under everyday practical conditions are still unsolved. Studies comparing different treatment interventions should respect these limitations. Sleep disordered breathing is a complex disorder; therapeutic effects cannot be simply measured by a reduction of surrogate parameters of a single night under laboratory conditions. Further research needs to respect subjective complaints and long term effects on cardiovascular morbidity more thoroughly.

The competition between conservative and surgical treatment and their representatives does not reflect the complexity of the disorder we want to treat. It is time for an open and critical discussion of the treatments available.