However, the success rate of sleep surgery remains moderate. A recent systematic review of sleep surgery based on the Apnea-Hypopnea Index (AHI) demonstrated significant variability [4]. Elshaug et al. [5] suggest redefining the success rate of airway surgery for OSA because the AHI levels used in surgical success criteria are higher than those for CPAP. Applying the same AHI criteria to both CPAP and sleep surgeries could lead to a perceived lower success rate for upper airway (UA) surgery, potentially making them less favorable. Nevertheless, numerous studies have confirmed the efficacy of surgical treatments for OSA. A multicenter, prospective, longitudinal study showed that UPPP significantly improves disease-specific quality of life and sleep apnea symptoms in OSA patients [6]. Another study comparing CPAP to sleep surgery in veterans with sleep apnea found that UPPP had a higher survival rate than CPAP (96.6% vs. 92.9%) after adjusting for age, gender, race, year of treatment, and comorbidity [7].

One reason for these differing outcomes could be the definition of CPAP compliance, usually accepted as using CPAP for with a minimum 4 hours of PAP usage per night for 70% of the nights. Using a mathematical function formula, Ravesloot and de Vries [8] demonstrated that the greater the severity of OSA, the more the use of CPAP is required to significantly reduce the AHI less than 5 or 10. They proposed that the mean AHI during CPAP therapy might be a more realistic index, and arbitrary compliance criteria might conceal insufficient reductions in AHI. Consider a patient with an AHI of 60/hour who sleeps for 8 hours per night for 30% of nights in a week and use CPAP, achieving an AHI of 5, for merely 4 hours per night for 70% of nights in a week (AHI=32.5/night). Despite an average AHI of 40.35/hour across all nights of a week, this patient would still be classified as “compliant” according to the current compliance definition. For a patient who underwent sleep surgery, with a postoperative AHI of 24 and a sleep duration of 8 hours per night, the average AHI across all nights of a week would be 24. Since developed by C. Guilleminault and W. Dement, AHI has been a dominant index for sleep apnea and used across nearly every field of sleep medicine.

However, AHI essentially quantifies the number of episodes of apnea and hypopnea per hour and is a simple frequency metric. Thus, it fails to reflect the duration or severity of abnormal respiratory episodes during sleep and cannot accurately measure the true severity of respiratory events. The oxygen desaturation index (ODI), another alternative, also represents the number of desaturation episodes (more than 3% or 4%) and so does not account for the duration or severity of abnormal respiratory episodes. Therefore, treatment outcomes based on AHI or ODI may be misleading because these indices do not capture the full extent of the total hypoxic burden (HB) caused by abnormal respiratory episodes, nor do they adequately reflect the patient’s symptoms or the potential risk of related complications such as cardiovascular disease, stroke, and cognitive decline.

Asano et al. [9] demonstrated that their new index, the integral area of desaturation, evaluated cardiovascular risk more effectively than AHI alone in patients with mild to moderate OSA but not in those with severe. HB is typically defined as the area under the oxygen desaturation curve relative to a pre-event saturation baseline, expressed as the total desaturation area (%minute) per hour of sleep in OSA patients. Recently, Azarbarzin et al. [10] reported that HB more accurately predicted incident heart failure in men than in women, whereas AHI predicted heart failure less effectively and less precisely than HB. From the perspective of HB, surgical treatment often reduces the severity of OSA by converting more severe apnea events into milder hypopneas. However, this improvement is not reflected in the AHI, as AHI assigns the same weight to both apneas and hypopneas. Moreover, it is important to note that positional dependency may change following UA surgery. The prevalence of positional OSA has been reported to increase after UA surgery and may be associated with persistent OSA postoperatively, potentially contributing to the low or moderate success rates of UA surgeries. Although subjective sleepiness scales and quality of life questionnaires are useful in assessing the burden and consequences of OSA, their results may not be accurate and can vary depending on the timing of assessments.

In summary, it might be unfair to sleep surgeons to assess postoperative results solely based on AHI (respiratory events per total sleep time on polysomnography), especially when CPAP treatments are considered compliant with a minimum 4 hours of PAP usage per night for only 70% of the nights, regardless of the total sleep time.

While all sleep physicians and surgeons recognize that AHI is the most critical index in sleep medicine, they also agree that it does not capture all aspects of total HB during sleep or the overall physiological improvements achieved through surgery. Therefore, when evaluating post-treatment results based on AHI, it is crucial to acknowledge that AHI, while important, does not encompass all aspects of sleep apnea.

In the future, a comprehensive metric that includes AHI, total HB, autonomic dysfunction (heart rate variability), and other parameters is expected to be developed more easily with the aid of artificial intelligence. This will allow for a more precise and fair assessment of both the efficacy of CPAP treatment and surgical outcomes.