Night-time heart rate variability identifies high-risk people among people with uncomplicated type 2 diabetes mellitus

Rakin Hadad,Bjørn Strøjer Larsen,Philip Weber,Dorte Stavnem,Ole P Kristiansen,Olav W Nielsen,Steen B Haugaard,Ahmad Sajadieh

DOI: https://doi.org/10.1111/dme.14559

Abstract:Background: Low heart rate variability (HRV) reflects cardiac autonomic neuropathy, which is associated with increased cardiovascular mortality in people with type 2 diabetes mellitus (T2DM). Measuring HRV is challenged by environmental noise, mental stress and physical activity during daytime. Night-time HRV during sleep may be a more valid tool to measure cardiac autonomic neuropathy and therefore may improve prediction of cardiovascular (CV) events in low-risk people with T2DM. Methods: Copenhagen Holter Study included 678 community-dwelling participants aged 55-75 years who were free of previous CV disease. Day and night-time HRV were available for 653 participants. The population included 133 people with well-controlled T2DM and newly recognized T2DM (mean HbA1c 55 mmol/mol [7.2%]). HRV is defined as standard deviation for the mean value of normal-to-normal complexes (SDNN). Night-time HRV measurements were pre-defined from 2:00 to 2:15 AM. Cardiovascular events were defined as CV death, myocardial infarction, stroke or coronary revascularization. Results: Median follow-up time was 14.4 years. During this period, 245 death and 149 CV events (CV death 36, myocardial infarction 42, revascularisation procedures 46, stroke 70) occurred in total. Among people with T2DM, 41 CV events were observed (CV death 13, myocardial infarction 13, revascularisation procedures 17, stroke 18). Night-time SDNN was inversely associated with CV events in people with T2DM, (hazard ratio [HR]: 0.74 95% confidence interval [CI]:0.61-0.89) for each 10-millisecond increment in SDNN after adjustment for the conventional risk factors sex, age, LDL cholesterol, smoking, systolic blood pressure and by also including glucose CRP and NT-proBNP in adjustment. Twenty-four-hour HRV was not associated with CV events, but associated with all-cause mortality in people with T2DM. Conventional risk factors had a receiver operating characteristic (ROC) value of 0.704 (95% CI 0.602-0.806) to predict CV events in people with T2DM. The prediction of CV events by conventional risk factors was improved in people with T2DM by the addition of night-time SDNN; ROC 0.765 (95% CI 0.669-0.862), p = 0.037, but ROC was not improved by addition of CRP and NT-proBNP in the model. In people with T2DM and night-time SDNN ≤30 ms, the 10-year risk of CV death and CV event rate was 12% and 45%, respectively, which re-allocated them to a 'very high-risk' group according to current guidelines. Conclusion: Reduced night-time HRV predicts increased risk of CV events in people with well-controlled T2DM, thus night-time HRV may add to traditional risk factors in predicting CV events in people with T2DM.