OSA: Why AHI alone isn't predicting your patients' heart risk

Obstructive sleep apnea is a chronic condition characterized by recurrent episodes of upper airway collapse during sleep, leading to intermittent hypoxia and sleep fragmentation. The prevailing diagnostic framework categorizes OSA severity based on AHI: mild (5-15 events/hour), moderate (15-30 events/hour), and severe (>30 events/hour). This classification dictates treatment pathways, primarily continuous positive airway pressure (CPAP) therapy, but also oral appliances, positional therapy, and surgical interventions. Despite its widespread use, the AHI often fails to correlate consistently with patient-reported symptoms, quality of life, or the long-term cardiovascular and metabolic risks associated with OSA. A patient with a moderate AHI might experience significant daytime sleepiness and hypertension, while another with a higher AHI could be relatively asymptomatic, highlighting the metric's inherent variability in reflecting individual disease burden.

The current reliance on AHI stems from its simplicity and ease of measurement via polysomnography (PSG), the gold standard diagnostic test. PSG records multiple physiological parameters during sleep, including electroencephalography (EEG), electrooculography (EOG), electromyography (EMG), electrocardiography (ECG), respiratory effort, airflow, and oxygen saturation. From these recordings, apneas (cessation of airflow for at least 10 seconds) and hypopneas (reduction in airflow by at least 30% for at least 10 seconds, accompanied by a 3% oxygen desaturation or an arousal) are identified and counted. The AHI is then calculated by dividing the total number of apneas and hypopneas by the total sleep time in hours. This standardized approach has facilitated research and clinical practice for decades, but it homogenizes a complex pathophysiology into a single number, overlooking critical aspects of disease expression.

The Limitations of a Single Number

The AHI's primary limitation lies in its inability to capture the qualitative aspects of respiratory events. It treats all apneas and hypopneas as equal, regardless of their duration, the depth of associated oxygen desaturation, or their impact on sleep architecture. A brief hypopnea with minimal desaturation is counted the same as a prolonged apnea leading to severe hypoxemia, yet their physiological consequences are vastly different. This lack of granularity means that two patients with identical AHI scores can have dramatically different clinical presentations and prognoses. The AHI also does not differentiate between central and obstructive events, which have distinct underlying mechanisms and treatment implications, although this distinction is typically made during PSG interpretation.

The impact of OSA on cardiovascular health, including hypertension, atrial fibrillation, coronary artery disease, and stroke, is well-established. But the AHI's predictive power for these outcomes is often weak. This disconnect has driven the search for alternative or complementary metrics that better reflect the physiological stress imposed by OSA. The intermittent hypoxia associated with OSA, characterized by repeated cycles of oxygen desaturation and reoxygenation, is a key driver of systemic inflammation, oxidative stress, and sympathetic nervous system activation, all contributing to cardiovascular morbidity. The AHI, by itself, provides only an indirect measure of this hypoxic burden.

Emerging Metrics and Their Promise

Several alternative metrics are gaining traction, aiming to provide a more comprehensive assessment of OSA severity and its clinical impact. One such metric is the oxygen desaturation index (ODI), which counts the number of times per hour that blood oxygen levels drop by a certain percentage (typically 3% or 4%) from baseline. The ODI directly quantifies the hypoxic burden, which is often considered a more direct mediator of cardiovascular damage than the mere frequency of respiratory events. Some studies suggest that ODI may be a better predictor of cardiovascular outcomes than AHI, particularly in patients with co-morbidities like heart failure or diabetes. But ODI still does not account for the depth or duration of desaturations, treating all drops beyond the threshold as equivalent.

Another area of focus is sleep fragmentation. OSA not only causes hypoxia but also repeatedly disrupts sleep continuity through respiratory effort-related arousals (RERAs) and micro-arousals. These sleep disturbances lead to chronic sleep deprivation, daytime sleepiness, and impaired cognitive function. Metrics like the arousal index (AI), which counts all arousals per hour of sleep, including those not associated with apneas or hypopneas, offer a more direct measure of sleep fragmentation. The AI can be particularly relevant for patients whose primary complaint is excessive daytime sleepiness, even if their AHI is only mildly elevated. But the scoring of arousals can be subjective, leading to inter-scorer variability.

More sophisticated metrics are also under investigation, including measures of cumulative hypoxic time (the total time spent below a certain oxygen saturation threshold), the severity of the lowest oxygen saturation (nadir SpO2), and the variability of heart rate during sleep. These parameters aim to quantify the physiological stress more precisely. For instance, the percentage of total sleep time spent with oxygen saturation below 90% (T90) provides a direct measure of sustained hypoxemia. The nadir SpO2, the lowest oxygen saturation recorded during sleep, can indicate the severity of individual hypoxic events. These metrics move beyond simple event counts to characterize the physiological insult more thoroughly. Clinicians can find comprehensive guidance on these and other diagnostic approaches in resources like the Oxford Handbook of Respiratory Medicine.

The Challenge of Integration

The proliferation of new metrics presents a challenge for clinical integration. While AHI is imperfect, it is universally understood and forms the basis of current diagnostic guidelines. Adopting new metrics would require re-educating clinicians, updating diagnostic criteria, and potentially re-evaluating treatment thresholds. Many of these advanced metrics require sophisticated analysis of PSG data, which may not be readily available in all sleep laboratories. The move towards home sleep apnea testing (HSAT), which typically measures fewer physiological parameters than PSG, also complicates the adoption of more complex metrics. HSAT often provides only AHI and ODI, limiting the ability to assess sleep architecture or differentiate central from obstructive events comprehensively. Still, the increasing use of wearable technology and AI in sleep diagnostics promises to make more granular data accessible, even outside of traditional lab settings.

The field is also exploring composite scores that combine multiple physiological parameters to create a more holistic picture of OSA severity. These scores might integrate AHI, ODI, AI, and measures of cardiovascular variability to provide a single, more predictive risk score. Such an approach could potentially overcome the limitations of any single metric by accounting for the complex nature of OSA pathophysiology. But developing and validating such composite scores requires extensive research and large patient cohorts to ensure their clinical utility and generalizability. The goal is to identify metrics that not only correlate with patient symptoms but also reliably predict long-term adverse outcomes, allowing for more personalized and effective treatment strategies.

The open-label design of many observational studies exploring these new metrics is an obvious caveat. The absence of randomized controlled trials specifically designed to compare the prognostic value of AHI versus other metrics means that definitive conclusions remain elusive. The trial was not powered to detect differences in specific cardiovascular outcomes based on these alternative metrics, and that gap matters for guideline development. Until robust, prospective data emerge, AHI will likely remain the primary diagnostic criterion, even as its shortcomings are increasingly acknowledged. The challenge lies in moving beyond simply identifying respiratory events to understanding their true physiological cost and guiding interventions that genuinely mitigate risk.