Modeling 24-Hour Blood Pressure Rhythms Using Mixed-Effects Cosinor Models

Single-component cosinor models are commonly used to assess circadian dysregulation in 24 h ambulatory blood pressure (BP) measurements, but they cannot capture more complex or asymmetric circadian patterns. This study evaluated 4 mixed-effects cosinor models by applying them to systolic and diastolic blood pressure (SBP and DBP) data collected from children with obstructive sleep apnea (OSA) and healthy controls (non-OSA) and compared the estimated circadian parameters between groups. The analysis included 24-h BP monitoring data from 219 age- and gender-matched children (117 controls, 52 with mild OSA, and 50 with moderate-to-severe OSA [MS-OSA]). Mixed-effects cosinor models with 1 to 4 components estimated various circadian parameters: acrophase, amplitude, and time arrived at peak velocity (TAPV). The 3-component mixed-effects cosinor model provided the best fit for SBP and DBP data. The estimated MESOR (midline estimating statistic of rhythm) was 106 mmHg for SBP and 64 mmHg for DBP in the control group; the MS-OSA group had a higher DBP MESOR (66 mmHg). Children with OSA had a dampened early afternoon BP peak and an increased late-evening BP peak. The timing differences for the first BP peak were more pronounced in the morning for SBP, and from morning to mid-day for DBP, particularly in MS-OSA. TAPV occurred in the morning in all models, with slight timing differences in SBP for mild OSA, and in DBP for MS-OSA, compared to controls. Although single-component cosinor models are traditionally used for 24-h BP rhythms, the multi-component mixed-effects models provided a better fit and captured disease-related differences.