Effect of Glucose Variability on the Mortality of the Very Old People During the First Year of the COVID-19 Pandemic

Abstract

Abstract Background To our knowledge, only one study has examined the association between glucose variability (GV) and mortality in the elderly population with diabetes. GV was assessed by HbA1c, and a J-shaped curve was observed in the relationship between HbA1c thresholds and mortality. No study of GV was conducted during the COVID-19 pandemic and its lockdown. This study aims to evaluate whether GV is an independent predictor of all-cause mortality in patients aged 75 years or older with and without COVID-19 who were followed during the first year of the COVID-19 pandemic and its lockdown measures. Methods This was a retrospective cohort study of 407,492 patients from the AGED-MADRID dataset aged 75 years and older; 63.2% were women, and 29.3% had diabetes. GV was measured by the coefficient of variation of fasting plasma glucose (CV-FPG) over 6 years of follow-up (2015–2020). The outcome measure was all-cause mortality in 2020. Four models of logistic regression were performed, from simple (age, sex) to fully adjusted, to assess the effect of CV-FPG on all-cause mortality. Results During follow-up, 34,925 patients died (14,999 women and 19,926 men), with an all-cause mortality rate of 822.3 per 10,000 person-years (95% confidence interval (CI), 813.7 to 822.3) (739 per 10,000; 95% CI 728.7 to 739.0 in women and 967.1 per 10,000; 95% CI 951.7 to 967.2 in men). The highest quartile of CV-FPG was significantly more common in the deceased group (40.1% vs. 23.6%; p < 0.001). The fully adjusted model, including basal FPG, showed an odds ratio for mortality that ranged from 2.48 to 2.88 according to different sensitivity analyses. Conclusions GV has clear implications for clinical practice, as its assessment as a risk prediction tool should be included in the routine follow-up of the elderly and in a comprehensive geriatric assessment. Electronic health records can incorporate tools that allow its calculation, and with this information, clinicians will have a broader view of the medium- and long-term prognosis of their patients.