0582 Apnea-Hypopnea Index at the US / Mexico Border: Prevalence and Sociodemographic Correlates of Clinical Cutoffs – Data from the NoCHeS Study

Abstract Introduction Sleep disordered breathing is under-studied among Hispanics/Latinos, especially in communities where sociocultural and clinical risk factors may contribute to disease burden. Methods Study participants were 992 adults ages 25-60 years residing along the US-Mexico Border. Apnea-Hypopnea Index (AHI) was determined from home sleep tests (Embletta) and scored by certified polysomnographic technicians using the 3% criterion. Results were characterized as None (0.0-4.9), Minimal (5.0-9.9), Mild (10.0-14.9), Moderate (15.0-29.9), and Severe (30+). Potential correlates included language (English or Spanish), age, gender, education, length of time in the US, financial status, hypertension, diabetes, and Anglo and Mexican acculturation (ARSMA-II), Americanism, Hispanicism, and Biculturism (BIQ), and objective Body Mass Index (BMI). Penalized logistic regressions using the LASSO regression with tenfold cross-validation was used to determine the optimal regularization parameter and select predictors that contributed meaningfully to classification of each cutoff. Following model convergence, the non-zero coefficients identified were extracted for downstream modeling. This was followed by a multinomial logistic regression to determine factors that differentiate each category from None. Results Prevalence of AHI categories was None (61.5%), Minimal (21.4%), Mild (7.7%), Moderate (5.6%), and Severe (4.2%). The most parsimonious model differed across categories. For Minimal (λ=0.0148; CV≈1.24), non-zero predictors included age, gender, BMI, immigration, financial status, Mexican acculturation, hypertension, and diabetes. For Mild (λ=0.0209; CV≈1.04), non-zero predictors were gender, immigration, and hypertension. For Moderate (λ=0.0139; CV≈0.53), the coefficients were age, BMI, and language. For Severe (λ=0.0126; CV≈0.41), the non-zero predictors were age, BMI, and hypertension. A multinomial logistic model with only these variables demonstrated significant associations between male gender (RRR=2.28) and hypertension (RRR=1.73) for Minimal; age (RRR=1.04/yr), male gender (RRR=2.42), and BMI (RRR=1.09/pt) for Mild; age (RRR=1.06/yr), male gender (RRR=2.47), BMI (RRR=1.11/pt), and hypertension (RRR=2.50) for Moderate; and age (RRR=1.07/yr), male gender (RRR=4.92), BMI (RRR=1.18/pt), and >20 years in the US (RRR=0.21) with being in the Severe range. Conclusion Sleep apnea was highly prevalent in this cohort at the US-Mexico border. Using a machine-learning strategy, the strongest correlates were age, gender, BMI, and hypertension. These may serve as effective risk factors for identifying individuals at high-risk in the Hispanic/Latinx border communities Support (if any) R01MD011600, R01MH135978