Abstract
The prevalence of diabetes increases with aging and reaches 14.5-26.7% in nursing home elderly dependent residents. In this population diabetes control is difficult, especially under insulin treatment, with an increased risk of severe hypoglycemia or chronic hyperglycemic excursions. HbA1c is of limited interest to detect hypoglycemia risk in this population.1,2 In the present report we describe EKILIDIAB™, an algorithm that allows one, using capillary blood glucose (BG), to classify the risk of hypoglycemia or hyperglycemic excursions. The algorithm is specific for the BG target in elderly dependent people, that is, 100-200 mg/dL according to the ADA guidelines, and allows one to monitor diabetes control every 3-4 weeks when BG is checked 2-3 times a day.
Studies were in compliance with the World Medical Association’s Declaration of Helsinki and were approved by the Ethical Committee of the State Geriatric Center. The algorithm was developed using Microsoft Excel (2010) and R Development Core Team. 3 Elderly dependent low blood glucose score (ED-LBGS) and elderly dependent high blood glucose score (ED-HBGS), which refers to the risk of hypoglycemia or chronic hyperglycemic excursions, respectively, were calculated according to Box and Cox 4 and Kovatchev et al 5 using 48,339 BG collected from medical records of 396 multiple injections insulin-treated residents. ED-LBGS and ED-HBGS are nonnegative numbers, the sum of which ranges from 0 to 100. The scores are based on a nonlinear transformation (resolved by the nleqslv R package 6 ) of the BG scale, applying symmetry (critical center: 140-150 mg/dL) to the distribution of BG readings for a subject:
Risk thresholds were calculated using a decision tree classification (CART) for ordinal variables. 7 The algorithm needs at least 60 BG values at a time.
A validation study was conducted during a two-month period. Informed consent was obtained from all subjects or legal guardians. 24 insulin-treated elderly dependent residents with diabetes either fairly controlled or with hypoglycemia or hyperglycemic excursion risks were investigated. Treatment was modified in 19 poorly controlled residents according to the algorithm diagnosis. New algorithm values were obtained 1 month later. In 5 residents with fair BG control, ED-LBGS and ED-HBGS showed comparable values at 1 month interval, demonstrating the reproducibility of the algorithm. Treatment modifications improved significantly BG control in 6/7 patients at risk for hypoglycemia resulting in a significant decrease in mean ED-LBGS and in hypoglycemic episodes (BG < 70 mg/dL). Chronic hyperglycemic excursions risk was unchanged in 5 patients and improved in 2 patients. Mean ED-HBGS and the number of hyperglycemic peaks (BG > 250 mg/dL) were not significantly changed. In all, 12 patients were at risk for hyperglycemic excursions (2 of them were also at risk for hypoglycemia). Treatment modifications improved significantly BG control in 9 patients and were ineffective in 3 patients, resulting in a significant decrease in mean ED-HBGS and in the number of hyperglycemic peaks. Hypoglycemia risk was worsened (from low to moderate) in 2 patients with no significant changes in mean ED-LBGS and in the number of hypoglycemic episodes (Table 1).
Algorithm Analysis of Blood Glucose Data Obtained From Elderly Dependent Residents With Diabetes in Long-Term Home Care.
Residents with fair control (upper part) or with hypoglycemia risks (middle part) or with hyperglycemic excursions risks (lower part) were studied. Elderly dependent low blood glucose score (ED-LBGS) and elderly dependent high blood glucose score (ED-HBGS) were measured with the algorithm at one month interval without (fairly controlled residents) or with treatment modification (poorly controlled residents). Hypoglycemia or hyperglycemic excursions classes risks are: 0: low, 1: moderate, 2: severe. Hypoglycemic episodes and hyperglycemic excursions were defined as blood glucose <70 or >250 mg/dL, respectively. Statistical analysis was performed using the paired Wilcoxon test (ED-LBGS and ED-HBGS) or the χ2 test (number of hypoglycemic episodes or of hyperglycemic excursions). Treatment modification: 0: no modification; 1: switch from basal to basal-bolus insulin regimen; 2: titrate basal and meal-time insulin to blood glucose and food intake with (*) or without irregular meal intake; 3: diet modification; 4: addition of metformin; 5: addition of insulin secretagogues (sulfonylurea, repaglinide, or GLP-1).
Footnotes
Acknowledgements
The authors thank Mrs F. Canavelli and Mr C. Gammelin (Department of Computational Systems) for their excellent technical assistance and Pr C. Deniau for his help in mathematic advices.
Abbreviations
ADA, American Diabetes Association; BG, blood glucose; ED-HBGS, elderly dependent high blood glucose score; ED-LBGS, elderly dependent low blood glucose score.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
