Mortality‐risk‐based apnea–hypopnea index thresholds for diagnostics of obstructive sleep apnea
Henri Korkalainen, Juha Töyräs, Sami Nikkonen, Timo Leppänen
Abstract
Open-access reader
Henri Korkalainen, Juha Töyräs, Sami Nikkonen, Timo Leppänen
Abstract
Open-access reader
The severity of obstructive sleep apnea is clinically assessed mainly using the apnea-hypopnea index. Based on the apnea-hypopnea index, patients are classified into four severity groups: non-obstructive sleep apnea (apnea-hypopnea index < 5); mild (5 ≤ apnea-hypopnea index < 15); moderate (15 ≤ apnea-hypopnea index < 30); and severe obstructive sleep apnea (apnea-hypopnea index ≥ 30). However, these thresholds lack solid clinical and scientific evidence. We hypothesize that the current apnea-hypopnea index thresholds are not optimal despite their global use, and aim to assess this clinical shortcoming by optimizing the thresholds with respect to the risk of all-cause mortality. We analysed ambulatory polygraphic recordings of 1,783 patients with suspected obstructive sleep apnea (mean follow-up 18.3 years). We simulated 79,079 different threshold combinations in 100 randomized subgroups of the population and studied the relative risk of all-cause mortality corresponding to each combination and randomization. The optimal thresholds were chosen according to three criteria: (a) the hazard ratios increase linearly between severity groups towards more severe obstructive sleep apnea; (b) each group includes at least 15% of the study population; (c) group sizes decrease with increasing obstructive sleep apnea severity. The risk of all-cause mortality varied greatly across simulations; the threshold defining non-obstructive sleep apnea group having the largest effect on the hazard ratios. The apnea-hypopnea index threshold combination of 3-9-24 was optimal in most of the subgroups. In conclusion, the assessment of obstructive sleep apnea severity based on the current apnea-hypopnea index thresholds is not optimal. Our novel approach provides methods for optimizing apnea-hypopnea index-based severity classification, and the revised thresholds better differentiate patients into severity groups, ensuring that an increase in the severity corresponds to an increase in the risk of all-cause mortality.
OpenAlex reports 25 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
The severity of obstructive sleep apnea is clinically assessed mainly using the apnea-hypopnea index. Based on the apnea-hypopnea index, patients are classified into four severity groups: non-obstructive sleep apnea (apnea-hypopnea index < 5); mild (5 ≤ apnea-hypopnea index < 15); moderate (15 ≤ apnea-hypopnea index < 30); and severe obstructive sleep apnea (apnea-hypopnea index ≥ 30). However, these thresholds lack solid clinical and scientific evidence. We hypothesize that the current apnea-hypopnea index thresholds are not optimal despite their global use, and aim to assess this clinical shortcoming by optimizing the thresholds with respect to the risk of all-cause mortality. We analysed ambulatory polygraphic recordings of 1,783 patients with suspected obstructive sleep apnea (mean follow-up 18.3 years). We simulated 79,079 different threshold combinations in 100 randomized subgroups of the population and studied the relative risk of all-cause mortality corresponding to each combination and randomization. The optimal thresholds were chosen according to three criteria: (a) the hazard ratios increase linearly between severity groups towards more severe obstructive sleep apnea; (b) each group includes at least 15% of the study population; (c) group sizes decrease with increasing obstructive sleep apnea severity. The risk of all-cause mortality varied greatly across simulations; the threshold defining non-obstructive sleep apnea group having the largest effect on the hazard ratios. The apnea-hypopnea index threshold combination of 3-9-24 was optimal in most of the subgroups. In conclusion, the assessment of obstructive sleep apnea severity based on the current apnea-hypopnea index thresholds is not optimal. Our novel approach provides methods for optimizing apnea-hypopnea index-based severity classification, and the revised thresholds better differentiate patients into severity groups, ensuring that an increase in the severity corresponds to an increase in the risk of all-cause mortality.
Key concepts: Obstructive sleep apnea, Medicine, Apnea, Hypopnea, Apnea–hypopnea index, Sleep apnea, Population, Polysomnography