Sleep apnea severity impacts all-cause mortality rates, odds in veterans

The findings of this study are surprising in that one would expect the survival curve to be proportional to the disease burden, meaning that the higher the severity of OSA, the higher the number of deaths expected. This was not the case here. However, there are several factors that need to be considered.

The AHI collected and the severity of disease burden were not clearly delineated. As the authors pointed out, they did not collect data about how much the subjects dropped their oxygen levels. Budhiraja and colleagues cited this as more predictive of survival, finding that if the oxygen levels are lower, there are more deaths. Specifically, oxygen levels less than 85% have been cited to correlate with higher risks for comorbid disease and death per a study by Khouzani and colleagues.

It was also not clear what the composition of the “apnea” vs. “hypopnea” was as common for home sleep apnea tests. In a review of the AHI, Malhotra and colleagues found an equal AHI of 15 per hour can have mostly apneas (complete cessation of breathing for 10 seconds or more) or mostly hypopneas (decrease in airflow by 30% to 50% for 10 seconds or more) with either a drop in oxygen of 3% or 4% or “arousal” with an in-lab study. The assumption is that the more apneas than hypopneas subjects exhibited, the more likely the subject would have daytime symptoms such as sleepiness or quality of life issues or comorbid conditions.

Other studies divide up severity differently: AHI less than 5 “normal,” AHI greater than 5 but less than 15 “mild,” AHI greater than 15 but less than 30 “moderate” and AHI greater than 30 “severe.” Despite the difference of classification, those with higher AHIs are generally expected to have more comorbid conditions.

Clinically, the results are still surprising in the “no” OSA group. This raises several other questions.

The group of veterans in the “no” OSA group aged younger than 40 years trended higher toward psychiatric disease. The nature of the illness obviously was beyond the scope of the study. Nonetheless, veterans that suffer from mood disorders and/or post-traumatic stress have been reported to have a more pronounced hyperarousal response. Whether this can pose an additional stress on the cardiovascular system is not clear. Furthermore, the morality rate did not report any from suicide, which is also prevalent amongst veterans.

Home sleep apnea tests (HSAT), and the proportion of these included, as compared with overnight polysomnography was not reported. False negative rates have been reported at variable rates from 16.3% in a study by Lipatov and colleagues, to higher amongst those with atrial fibrillation and in minorities per a study by Duggal and colleagues. Did the home sleep apnea test provide adequate assessment for the “no” OSA group, or missed the diagnosis?

Comorbid conditions such as COPD, arrhythmia, smoking history, neurological status such as prior stroke or respiratory impairment, cancer risks, alcohol and other substance(s) use were not reported in any of the groups. Medications were not included. It is difficult to make comparison without knowing if more subjects in the “no OSA” group had more or fewer underlying medical conditions contributing to higher mortality rates.

Central apneas are not accurately measured by HSATs. Those with neurological conditions or congestive heart failure at baseline again may have an inaccurate assessment based on HSAT and be misclassified since oxygen levels were not included in the “no OSA” group.

Compliance data of CPAP use, not just how many were used, can have implications in risk reduction. The report of higher CPAP usage rates in the “severe OSA” category and higher mortality again did not account for underlying medical conditions such as comorbid diabetes.

In a review, Agarwal and colleagues reported that minorities and those socioeconomically disadvantaged despite being in the VA system tend to have higher disease burden, more likelihood of false negative HSAT, less ability for follow-up visits and access to treatment. Whether any of these factors affect the “no OSA” group for retesting or the severe group for adequate treatment and follow-up is unknown.

Thinking about how doctors can use these findings to improve care, the theory that subjects gradually adapted to hypoxemia from OSA in the “mild-moderate OSA” group is an interesting theory. Perhaps this group had more access to care rather than from an adaptation response. Women also present differently with OSA symptoms and severity. As with most research, there was a male predominance in this study as well.

For the next step in this research, beside improved technology to help define respiratory abnormalities, the importance of the arousal response, as counted in the respiratory disturbance index instead of the typical AHI, may help separate out groups at risk for sleep fragmentation as compared with those that have “adapted” to repeated hypoxia without arousals and what that does to the cardiovascular, endocrine functions and resultant daytime performance. Better detection in women, particularly following menopause, may help increase our understanding of the best strategies to decrease risks for morbidity and mortality in the female population. Current technology tends to underestimate severity of OSA in women.

Better detection of arrhythmias is currently in the works. To be able to evaluate incidents of arrhythmia and any respiratory abnormalities before the conditions worsen can reduce those from future disability or impairment.

The question remains whether correcting hypoxia saves brain function. A study by Maestri and colleagues pointed out that treatment of OSA amongst those with dementia vs. those not treated may delay symptom progression. We do not know yet whether this is the case. More studies will need to be done.

This paper is important in the large size of each category included, and as the authors pointed out, OSA also has night-to-night variations, depending on how much sleep was obtained, body positions of the subjects tested, and the “category” to which they were placed.

References:

Agarwal S, et al. Curr Sleep Med Rep. 2024;doi:10.1007/s40675-024-00308-6.

Budhiraja R, et al. Southwest J Pulm Crit Care. 2021;doi:10.13175/swjpcc025-21.

Duggal A, et al. Sleep. 2024;doi:10.1093/sleep/zsae067.0559.

Khouzani MM, et al. CHEST. 2024;doi:10.1016/j.chpulm.2024.100087.

Lipatov K, et al. J Clin Sleep Med. 2018;doi:10.5664/jcsm.7486.

Malhotra A, et al. Sleep. 2021;doi:10.1093/sleep/zsab030.

Maestri M, et al. J Clin Sleep Med. 2020;doi:10.5664/jcsm.8536.