What the Apnea Hypopnea Index Does and Does Not Tell You

The number, and its limits

The apnea hypopnea index (AHI) counts how often per hour of sleep your breathing stops or partly collapses, and it is the number that sorts obstructive sleep apnea into mild, moderate, or severe. It is useful, communicable, and reproducible enough to guide many real decisions. But it is a frequency count that ignores how long each event lasts, how far your oxygen falls, and whether you wake up exhausted or feel fine, and its very definition shifts with the scoring rule a given lab applies. Treated as a single verdict on how sick you are or how much your heart is at risk, the AHI promises more precision than the measurement can support.

How the index is built

An apnea is a near-total stop in airflow, at least a 90 percent reduction lasting 10 seconds or longer. A hypopnea is a partial event, roughly a 30 percent drop in airflow for at least 10 seconds, paired with a physiological consequence. You add the apneas and hypopneas across the night and divide by hours of sleep. By long-standing clinical convention, an AHI of 5 to 15 is mild, 15 to 30 is moderate, and above 30 is severe. Those cut points are useful administrative lines, not biological thresholds where risk suddenly switches on.

The hypopnea definition is a moving target

The "physiological consequence" is where the number gets slippery. The American Academy of Sleep Medicine's recommended rule scores a hypopnea when airflow drops for at least 10 seconds and is followed by either a 3 percent oxygen desaturation or a cortical arousal. Its accepted alternative keeps the airflow and duration criteria but requires a 4 percent desaturation and gives no credit for arousals, and that stricter version is the one Medicare's coverage criteria have long relied on. Run the same night's recording through both rules and you get two different indices. The 3 percent rule reliably produces the higher number, moving many people from normal to mild, or from mild to moderate, on the strength of the rulebook alone. The shift does not fall evenly: because events that end in an arousal rather than a deep oxygen drop are more common in women, the choice of rule tends to change women's classifications more often. A diagnosis that can flip based on which manual the scorer opened is not a fixed property of the patient.

One night is a thin sample

Even with the scoring settled, a single night is a small sample of a variable process. Breathing during sleep changes with body position, alcohol, nasal congestion, sleep stage, and how much REM you happen to get. A large study by Lechat and colleagues used an in-home sensor to record more than 67,000 people across roughly 170 nights each, then asked how often a single night would have misled them. About one in five people with sleep apnea would have been told they did not have it, and the mild and moderate labels were correct on only about half of individual nights, while severe disease held up better. Diagnostic accuracy kept improving until roughly two weeks of recording. A single laboratory night, in other words, can place you in the wrong severity bin through nothing but chance. This article is educational and is not medical advice.

Frequency is not severity

The AHI also flattens differences that matter. It scores a 10-second event and a 90-second event as one apiece, a shallow dip to 94 percent oxygen the same as a plunge into the high 70s, and an event in light sleep the same as one in REM. Two people with an AHI of 20 can have very different nights. This is why researchers have looked past event frequency toward the physiological cost of the events. Hypoxic burden, which measures the area under the oxygen desaturation curve and so captures depth and duration together, has performed strikingly better as a risk marker. In the Osteoporotic Fractures in Men Study and the Sleep Heart Health Study, hypoxic burden predicted cardiovascular mortality in both cohorts, and all-cause mortality in the men's cohort, with the association holding even after adjusting for the AHI, while the AHI itself did not show the same relationship. The count told you an event happened; it did not tell you how much damage the event carried.

What it misses about how you feel

Nor does the index track symptoms closely. Some people with a high AHI report no sleepiness, while others with a modest AHI are exhausted, foggy, and hypertensive. Research on symptom subtypes finds that patients cluster into groups with different complaints and different cardiovascular trajectories that the AHI alone does not separate. So the number cannot tell you, on its own, whether treatment will make you feel better or protect your heart.

Using the number well

None of this makes the AHI worthless. It is a reasonable triage and screening tool, a starting point that becomes more informative alongside oxygen measures, symptoms, comorbidities, and, where variability is a concern, more than one night of data. The direction of the field is toward reading breathing during sleep as a physiological picture rather than a single tally. The number also cannot settle a population-level question on its own: the US Preventive Services Task Force concluded that current evidence is insufficient to screen asymptomatic adults for sleep apnea, an I statement that reflects how uncertain the path from a screening number to a better health outcome remains. That caution is about screening people without symptoms, not a reason for someone with loud snoring, witnessed pauses, or daytime sleepiness to skip an evaluation. The AHI is a starting question, not the final answer.