Is the oxygen desaturation index a better predictor of memory problems than the apnea-hypopnea index in sleep apnea?

The memory study tested oxygen measures — but not the ODI

The closest evidence comes from a study of 207 adults with obstructive sleep apnea who each spent one night in a sleep laboratory and completed two neuropsychological tests, including a verbal learning and memory test [1]. The researchers checked whether memory scores tracked with the standard AHI, with newer measures that capture how deep and how long each oxygen dip lasts, and with simpler oxygen summaries such as average oxygen saturation and the share of the night spent below 90% [1].

In the unadjusted analyses, several measures looked promising: immediate recall was associated with the arousal index, time below 90% oxygen, average oxygen saturation, and one of the newer hypoxic-load measures, while total recall was associated with all the apnea parameters tested [1]. But once age, sex, and education were added to the models, the picture narrowed sharply. Average oxygen saturation during sleep remained a significant predictor of both total recall and immediate recall, explaining about 21% and 11% of the variation in those scores respectively, while neither the traditional AHI nor the newer hypoxic-load measures predicted memory after adjustment [1].

Crucially, the oxygen desaturation index — the count of oxygen dips per hour — was not among the measures this study evaluated [1]. That means the study can tell us the AHI is a weak memory predictor, and that average oxygen saturation may be a stronger one, but it cannot tell us where the ODI stands.

Why oxygen measures might matter more than breathing-event counts

The AHI counts how often breathing pauses or shallow breaths occur per hour of sleep. The ODI counts how often oxygen saturation drops per hour. The two are related but not identical: a breathing event may or may not produce a measurable desaturation, and desaturations can occur for other reasons. Studies comparing them for diagnosis find they move together reasonably well — one study of 67 patients found a correlation of 0.73 between home oximetry ODI and laboratory AHI [2], and another found good agreement (87% concordance) in classifying severe sleep apnea [5].

That correlation is why the ODI is attractive as a cheaper, home-friendly screening tool [2][5]. But correlation for diagnosis is a different question from prediction of cognitive harm. The memory study's finding that average oxygen saturation — a steady level, not a count of dips — was the surviving predictor suggests that the brain's memory circuits may be sensitive to overall oxygen exposure rather than to the number of discrete events [1]. If that interpretation holds, an index built on counting desaturation events (the ODI) might behave more like the AHI than like average saturation, but this is a hypothesis the memory study did not test.

The ODI studies are about diagnosis, not memory

The remaining studies in this set examine how well the ODI tracks the AHI for diagnosing or grading sleep apnea, not how well either predicts memory. One retrospective review of 65 patients found only a weak positive association between ODI and AHI (correlation 0.33), which disappeared after adjusting for age, sex, body mass index, and ethnicity [3]. A larger retrospective study of 200 patients found significant AHI–ODI correlations in all severity groups, strongest in severe apnea, and concluded the ODI carries similar diagnostic weight to the AHI [4]. A screening study found that an ODI above 20 detected severe sleep apnea with 96.6% sensitivity but 69.6% specificity [5].

These results describe how the two indices relate to each other, and they disagree somewhat on how tightly — from weak [3] to strong [4][5] — depending on the population and setting. None of them measured memory, so none can say whether the ODI is a better memory predictor than the AHI. The memory study itself notes that the AHI is poorly associated with the cognitive decline often seen in sleep apnea patients, which is the gap the ODI would need to fill [1].

What would actually answer the question

To know whether the ODI beats the AHI for memory, a study would need to measure both indices in the same patients alongside memory tests, then compare how much of the memory variation each explains after accounting for age, sex, and education. The existing memory study provides the template — it did exactly this kind of head-to-head comparison among several sleep measures — but left the ODI out [1].

For now, the practical takeaway is modest. The AHI has not performed well as a memory predictor in the one study that tested it directly [1], and average oxygen saturation outperformed it there. The ODI remains a reasonable stand-in for the AHI when diagnosing sleep apnea [2][4][5], but there is no evidence yet that it is a better guide to who will develop memory problems.

Sources used in this answer

Related research questions