High Cardiometabolic Index Strongly Predicts Obstructive Sleep Apnea Risk
A large NHANES study finds that the Cardiometabolic Index — combining visceral fat and lipid markers — independently predicts OSA symptoms with 75% higher odds.
Summary
Researchers analyzed data from 8,460 U.S. adults in the NHANES database to examine whether the Cardiometabolic Index (CMI) — a composite measure of visceral fat and lipid metabolism calculated from triglycerides, HDL cholesterol, waist circumference, and height — is associated with obstructive sleep apnea (OSA). After adjusting for age, sex, race, lifestyle factors, and comorbidities, higher CMI was significantly linked to greater odds of self-reported OSA symptoms. The association held across all demographic subgroups. The CMI's ROC curve yielded an AUC of 0.605, suggesting modest but meaningful predictive value. The findings position CMI as a simple, low-cost screening tool that clinicians could use to flag patients at elevated OSA risk before expensive polysomnography.
Detailed Summary
Obstructive sleep apnea affects an estimated 17–34% of the general population and is strongly linked to hypertension, type 2 diabetes, cardiovascular disease, cognitive decline, and reduced quality of life. Despite its prevalence and consequences, OSA remains widely underdiagnosed because the gold-standard diagnostic test — polysomnography — is expensive, time-consuming, and resource-intensive. This study investigated whether the Cardiometabolic Index (CMI), a composite marker of visceral adiposity and lipid dysfunction, could serve as a practical screening tool for identifying individuals at elevated OSA risk using routine clinical measurements.
The CMI is calculated as: (Triglycerides / HDL-C) × (Waist Circumference / Height). It was first introduced in 2015 and has since been validated as a predictor of type 2 diabetes, insulin resistance, cardiovascular disease, non-alcoholic fatty liver disease, and stroke. Because OSA is closely intertwined with metabolic dysfunction — including dyslipidemia and central obesity — the authors hypothesized that CMI would also correlate with OSA symptom burden in a large, nationally representative population.
The study drew on NHANES data from four survey cycles (2005–2008 and 2015–2018), ultimately enrolling 8,460 adults aged 20 and older after excluding participants with missing OSA questionnaire data, incomplete CMI variables, or missing sampling weights. OSA was defined by self-report: participants were classified as OSA-positive if they endorsed at least one of three symptom criteria — snoring ≥3 nights/week, snorting/gasping/stopping breathing ≥3 nights/week, or excessive daytime sleepiness ≥16 times/month despite adequate sleep. The overall weighted OSA prevalence in the sample was 48.51%. CMI was analyzed both as a continuous variable and in quartiles (Q1: <0.28; Q2: 0.28–0.51; Q3: 0.51–0.91; Q4: >0.91).
In the fully adjusted logistic regression model (Model 3, controlling for age, sex, race, education, marital status, smoking, alcohol use, diabetes, hypertension, LDL-C, and poverty-to-income ratio), each unit increase in CMI was associated with 75% higher odds of OSA (OR = 1.75, 95% CI: 1.46–2.10). Quartile analysis showed a clear dose-response gradient: compared to Q1, participants in Q4 had significantly elevated OSA odds. Restricted cubic spline analysis confirmed a predominantly linear positive relationship between CMI and OSA probability, with no significant nonlinearity detected. Subgroup analyses and interaction tests across age, sex, BMI, and race found no significant effect modification, indicating the CMI–OSA association is robust across diverse demographic groups.
The receiver operating characteristic (ROC) curve analysis yielded an AUC of 0.605 for CMI as a predictor of OSA, indicating modest discriminatory power. While this AUC is not sufficient for standalone diagnostic use, it suggests CMI adds meaningful predictive information beyond chance and could be incorporated into multi-variable screening algorithms. Baseline data confirmed that participants in the highest CMI quartile were more likely to be male, obese, hypertensive, diabetic, and have lower educational attainment — all established OSA risk factors — yet the CMI–OSA association persisted after adjusting for these variables, suggesting CMI captures independent metabolic risk. The authors propose that CMI's integration of both central adiposity and lipid dysregulation makes it a uniquely informative and easily calculated clinical marker for OSA risk stratification.
Key Findings
- In the fully adjusted model, each unit increase in CMI was associated with 75% higher odds of OSA (OR = 1.75, 95% CI: 1.46–2.10, p<0.001)
- OSA prevalence in the 8,460-participant NHANES sample was 48.51%, with higher CMI quartiles showing progressively greater OSA rates
- Participants in the highest CMI quartile (Q4, CMI >0.91) were more likely to be male (61.3% vs 37.4% in Q1), obese, hypertensive, and diabetic
- ROC curve analysis yielded an AUC of 0.605 for CMI as a predictor of self-reported OSA symptoms
- Restricted cubic spline analysis confirmed a predominantly linear positive dose-response relationship between CMI and OSA probability
- Subgroup interaction tests found no significant effect modification by age, sex, BMI, or race, confirming the association's robustness across demographic groups
- Multiple imputation sensitivity analyses reproduced the primary findings, supporting the validity of the observed CMI–OSA association
Methodology
This cross-sectional study used NHANES data from four cycles (2005–2008 and 2015–2018), enrolling 8,460 adults ≥20 years after exclusions for missing data. OSA was defined by self-reported questionnaire responses across three symptom domains (snoring, nocturnal breathing events, excessive daytime sleepiness). Three weighted logistic regression models of increasing covariate adjustment were used, alongside restricted cubic splines (4 knots) for nonlinearity testing, subgroup interaction analyses, and ROC curve evaluation. NHANES complex survey design was accounted for using appropriate sample weights, clustering, and stratification.
Study Limitations
OSA was defined by self-reported symptoms rather than polysomnography, which may introduce misclassification bias and limits diagnostic precision. The cross-sectional design precludes causal inference — it is unclear whether metabolic dysfunction drives OSA or vice versa. The authors did not report conflicts of interest, and the study's reliance on fasting subsample weights reduced the available sample from the full NHANES cohort.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.