Combined insomnia, OSA phenotype predicts incident diabetes in men

Key takeaways:

• The ability to predict incident diabetes in men depended on the index used to define obstructive sleep apnea in combination with insomnia.
• Future studies are needed to see if this holds true in women.

In older men, the sleep breathing impairment index definition of combined insomnia-obstructive sleep apnea predicted incident diabetes, according to results published in Annals of the American Thoracic Society.

This result was not found with use of the apnea-hypopnea index (AHI) definition of combined insomnia and obstructive sleep apnea (COMISA), according to researchers.

“These findings suggest that this phenotype may be used for risk stratification and may inform clinical interventions for diabetes prevention such as targeted treatment of OSA (eg, with CPAP) and insomnia (with behavioral therapies),” Junwei Guo, of the department of respiratory and critical care medicine at Peking Union Medical College Hospital, and colleagues wrote.

Using data from the Osteoporotic Fractures in Men study, Guo and colleagues analyzed 2,365 men (mean age, 76 years; 91.5% white; median BMI, 26.49 kg/m²) without diabetes at baseline to assess the ability of the comorbid insomnia and sleep breathing impairment index (SBII) for predicting incident diabetes in COMISA. Researchers also evaluated the predictive performance of this index against that of COMISA-AHI using Cox proportional models.

Respiratory event duration and frequency, as well as the accompanying baseline desaturation area, made up SBII, according to researchers. Patients above the 50th percentile of this measure had sleep breathing impairment/OSA.

Most men in this study population had either mild (38.1%) or moderate (38.2%) OSA, followed by severe OSA (18.5%) and no OSA (5.2%).

The average follow-up period was 8 years, and at this checkpoint, 181 men (mean age, 75 years; 85.64% white; median BMI, 27.74 kg/m²) now had diabetes, whereas the remaining 2,184 men (mean age, 76 years; 91.99% white; median BMI, 26.37 kg/m²) still did not have diabetes.

Of the total cohort, 231 men (9.8%) had COMISA based on the AHI definition, and fewer men had COMISA based on the SBII definition (n = 212; 9%). Within these populations, diabetes was found in a higher proportion of those with COMISA-SBII vs. COMISA-AHI (n = 28; 13.2% vs. n = 23; 10%).

The risk for incident diabetes was heightened among those with COMISA-SBII vs. those without sleep disorders (adjusted HR = 1.82; 95% CI, 1.15-2.89) in a regression model that considered several covariates (demographics, comorbidities and behavioral risk factors).

Researchers continued to observe this significant relationship in an analysis adjusted for total sleep time, arousal index, hypnotic drugs, hypoxemia metrics and comorbid depression, as well as the competing risk model.

In contrast, COMISA-AHI was not significantly linked to incident diabetes in all three of the above models/analyses.

Further, researchers found that the predictive value for incident type 2 diabetes was significantly better when COMISA-SBII vs. COMISA-AHI was included in a crude model with established risk factors.

Exploratory analysis revealed that there was no significant relationship between incident diabetes and AHI, SBII or insomnia by themselves in the fully adjusted models and competing risk models.

“Our results identify a novel approach for defining a COMISA phenotype that more directly incorporates information on event duration and hypoxemia with insomnia symptoms,” Guo and colleagues wrote. “This measure predicted incident diabetes in older men, an outcome not previously consistently identified as a COMISA outcome.”

Future studies on this topic should include women and a diverse population since these were limitations of the current study.