Should We Treat Mild Obstructive Sleep Apnea?

Aaron B. Holley, MD


August 23, 2016

Obstructive Sleep Apnea

In 2006, the American Academy of Sleep Medicine (AASM) published a position statement on the treatment of obstructive sleep apnea (OSA).[1] They recommended treating all moderate to severe OSA. Treatment of mild OSA was considered optional because data on treatment outcomes were inconsistent. Clinicians and researchers have been debating the value of treating mild OSA ever since,[2,3] and the AASM has not updated its guidelines. In May 2016, the American Thoracic Society (ATS) published a research statement[4] on the impact of mild OSA in adults. This article reviews the definition and prevalence of mild OSA and assesses the importance of the recent ATS statement.

OSA is characterized by changes in airflow that cause repetitive disruptions to sleep. The apnea-hypopnea index (AHI) quantifies the number of respiratory sleep disruptions per hour. The AHI is associated with adverse health outcomes in a linear fashion. As the AHI increases, the risk for physiologic abnormalities also rises.[5]

The AASM recommends using thresholds to define OSA severity.[6] An AHI ≥ 5 episodes/hour is considered abnormal, and AHI ≥ 5, 15, and 30 episodes/hour correspond to mild, moderate, and severe disease. The thresholds were defined by expert opinion without prospective data to support association with outcomes. Thresholds are often used as inclusion criteria in treatment trials because they are helpful for comparing outcomes across studies.

How Much OSA Is 'Mild'?

Most authors cite prevalence rates for OSA syndrome (OSAS—defined as an AHI ≥ 5 episodes/hour and daytime sleepiness) of 4% for men and 2% for women. These data come from the Wisconsin Sleep Cohort, a large, longitudinal study of cardiopulmonary sleep disorders in the general population.[7] Study participants were middle-aged (30-60 years) men and women employed in the state of Wisconsin. The study was slightly oversampled to include persons who snore. An AHI ≥ 5 episodes/hour was present in 24% of men and 9% of women. Most (15.6% of men and 7.6% of women) of the OSA among the entire cohort was mild.

There are a few important caveats to these data. Patients in the Wisconsin Sleep Cohort were studied with home sleep tests using only a thermistor (and not a pressure transducer) to detect changes in airflow. Current guidelines recommend using a thermistor and a pressure transducer to detect respiratory events.[8] A thermistor alone will miss 25%-50% of respiratory events detected by pressure transducers.[9] It's difficult to know exactly how much the prevalence of AHI ≥ 5 would have changed had a pressure transducer been used. The AHI certainly would have increased.

The definition of an episode of hypopnea has also changed over time. The Wisconsin Sleep Cohort investigators used a level of 4% oxygen desaturation as the threshold for defining an episode of hypopnea. The latest version of the AASM guidelines state only that an arousal is required to define hypopnea.[8] Several authors have modeled the changes in AHI based on the criteria for an episode of hypopnea.[10,11] In short, the AHI is approximately three times higher when arousal is used in lieu of a 4% oxygen desaturation. The prevalence of OSA is considerably higher if AASM 2012 scoring guidelines are used.

Furthermore, the International Classification of Sleep Disorders-3 (ICSD-3)[12] has changed the definition for OSA. They no longer use the term "OSAS." The ICSD-3 states that a patient has OSA if the AHI rate is mild (≥ 5 but < 15 episodes/hour) and the patient has sleep-related symptoms—fatigue, snoring, insomnia, breathing pauses, or non-refreshing sleep. Furthermore—and this is new—OSA is also diagnosed in the patient with both a mild AHI rate and an associated medical (to include hypertension, atrial fibrillation, congestive heart failure, stroke, diabetes) or psychiatric (mood disorders, neurocognitive deficits) disorder. The terminology has changed slightly, but essentially they are saying that mild OSA can be diagnosed and treated if any of these conditions are present.

Does Mild OSA Matter?

The bottom line is that OSA is very common and most OSA is mild. Changes in technology, scoring criteria, and the ICSD-3 have all increased the number of patients who meet diagnostic criteria. It is therefore important to know whether mild OSA is associated with adverse outcomes, and if so, whether treatment improves said outcomes. If treating mild OSA improves health outcomes, we should aggressively diagnose it and prescribe therapy. If there is no effect, we should not risk wasting our limited health resources. The ATS position statement aimed to quantify this with sleep-associated endpoints and determine treatment response.

Neurocognitive associations were assessed using five specific measures: sleepiness, motor-vehicle accidents, quality of life (QOL), neurocognitive function and mood. For subjective sleepiness, the authors specified that a mean increase of 0.5 points on the Epworth Sleepiness Scale was present in those with mild disease compared with those who had no OSA. The limited data on motor-vehicle accidents and QOL were conflicting, and no large population-based studies assessed neurocognitive function. One large population-based study found a relationship between mild OSA and mood, but this was not confirmed in three smaller studies.[4]

CV relationships were also assessed using five specific measures: hypertension, CV events, strokes, CV and all-cause mortality, and arrhythmias. No definitive association with hypertension, CV mortality, or all-cause mortality could be identified. The relationship to CV events was unclear and data were too limited to assess mild OSA and risk for arrhythmias. Patients with preexisting coronary artery disease and mild OSA may have an increase in stroke risk.[4]

Is Treating Mild OSA Beneficial?

The report found a modest improvement in subjective (but not objective) measures of sleepiness and variable effects on different QOL outcomes. Data on neurocognitive effects were inconsistent, and data on motor-vehicle accidents and mood outcomes were too limited to provide an assessment. No studies on stroke, CV events, atrial fibrillation, or other arrhythmias were available. Studies of hypertension were inconsistent, and one nonrandomized study did not provide evidence that treating mild OSA influences all-cause mortality.[4]

Although there does appear to be a little "noise" suggesting isolated improvement in subjective outcomes, it's hard to argue that placing the15.6% (or higher) of the male population with mild OSA on continuous positive airway pressure (CPAP) is cost-effective. The benefits are quite small at best and nonexistent at worst. It's difficult to understand why the ICSD-3 expanded criteria for diagnosing mild OSA when, according to the ATS statement, there is little evidence of an association with CV disease, neurocognitive symptoms, or mood disorders, much less evidence for treatment effect.

The CPAP and OSA proponents say that efficacy is all a matter of adherence, and that CPAP should only be used for the subgroup of mild-OSA patients who have OSAS. They urge sleep physicians to diagnose OSAS correctly and improve adherence. Unfortunately, poor adherence is a reflection of reality and may be worse with milder disease.[13,14] As for OSAS, the ATS statement casts serious doubt as to whether mild OSA is associated with the symptoms and diseases used to define it in the ICSD-3. In short, it seems that the research community, much less the physician seeing the patient in the office, doesn't know what symptoms and diagnoses are associated with mild OSA. How is it possible, then, to target only OSAS?

In summary, treating mild OSA/OSAS means placing a significant proportion of the population on therapy. At best, current data suggest that treatment provides a small benefit in subjective outcomes like QOL or sleepiness. There likely is a subset of patients with mild disease who will feel better with therapy. It seems poor policy to recommend treating 15.6% of the middle-aged male population with the hope that an undetermined, much smaller number will incur a slight improvement.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.