The AHI: not all that it’s cracked up to be

Citation:

Won CHJ. The AHI: not all that it’s cracked up to be. J Clin Sleep Med. 2020;16(11):1987–1988.

The diagnosis of obstructive sleep apnea (OSA) currently hinges upon the apnea-hypopnea index (AHI); this is true in research, in policy, and in clinical practice. This assumes that the AHI is the best metric for predicting disease, and that targeting treatment to the AHI best achieves outcomes. However, the AHI only modestly predicts cardiovascular and noncardiovascular disease, sleepiness, and quality of life.¹ Moreover, emerging data suggest different metrics on polysomnography may provide as good if not better risk categorization for several outcomes, including mortality.² Hunasikatti’s³ letter concedes that it is unlikely that a single metric will perfectly define OSA risk and outcomes. In addition to other polysomnographic measurements, the patient’s age and baseline health also likely contribute to defining pathologic upper airway obstruction during sleep. To solely rely on the AHI, as a single and distilled metric, for defining disease and disease severity, risk-stratifying patients, and determining treatment, is unjustified.

An intrinsic problem with the AHI is the variability by which a hypopnea is defined. The “best” definition of a hypopnea (ie, partial airway obstruction) has yet to be defined, as it pertains to disease and symptomology relevance. In this regard, research studies have utilized a range of AHI definitions and cutoffs to explore disease correlations. It seems problematic then to define OSA by a metric that remains unstandardized. The arbitrariness of the hypopnea definition is also illustrated in clinical practice by the varying insurance criteria for treatment qualification. Most private insurers cover positive airway pressure treatment based on an AHI that captures hypopneas with 3% desaturations or arousals. Meanwhile, the Centers for Medicare and Medicaid Services rejects this hypopnea criteria in favor of 4% desaturations. As Ho et al⁴ demonstrated, depending on the definition of hypopnea, one can misclassify up to 30% of “normal” persons and up to 10% of patients with severe OSA.

One of the most important reasons for advocating for more accurate disease capture and phenotypes is to be able to appropriately intervene and improve health outcomes. There have been several randomized clinical trials with positive airway pressure therapy,⁵ none of which have demonstrated a cardiovascular benefit. At this point, we need to consider that either treating OSA is inconsequential to cardiac outcomes or that we are not accurately capturing at-risk individuals with OSA. It has been argued that these studies failed to show an effect with positive airway pressure therapy because they omitted the highest-risk phenotype of the sleepy patient with OSA. In this case, the AHI alone was not appropriate for identifying at-risk individuals who might benefit from positive airway pressure therapy.

The AHI is certainly an important piece of data that is informative and predictive to some degree, but limited as a sole disease-defining metric. As discussed in Won’s commentary,⁶ the overall AHI may miss other important disease phenotypes such as rapid eye movement–OSA or supine-OSA and ignore other risk-defining metrics such as hypoxemic burden or sympathetic activation. These are lost opportunities for refining treatment targets. Therefore, it is time to rise to the sophistication of current understanding of the disease; it is time to ditch the AHI.

DISCLOSURE STATEMENT

The author has seen and approved the manuscript. The author reports no conflicts of interest.