An AHRQ report raises an important point that an individual with obstructive sleep apnea’s cumulative exposure to diminished inhaled air could contribute to sequelae, much like total exposure to cigarette smoke contributes to an increased risk of malignancies. 

By Sree Roy

Sleep medicine loves a good index. The metrics that sleep specialists use most commonly to measure obstructive sleep apnea (OSA) severity—like the apnea-hypopnea index (AHI), oxygen desaturation index, and respiratory disturbance index—quantify the number of respiratory events per unit of time. 

Reading the Agency for Healthcare Research and Quality’s (AHRQ) report on long-term health outcomes in OSA opened my eyes to an alternative way of measuring sleep apnea, a paradigm that already predicts health outcomes for other inputs: cumulative exposure.1

An index “cannot distinguish between patients who have the same average intensity of events but for different durations,” says AHRQ report investigator Thomas Trikalinos, PhD, MD, director of the Center for Evidence Synthesis in Health at the School of Public Health at Brown University. “It may be that intensity is what matters irrespective of total exposure. But there are some indications that some measure that quantifies total exposure to reduced oxygenation is also of interest.”

Metrics of cumulative exposure would incorporate sleep duration and could include the total number of events in a sleep study, total duration of events by type, and the time spent with oxygen saturation below a specified threshold. 

“At this point, we do not know which are likely to be better metrics,” Trikalinos says. “There are a lot of permutations for proposing such metrics: eg, which events should be aggregated, and whether all types of events should weight the same towards the total index, or whether events should be weighted by the extent of hypoxia during the event. It is practical to explore many metrics at the same time.” 

Common Cumulative Metrics

There are many cumulative metrics used in healthcare. A few examples include:

  • radiation exposure,
  • pulmonary inhalation exposures (such as to asbestos),
  • pack-years in smoking, and
  • hemoglobin A1C (HbA1c).

“The measurement of HbA1c (glycosylated hemoglobin) cumulates (aggregates, integrates) the instantaneous exposure to glucose levels as they fluctuate over time,” Trikalinos says. “We do not, to our knowledge, have an analogous natural aggregator for oxygenation levels—so we must do the aggregation by measurement and computation.”

How Cumulative Sleep Apnea Metrics Would Help

Cumulative metrics could redefine severity for specific subpopulations and could even help certain populations get diagnosed in the first place.

For example, take two people with OSA: one with an AHI of 80 and one with an AHI of 25. Based on that index, the person with 80 events per hour might think their disease course is extremely severe, while the person with the 25 might think their “moderate” case isn’t so bad. 

“But if you think about it: If the AHI is 80, that person had a lot of events, but they may not necessarily have been long events. So, their desaturations may not even have been that bad,” says AHRQ report peer reviewer Nancy A. Collop, MD, director of the Emory Sleep Center in Atlanta. “But if the other person had events that lasted longer, say, 45 seconds, and had oxygen desaturations into the 70s or 60s, they might be at much higher risk” of poor long-term outcomes. 

Focusing on cumulative measures could also help women—whose sleep apnea is often overlooked—get diagnosed.

“It’s well documented that women have typically most of their respiratory events in REM sleep, and they have classically been considered mild sleep apnea or not diagnosed because AHI is an index,” says AHRQ report investigator Carolyn M. D’Ambrosio, MS, MD, FCCP, vice chief, fellowship training and mentorship in the section of pulmonary, critical care, and sleep medicine at the Yale University School of Medicine. “If we had, say, 5,060 apneas or hypopneas in a night of sleep as a measurement, you’d be more likely to get the diagnosis for them.”

What’s more, a cumulative measurement would better facilitate prospective studies. Most studies to date linking sleep apnea to long-term health outcomes have been retrospective.

Consider the Lifetime Impact of Sleep Apnea

In cigarette smokers, the risk of lung cancer is established using “pack years,” that is, the amount of tobacco smoked and the years exposed. 

I asked experts if sleep medicine should consider similar cumulative metrics over a lifetime, such as sleep apnea onset and offset, with “offset” representing when a person’s disease is below a specified threshold of respiratory events.

“The analogy in the [AHRQ] assessment is narrower: We propose that these metrics cumulate the intensity of respiratory events during the sleep study. So, they would aggregate over, say, a night,” Trikalinos says. 

“In longitudinal studies that assess the association of sleep apnea indices with cardiovascular events, it is possible to do something along the lines you described by counting essentially the length of time after sleep apnea diagnosis as the exposure period. If we had detailed nightly measurements of respiratory event intensity, we could use them exactly as you propose.”

Several experts commented that, unlike the smoking analogy, it is difficult to tease out the onset of sleep apnea. The symptoms tend to come on gradually. But “with the ambulatory monitoring people now have on themselves, like the Apple Watch, maybe we’ll get there,” Collop says.

Calculating Sleep Apnea Onset

I was pleasantly surprised to discover that a team of researchers, led by Michelle Olaithe, PhD, MClinPsych, an associate research fellow at the University of Western Australia and the clinical director and Health-Bright clinical psychologist at, recently developed an “age of OSA Onset” algorithm that takes sleep medicine in the direction of calculating lifetime exposure to sleep apnea.2

They used longitudinal data from the Wisconsin Sleep Cohort from participants who had two sleep studies and were not using CPAP and from Sleep Heart Health Study participants who had an initial sleep study showing no significant sleep apnea and a later sleep study showing moderate to severe OSA.2

As it turns out, regression analyses found there are only three variables needed to determine the age of OSA onset: sex, AHI at the final observed sleep study, and body mass index (BMI) at the final observed sleep study.2 

“In the development phase, we tested a range of variables shown to be associated with OSA, including apnea-hypopnea index, body mass index, weight gain, snoring, diabetes diagnosis, history of cardiovascular events, hypertension, daytime sleepiness, mood, and sex,” Olaithe says. They even tested excessive daytime sleepiness, even though its onset is notoriously hard to determine. “If the individual’s excessive daytime sleepiness started before the data set, we did not have that; however, we did have data from 1989 to the time of the analysis in 2017,” she says. 

I was surprised by how many variables were not predictive, for instance, weight change. “We found it surprising a lot of things fell out after the initial regressions,” Olaithe says. “However, in a way, change in BMI is accounted for in the regression weights. The regression equation weights for AHI, BMI, and sex are from the individuals’ initial and subsequent sleep studies. It is just when predicting that you enter in the current AHI, BMI, and sex of the individual.” 

All in all, the OSA-onset algorithm estimated years of exposure to OSA with an accuracy of between 6.6 and 7.8 years. In 2024, Olaithe plans to start applying for funding to test the algorithm in a real-world sleep clinic population.

Challenges to Adopting a Cumulative Sleep Apnea Metric

So why doesn’t sleep medicine use cumulative metrics?

Several experts noted a practical reason: Cumulative calculations aren’t built into all sleep scoring software. “It’s hard to change the dogma associated with these studies that we’ve done for years and years and years,” Collop says.

Other sleep specialists, like AHRQ report peer reviewer Charles W. Atwood Jr., MD, FCCP, FAASM, an associate professor of medicine and director of the Sleep Disorders Program at the VA Pittsburgh Healthcare System, says cumulative measures would correlate closely with the metrics already in use, making them a moot point.

“I don’t see a big advantage,” Atwood says. “For example, one of the measures used in sleep medicine is the CT90 [the cumulative amount of time a patient is below a specified oxygen saturation, typically under 90%]. But that correlates closely with AHI. CT90 is reported on pretty much all sleep studies already.”

Would You Prefer Cumulative Metrics or Indices?

Finally, I asked the experts: If they could determine the future for all of sleep medicine, would they select a cumulative metric or an index?

Atwood says, “Even in the case of women who have more REM-related events, I wouldn’t throw away the AHI. You do have to take into account the whole picture, such as their symptoms and all abnormalities on their sleep study. But I’m not sure we have anything better than AHI, even in that situation, right now.”

Any new metric would have to be studied extensively first, D’Ambrosio notes. “I would put my money down on a cumulative metric and say that’s where we should be going. But we’d need to look at it and ask: Does that predict the outcomes we care about?”

Outcome of interest does, of course, make this answer complicated. 

“Sleep apnea is associated with so many comorbidities,” Collop says. “Is your outcome of interest if they’ve had a stroke; are they sleepy; or what their blood pressure is?”

She ideally wants a metric, such as the currently popular AHI, that encompasses sleep disruption. 

“If you just use a cumulative score, you’re probably not going to capture people with frequent sleep disruptions. I don’t know that I would want to discount that piece,” Collop says. “We need to keep working toward something better than AHI but not get stuck just looking at one aspect of sleep apnea.”


1. Balk EM, Adam GP, Cao W, et al. Long-term health outcomes in obstructive sleep apnea: A systematic review of comparative studies evaluating positive airway pressure and validity of breathing measures as surrogate outcomes. Project ID: SLPT0919. (Prepared by the Brown Evidence-based Practice Center under Contract No. 290-2015-00002-I/Task Order No. 75Q80119F32017.) Agency for Healthcare Research and Quality. 2022 Dec 1. 

2. Olaithe M, Hagen EW, Barnet JH, et al. OSA-Onset: An algorithm for predicting the age of OSA onset. Sleep Med. 2023 Aug;108:100-4.

Illustration 30265960 © Sashazamarasha |