Abstract
Context
Glycemic variation had been demonstrated to be associated with several complications of diabetes.
Objective
Investigation of the association between visit to visit hemoglobin A1c (HbA1c) variation and the long-term risk of major adverse limb events (MALEs).
Methods
Retrospective database study. Average real variability was used to represent glycemic variations with all the HbA1c measurements during the 4 following years after the initial diagnosis of type 2 diabetes. Participants were followed from the beginning of the fifth year until death or the end of the follow-up. The association between HbA1c variations and MALEs was evaluated after adjusting for mean HbA1c and baseline characteristics. Included were 56 872 patients at the referral center with a first diagnosis of type 2 diabetes, no lower extremity arterial disease, and at least 1 HbA1c measurement in each of the 4 following years were identified from a multicenter database. The main outcome measure was incidence of a MALE, which was defined as the composite of revascularization, foot ulcers, and lower limb amputations.
Results
The average number of HbA1c measurements was 12.6. The mean follow-up time was 6.1 years. The cumulative incidence of MALEs was 9.25 per 1000 person-years. Visit to visit HbA1c variations were significantly associated with MALEs and lower limb amputation after multivariate adjustment. People in the highest quartile of variations had increased risks for MALEs (HR 1.25, 95% CI 1.10-1.41) and lower limb amputation (HR 3.05, 95% CI 1.97-4.74).
Conclusion
HbA1c variation was independently associated with a long-term risk of MALEs and lower limb amputations in patients with type 2 diabetes.
Keywords: type 2 diabetes, glycemic variability, lower extremity arterial disease, major adverse limb events, lower limb amputation
Lower extremity arterial disease (LEAD) is a clinical manifestation of atherosclerosis and a marker of atherothrombotic disease in other vascular beds in patients with type 2 diabetes, which is associated with high risks of foot ulcers, nontraumatic amputation, major adverse cardiovascular events, major adverse limb events (MALEs), and all-cause mortality (1, 2). The prevalence of LEAD is 2 to 4 times higher in people with type 2 diabetes than in the general population (3, 4). Diabetes is a predictor for worse outcomes in patients with LEAD, including increased risk of MALEs (such as revascularization, critical limb ischemia, and amputation), and mortality (5). Moreover, patients with a MALE exhibit significantly higher risks of subsequent death or vascular amputation (6).
In patients with diabetes, conventional risk factors for MALEs include smoking, long-term diabetes, poor glycemic control, hypertension, hypercholesterolemia, prior foot ulcers, prior amputation, insulin dependence, chronic kidney disease, and neuropathy (3, 7, 8). Glycemic variability (GV) is a measure of fluctuations in blood glucose in a given interval of time and is as a novel predictor for several diabetic complications (9). Plausible biological mechanisms underlying GV in the pathogenesis of diabetic complications include hypoglycemia, hyperglycemia, oxidative stress, inflammatory cytokines, platelet activation, and endothelial dysfunction (10, 11). However, data on the association between GV and the risk of MALEs is limited. Therefore, this study investigated the association between visit to visit glycated hemoglobin (HbA1c) variations and risk of MALEs in more than 50 000 patients with type 2 diabetes by using a large multicenter-based electronic medical registry database.
Material and Methods
Database
Patients for this study were identified from data in the Chang Gung Research Database, which are derived from the original patient-level electronic medical records of the Chang Gung Memorial Hospital group, the largest health care provider in Taiwan, which comprises 7 institutes. The database has compiled the integrated and standardized electronic medical records of 1.3 million patients since 2000. An advantage of the database over other claims databases is that it includes detailed laboratory and examination results. The data structure and representativeness of the database have been detailed, and many diagnostic codes have been validated elsewhere (12). Disease diagnosis in the in database was based on the International Classification of Disease, Clinical Modification diagnostic codes.
Study Design
Figure 1 presents the patient recruitment process. A total of 163 445 patients with diabetes between January 1, 2007, and December 31, 2010, were identified. We excluded patients who were aged younger than 18 years, had type 1 diabetes, and had known LEAD. Diabetes and LEAD were defined as any inpatient or at least 2 outpatient diagnoses.
Figure 1.
Flowchart of patient enrollment in the study. From January 1, 2007, to December 31, 2010, a total of 56 872 patients with type 2 diabetes and at least 1 HbA1c examination in each of the subsequent 4 run-in years were enrolled in the present study. Patients were followed up from the end of the run-in period until the occurrence of the primary outcome, mortality, the latest visit date, or the end of the study period (December 31, 2018). HbA1c, glycated hemoglobin A1c.
Currently, there is no consensus regarding the optimal or minimal period of time required to capture HbA1c or fasting plasma glucose (FPG) measurements to reflect long-term GV (13). Based mainly on a large retrospective cohort study investigating the association between HbA1c variability and mortality, the 48 months following the diagnosis of diabetes was defined as the run-in period (14). All HbA1c measurements taken during the run-in period were identified for GV calculation. Patients who had no prescriptions of glucose-lowering therapies and those who had all HbA1c measurements <6.5% during the run-in period were excluded. To calculate GV by using visit to visit HbA1c, we excluded those who had no HbA1c measurement at any 12-month interval during the run-in period. Patients who died or had MALEs during the run-in period and who were lost to follow-up within the first year of the observational period were excluded. After the exclusion of these patients, 56 872 adult patients (29 255 male, 51.4%) with type 2 diabetes remained for analysis. The index date was the next day after the end of the run-in period. Patients were followed up from the index date until the occurrence of the primary outcome, death, the latest visit date, or the end of the study period (December 31, 2018). This study was approved by the Institutional Review Board of Chang Gung Memorial Hospital, Linkou (IRB number 201701401B). Each patient's personal information was de-identified by an encrypting procedure; therefore, informed consent was waived.
Covariates
Baseline covariates were demographics (age, sex, body mass index, and smoking), comorbidities (hypertension and 10 others), as well as the Charlson Comorbidity Index, medical therapies (glucose-lowering therapies, antihypertensives, and lipid-lowering and antithrombotic therapies), systolic and diastolic blood pressure, and laboratory findings. In addition, all measurements of systolic blood pressure, diastolic blood pressure, total cholesterol, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, and triglyceride during the run-in period were identified and are presented as mean values. Comorbidities were defined if any inpatient or at least 2 outpatient diagnoses were recorded before or on the index date. Chronic kidney disease was defined as 2 consecutive reports of estimated glomerular filtration rate <60 mL/min/1.73 m2 before the index date. Data on body mass index, medications, and laboratory findings 3 months before the index date were obtained. Hypoglycemic and hyperglycemic events requiring admission to the emergency department or hospitalization during the run-in period were also identified.
Outcomes
The primary outcome was MALEs, defined as the composite of, endovascular therapies or bypass surgery for LEAD, newly developed LEAD-related foot ulcers, or nontraumatic above-ankle lower limb amputation (LLA). Endovascular therapies or bypass surgery for LEAD was defined as having a procedure code plus a discharge diagnosis of LEAD. Newly developed foot ulcers were defined as having at least 2 new-onset outpatient diagnoses or 1 new-onset discharge diagnosis of foot ulcers plus a diagnosis of LEAD. LLA was defined as having a procedure code for above-ankle amputation along with a discharge diagnosis of LEAD. Additional outcomes were all-cause mortality. The codes for diagnoses and procedures are presented elsewhere (Table 1 (15)). Patients were followed up from the next day of the end of the 4 run-in years until death, the latest visit date in the database, or the end of the study period (December 31, 2018).
Visit to Visit Variability of HbA1c
GV was derived using all the visit to visit HbA1c measurements taken during the run-in period. Four metrics were used to represent GV, namely average real variability (ARV) (16), variability independent of the mean (VIM) (17), SD, and coefficient of variation (CV). The HbA1c variation (eg, SD, ARV) is likely to be positively correlated with a greater mean HbA1c level. Two variation indices are frequently used to consider the effect of mean level: CV and VIM. VIM is calculated as the SD divided by the mean to the power x and multiplied by the population mean to the power x. Compared with CV, VIM more thoroughly deals with the collinearity between variation and mean level. ARV averages the absolute differences between any 2 successive measurements, which might be a reliable index for time series variability. In this study, we opted for ARV primarily because it aligns with clinical practice and is more intuitive to comprehend and calculate than other measures.
Statistical Analysis
The linear trend of patient demographics and characteristics across different extents of GV (represented as the quartiles of ARV) was tested using the linear contrast of the general linear model for continuous variables or Cochran–Armitage analysis for categorical variables. The association between GV and patient outcomes (MALEs, LLA, and all-cause mortality) was evaluated using the Cox proportional hazard model, in which the quartiles of GV were considered the explanatory variable and the first quartile (the smallest GV group) was used as the reference category. Three models with the following adjustment statuses were used: (1) Model 1, unadjusted; (2) Model 2, adjusted for age, sex, body mass index, smoking, all comorbidities, all laboratory data, all medications, hypoglycemic or hyperglycemic events, and the mean of systolic/diastolic blood pressure, total cholesterol, LDL cholesterol, HDL cholesterol, and triglyceride during the 4 run-in years; and (3) Model 3, further adjusted for the average HbA1c level during the 4 run-in years. We also conducted a sensitivity analysis by calculating the variability based on HbA1c measurements by defining the run-in period as 2 and 3 years. In an alternative Cox model, the quartile of GV was considered to be a continuous variable. The linear trend across the quartile of GV for the risk of outcomes was also tested. A subgroup analysis was also performed to assess the association between ARV and the risk of each outcome, stratified by the mean HbA1c level (<6%, 6-8%, and ≥8%) during the run-in period. Because some values of blood pressure and laboratory data were missing, the data were imputed using a single expectation–maximization algorithm. A 2-sided P < .05 was considered to be statistically significant. All statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA).
Results
Baseline Characteristics
A total of 56 872 patients with type 2 diabetes (29 255 men, 51.4%; average age, 64 ± 11.6 years) were enrolled for analysis. Approximately 70% of the patients had hypertension or dyslipidemia, and 12.3% of the participants were current or previous smokers. The most prevalent comorbidities were chronic kidney disease (23.2%) and coronary heart disease (18.9%). Moreover, 5.8% and 2.2% of the patients had prior stroke and myocardial infarction, respectively. The average number of HbA1c measurements during the 4 run-in years was 12.6 ± 3.9, and the average follow-up time was 6.1 ± 2.1 years.
Table 1 presents the baseline characteristics of the patients with type 2 diabetes categorized into quartiles of visit to visit variations of HbA1c represented by ARV. In general, patients with a higher GV were younger, more likely to be male and smokers, and have more comorbidities. Patients with a higher GV had a lower estimated glomerular filtration rates and a higher triglyceride and LDL. A higher proportion of patients in higher GV quartiles received ezetimibe, fenofibrate, antiplatelet agents, and multiple glucose-lowering therapies.
Table 1.
Demographics and characteristics of patients by variability in glycated hemoglobin (represented using ARV)
| Variable | Available number | Total | Q1 (n = 14 244) | Q2 (n = 14 181) | Q3 (n = 14 240) | Q4 (n = 14 207) | P trend |
|---|---|---|---|---|---|---|---|
| Value of HbA1c ARV | — | — | ≤0.331 | 0.332-0.521 | 0.522-0.785 | >0.785 | |
| Demographics | |||||||
| Age, years | 56 872 | 64.0 ± 11.6 | 66.0 ± 11.0 | 64.8 ± 11.2 | 63.2 ± 11.4 | 62.1 ± 12.3 | <.001 |
| Male | 56 872 | 29 255 (51.4) | 7151 (50.20) | 7151 (50.4) | 7267 (51.0) | 7686 (54.1) | <.001 |
| BMI, kg/m2 | 39 206 | 26.3 ± 4.1 | 25.8 ± 3.8 | 26.3 ± 3.9 | 26.5 ± 4.1 | 26.7 ± 4.4 | <.001 |
| Weight status | 56 872 | <.001 | |||||
| Unknown | 17 666 (31.1) | 4437 (31.15) | 4453 (31.4) | 4391 (30.8) | 4385 (30.9) | ||
| <18.5 kg | 467 (0.8) | 125 (0.88) | 89 (0.6) | 113 (0.8) | 140 (1.0) | ||
| 18.5-23.9 kg | 10 960 (19.3) | 3084 (21.65) | 2739 (19.3) | 2602 (18.3) | 2535 (17.8) | ||
| 24-26.9 kg | 12 752 (22.4) | 3394 (23.83) | 3205 (22.6) | 3160 (22.2) | 2993 (21.1) | ||
| ≥27 kg | 15 027 (26.4) | 3204 (22.49) | 3695 (26.1) | 3974 (27.9) | 4154 (29.2) | ||
| Smoker | 56 872 | 6988 (12.3) | 1203 (8.45) | 1459 (10.3) | 1788 (12.6) | 2538 (17.9) | <.001 |
| Comorbidity | |||||||
| Hypertension | 56 872 | 40 774 (71.7) | 10 088 (70.82) | 10 177 (71.8) | 10 132 (71.2) | 10 377 (73.0) | <.001 |
| Dyslipidemia | 56 872 | 39 597 (69.6) | 9725 (68.27) | 9902 (69.8) | 10 024 (70.4) | 9946 (70.0) | .001 |
| Coronary heart disease | 56 872 | 10 734 (18.9) | 2593 (18.20) | 2727 (19.2) | 2612 (18.3) | 2802 (19.7) | .012 |
| Atrial fibrillation | 56 872 | 1824 (3.2) | 501 (3.52) | 426 (3.0) | 410 (2.9) | 487 (3.4) | .549 |
| Malignancy | 56 872 | 4848 (8.5) | 1228 (8.62) | 1220 (8.6) | 1166 (8.2) | 1234 (8.7) | .832 |
| Chronic kidney disease | 56 872 | 13 209 (23.2) | 2611 (18.33) | 3123 (22.0) | 3463 (24.3) | 4012 (28.2) | <.001 |
| Dementia | 56 872 | 2025 (3.6) | 477 (3.35) | 495 (3.5) | 472 (3.3) | 581 (4.1) | .003 |
| Myocardial infarction | 56 872 | 1247 (2.2) | 205 (1.44) | 273 (1.9) | 353 (2.5) | 416 (2.9) | <.001 |
| Stroke | 56 872 | 3299 (5.8) | 616 (4.32) | 690 (4.9) | 866 (6.1) | 1127 (7.9) | <.001 |
| Hospitalization for heart failure | 56 872 | 1497 (2.6) | 219 (1.54) | 295 (2.1) | 377 (2.6) | 606 (4.3) | <.001 |
| Dialysis | 56 872 | 1297 (2.3) | 145 (1.02) | 285 (2.0) | 364 (2.6) | 503 (3.5) | <.001 |
| Charlson Comorbidity Index | 56 872 | 2.9 ± 2.0 | 2.7 ± 1.8 | 2.9 ± 1.9 | 3.0 ± 2.0 | 3.2 ± 2.2 | <.001 |
| Laboratory data | |||||||
| Blood urea nitrogen, mg/dL | 20 355 | 25.5 ± 20.5 | 22.1 ± 17.2 | 24.9 ± 20.5 | 25.9 ± 20.4 | 28.3 ± 22.4 | <.001 |
| Creatinine, mg/dL | 55 526 | 1.2 ± 1.4 | 1.0 ± 1.0 | 1.2 ± 1.4 | 1.2 ± 1.4 | 1.4 ± 1.7 | <.001 |
| eGFR, mL/min/1.73 m2 | 55 526 | 76.6 ± 33.6 | 79.1 ± 30.2 | 77.5 ± 31.6 | 76.9 ± 34.6 | 72.8 ± 37.3 | <.001 |
| UACR | 15 131 | 210.6 ± 726.6 | 96.6 ± 372.0 | 160.2 ± 636.0 | 209.0 ± 664.5 | 382.3 ± 1050.0 | <.001 |
| High-density lipoprotein, mg/dL | 52 734 | 48.8 ± 13.6 | 51.5 ± 13.7 | 49.3 ± 13.3 | 48.1 ± 13.4 | 46.4 ± 13.3 | <.001 |
| Low-density lipoprotein, mg/dL | 53 759 | 101.2 ± 42.4 | 99.2 ± 33.4 | 99.3 ± 38.2 | 101.3 ± 43.2 | 105.0 ± 52.3 | <.001 |
| Triglyceride, mg/dL | 54 616 | 144.7 ± 125.0 | 125.0 ± 76.6 | 138.0 ± 108.3 | 149.4 ± 124.8 | 166.6 ± 168.8 | <.001 |
| Total cholesterol, mg/dL | 54 816 | 173.2 ± 36.0 | 172.3 ± 32.0 | 171.5 ± 33.9 | 173.1 ± 36.0 | 175.8 ± 41.2 | <.001 |
| Hemoglobin, g/dL | 25 777 | 12.6 ± 2.1 | 12.9 ± 1.9 | 12.7 ± 2.0 | 12.6 ± 2.1 | 12.4 ± 2.2 | <.001 |
| Antidiabetic drug | |||||||
| Metformin | 56 872 | 41 697 (73.3) | 9815 (68.91) | 10 771 (76.0) | 10 925 (76.7) | 10 186 (71.7) | <.001 |
| Sulfonylurea | 56 872 | 22 226 (39.1) | 4666 (32.76) | 6110 (43.1) | 5960 (41.9) | 5490 (38.6) | <.001 |
| Glinide | 56 872 | 3238 (5.7) | 716 (5.03) | 793 (5.6) | 802 (5.6) | 927 (6.5) | <.001 |
| Alpha-glucosidase inhibitors | 56 872 | 7681 (13.5) | 1260 (8.85) | 1726 (12.2) | 2114 (14.8) | 2581 (18.2) | <.001 |
| Thiazolidinedione | 56 872 | 7212 (12.7) | 1022 (7.17) | 1873 (13.2) | 2194 (15.4) | 2123 (14.9) | <.001 |
| DPP4i | 56 872 | 14 380 (25.3) | 1567 (11.00) | 3268 (23.0) | 4458 (31.3) | 5087 (35.8) | <.001 |
| Insulin | 56 872 | 8131 (14.3) | 404 (2.84) | 1351 (9.5) | 2654 (18.6) | 3722 (26.2) | <.001 |
| Number of drugs | 56 872 | 2.6 ± 1.4 | 1.7 ± 1.0 | 2.5 ± 1.2 | 2.9 ± 1.3 | 3.3 ± 1.4 | <.001 |
| Antihypertensive agent | |||||||
| ACEi | 56 872 | 4865 (8.6) | 1240 (8.71) | 1246 (8.8) | 1220 (8.6) | 1159 (8.2) | .076 |
| ARBs | 56 872 | 22 373 (39.3) | 5522 (38.77) | 5614 (39.6) | 5564 (39.1) | 5673 (39.9) | .104 |
| Beta blocker | 56 872 | 15 345 (27.0) | 3777 (26.52) | 3885 (27.4) | 3736 (26.2) | 3947 (27.8) | .113 |
| DCCBs | 56 872 | 16 229 (28.5) | 4113 (28.88) | 4079 (28.8) | 3983 (28.0) | 4054 (28.5) | .284 |
| Thiazide | 56 872 | 2675 (4.7) | 709 (4.98) | 679 (4.8) | 655 (4.6) | 632 (4.4) | .025 |
| MRAs | 56 872 | 698 (1.2) | 119 (0.84) | 140 (1.0) | 168 (1.2) | 271 (1.9) | <.001 |
| Alpha blocker | 56 872 | 1282 (2.3) | 320 (2.25) | 344 (2.4) | 306 (2.1) | 312 (2.2) | .442 |
| Other hypertension drugs | 56 872 | 7280 (12.8) | 1632 (11.46) | 1808 (12.7) | 1874 (13.2) | 1966 (13.8) | <.001 |
| Other medication | |||||||
| Statin | 56 872 | 26 840 (47.2) | 6625 (46.51) | 6763 (47.7) | 6843 (48.1) | 6609 (46.5) | .832 |
| Ezetimibe | 56 872 | 3471 (6.1) | 764 (5.36) | 823 (5.8) | 886 (6.2) | 998 (7.0) | <.001 |
| Fibrate | 56 872 | 6051 (10.6) | 1057 (7.42) | 1376 (9.7) | 1681 (11.8) | 1937 (13.6) | <.001 |
| Aspirin | 56 872 | 16 921 (29.8) | 4021 (28.23) | 4267 (30.1) | 4294 (30.2) | 4339 (30.5) | <.001 |
| Clopidogrel | 56 872 | 3377 (5.9) | 784 (5.50) | 823 (5.8) | 805 (5.7) | 965 (6.8) | <.001 |
| Cilostazol | 56 872 | 651 (1.1) | 100 (0.70) | 156 (1.1) | 177 (1.2) | 218 (1.5) | <.001 |
| Warfarin | 56 872 | 731 (1.3) | 197 (1.38) | 185 (1.3) | 169 (1.2) | 180 (1.3) | .270 |
| The average level in the 4 run-in years | |||||||
| Low-density lipoprotein | 55 675 | 102.3 ± 24.5 | 101.0 ± 22.3 | 100.7 ± 22.6 | 102.3 ± 24.1 | 105.3 ± 28.1 | <.001 |
| Systolic blood pressure | 54 170 | 139.6 ± 14.2 | 137.5 ± 13.4 | 139.7 ± 14.0 | 140.4 ± 14.1 | 140.9 ± 15.0 | <.001 |
| Diastolic blood pressure | 54 166 | 76.2 ± 8.3 | 75.0 ± 8.0 | 75.9 ± 8.2 | 76.5 ± 8.2 | 77.4 ± 8.7 | <.001 |
| TChol | 56 546 | 176.5 ± 26.5 | 174.6 ± 24.7 | 174.6 ± 24.8 | 176.6 ± 26.2 | 180.4 ± 29.7 | <.001 |
| HDL | 55 507 | 47.1 ± 11.6 | 49.4 ± 12.0 | 47.4 ± 11.4 | 46.4 ± 11.4 | 45.1 ± 11.3 | <.001 |
| TG | 56 406 | 153.2 ± 86.2 | 132.0 ± 63.8 | 146.3 ± 76.2 | 158.0 ± 88.4 | 176.4 ± 104.8 | <.001 |
| Hypoglycemia during run-in period | 1759 (3.1) | 232 (1.63) | 327 (2.3) | 513 (3.6) | 687 (4.8) | <.001 | |
| Hyperglycemia during run-in period | 1520 (2.7) | 48 (0.34) | 130 (0.9) | 364 (2.6) | 978 (6.9) | <.001 | |
| Information of HbA1c | |||||||
| Average level in the 1st year | 56 872 | 7.9 ± 1.6 | 6.7 ± 0.7 | 7.5 ± 1.0 | 8.3 ± 1.4 | 9.1 ± 1.9 | <.001 |
| Average level in the 2nd year | 56 872 | 7.6 ± 1.4 | 6.6 ± 0.6 | 7.3 ± 1.0 | 7.9 ± 1.3 | 8.6 ± 1.7 | <.001 |
| Average level in the 3rd year | 56 872 | 7.6 ± 1.4 | 6.6 ± 0.6 | 7.3 ± 1.0 | 7.9 ± 1.3 | 8.6 ± 1.6 | <.001 |
| Average level in the 4th year | 56 872 | 7.6 ± 1.4 | 6.7 ± 0.7 | 7.3 ± 1.0 | 7.9 ± 1.3 | 8.5 ± 1.6 | <.001 |
| Average level in the 4 run-in years | 56 872 | 7.7 ± 1.2 | 6.7 ± 0.6 | 7.3 ± 0.9 | 8.0 ± 1.1 | 8.7 ± 1.2 | <.001 |
| Number of HbA1c examinations in the 4 run-in years | 56 872 | 12.6 ± 3.9 | 12.4 ± 4.0 | 12.9 ± 3.9 | 13.0 ± 3.7 | 11.9 ± 3.6 | <.001 |
| Follow-up years | 56 872 | 6.1 ± 2.1 | 6.3 ± 1.9 | 6.3 ± 2.0 | 6.2 ± 2.1 | 5.8 ± 2.2 | <.001 |
Data are presented as frequencies (percentages) or means ± SD.
Abbreviations: Q, quartile; ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; ARV, average real variability; DCCB, dehydropyridine calcium-channel blocker; DPP4i, dipeptidyl peptidase-4 inhibitor; eGFR, estimated glomerular filtration rate; HbA1c, hemoglobin A1c; HDL, high-density lipoprotein cholesterol; MRA, mineralocorticoid receptor antagonist; TChol, total cholesterol; TG, triglyceride; UACR, urine albumin to creatinine ratio.
Patients in a higher GV quartile had higher mean HbA1c values during the run-in period. Moreover, hypoglycemia or hyperglycemic events requiring hospital visits were significantly more prevalent in patients with higher GV. The mean systolic/diastolic blood pressure, total cholesterol, LDL cholesterol, and triglyceride values during the run-in period increased from the first to the fourth GV quartiles. By contrast, the mean HDL cholesterol values were lower in patients with higher GV.
Visit to Visit HbA1c Variations (by ARV) and Clinical Outcomes
The incidence of MALEs was 9.25 events per 1000 person-years (9.73 for men and 8.74 for women); that of LLA was 1.25 events per 1000 person-years (1.40 for men and 1.08 for women); and that of death was 11 events per 1000 person-years (11.32 for men and 10.66 for women; data not shown). The cumulative incidence curves of MALEs, LLA, and all-cause mortality against visit to visit HbA1c variations by ARV are presented in Fig. 2. The risk of MALEs was higher in patients with third (0.522 to 0.785) and fourth (>0.785) ARV quartiles (log-rank P < .001). The cumulative incidence of MALEs was similar between patients in the first and second GV quartiles (Fig. 2A). Patients in the higher quartiles had a higher cumulative incidence of LLA (P for trend < .001). A dose–response relationship was observed, except for those in the second and third GV quartiles, who exhibited similar risks for LLA (Fig. 2B). A dose–response relationship was observed between visit to visit HbA1c variations by ARV and all-cause mortality during the long-term follow-up (P for trend <.001; Fig. 2C).
Figure 2.
Cumulative incidence curves of major adverse limb events (MALEs), lower limb amputation (LLA), and all-cause mortality among patients in different quartiles of visit-to-visit HbA1c variations by average real variability (ARV). The risk of MALEs was higher in patients in the third (0.522 to 0.785) and fourth (>0.785) quartiles of HbA1c ARV (log-rank P < .001). The cumulative incidence of MALEs was comparable between patients in the first and second quartiles (A). Patients in the higher quartiles of HbA1c ARV had higher cumulative incidence of LLA (P for trend < .001). A dose–response relationship was observed in all patients, except in those in the second and third quartiles with a comparable risk of LLA (B). An evident dose–response relationship was noted between visit to visit HbA1c variations by ARV and all-cause mortality at long-term follow-up (P for trend < .001) (C). ARV, average real variability; HbA1c, glycated hemoglobin A1c; LLA, lower limb amputation; MALE, major adverse limb event.
Sensitivity Analysis
Table 2 (also Tables 2 and Table 3 (15)) present the association of the risks of incident MALEs, LLA, and all-cause mortality with an increment of visit to visit HbA1c variations before and after multivariate adjustment. An increase in the risk of incident MALEs, LLA, and all-cause mortality was noted with visit to visit HbA1c variation before and after multivariate adjustment. This finding was consistent for all multivariate adjustments (models 2 and 3) and metrics (VIM, SD, and CV) used to calculate the HbA1c variations.
Table 2.
Association between variability in glycated hemoglobin and risk of major adverse limb events
| Variability index | N | n (%) | Model 1 | Model 2 | Model 3 | |||
|---|---|---|---|---|---|---|---|---|
| HR (95% CI) | P | HR (95% CI) | P | HR (95% CI) | P | |||
| VIM quartile (P trend) | (<.001) | (.002) | (.012) | |||||
| Q1: ≤ 0.011 | 14 218 | 568 (4.0) | Reference | Reference | Reference | |||
| Q2: 0.012-0.016 | 14 218 | 727 (5.1) | 1.30 (1.17-1.45) | <.001 | 1.11 (0.99-1.24) | .077 | 1.07 (0.96-1.20) | .243 |
| Q3: 0.017-0.024 | 14 218 | 895 (6.3) | 1.68 (1.51-1.86) | <.001 | 1.20 (1.08-1.34) | .001 | 1.15 (1.03-1.28) | .012 |
| Q4: >0.024 | 14 218 | 941 (6.6) | 1.89 (1.70-2.10) | <.001 | 1.18 (1.06-1.32) | .004 | 1.14 (1.02-1.28) | .021 |
| SD quartile (P trend) | (<.001) | (<.001) | (<.001) | |||||
| Q1: ≤0.433 | 14 218 | 542 (3.8) | Reference | Reference | Reference | |||
| Q2: 0.434-0.704 | 14 219 | 681 (4.8) | 1.26 (1.12-1.41) | <.001 | 1.10 (0.98-1.24) | .096 | 1.04 (0.93-1.17) | .504 |
| Q3: 0.705-1.103 | 14 217 | 834 (5.9) | 1.58 (1.42-1.76) | <0.001 | 1.27 (1.13-1.42) | <.001 | 1.14 (1.01-1.28) | .033 |
| Q4: >1.103 | 14 218 | 1074 (7.6) | 2.21 (1.99-2.45) | <.001 | 1.47 (1.31-1.65) | <.001 | 1.26 (1.11-1.42) | <.001 |
| CV quartile (P trend) | (<.001) | (<.001) | (<.001) | |||||
| Q1: ≤0.061 | 14 218 | 536 (3.8) | Reference | Reference | Reference | |||
| Q2: 0.062-0.091 | 14 219 | 704 (5.0) | 1.33 (1.19-1.48) | <.001 | 1.12 (1.00-1.26) | .048 | 1.06 (0.94-1.19) | .329 |
| Q3: 0.092-0.136 | 14 217 | 879 (6.2) | 1.72 (1.54-1.91) | <.001 | 1.28 (1.14-1.43) | <.001 | 1.17 (1.04-1.31) | .008 |
| Q4: >0.136 | 14 218 | 1012 (7.1) | 2.14 (1.92-2.37) | <.001 | 1.36 (1.21-1.52) | <.001 | 1.22 (1.09-1.37) | .001 |
| ARV quartile (P trend) | (<.001) | (<.001) | (<.001) | |||||
| Q1: ≤0.331 | 14 244 | 547 (3.8) | Reference | Reference | Reference | |||
| Q2: 0.332-0.521 | 14 181 | 655 (4.6) | 1.21 (1.08-1.35) | .001 | 1.06 (0.94-1.19) | .349 | 1.003 (0.89-1.13) | .956 |
| Q3: 0.522-0.785 | 14 240 | 821 (5.8) | 1.54 (1.38-1.72) | <.001 | 1.25 (1.11-1.40) | <.001 | 1.12 (1.00-1.26) | .054 |
| Q4: >0.785 | 14 207 | 1108 (7.8) | 2.24 (2.03-2.49) | <.001 | 1.48 (1.32-1.66) | <.001 | 1.25 (1.10-1.41) | <.001 |
Model 1: Unadjusted.
Model 2: Adjusted for age, sex, body mass index, smoking, all comorbidities, all laboratory data, all medications, the average of lipid profiles (low-density lipoprotein, high-density lipoprotein, total cholesterol, and triglyceride), systolic blood pressure, diastolic blood pressure during the 4 run-in years, hypoglycemia during the run-in period, and hyperglycemia during the run-in period.
Model 3: Further adjusted for the average HbA1c level during the 4 run-in years.
Abbreviations: ARV, average real variability; CV, coefficient of variation; HR, hazard ratio; Q, quartile; VIM, variability independent of the mean.
The association between visit to visit HbA1c variability and outcomes was consistent, regardless of whether the run-in period was defined as 2 or 3 years (Tables 4 and 5 (15)).
Subgroup Analysis
The results of subgroup analysis stratified by mean HbA1c during the run-in period are presented elsewhere (Table 6 (15)). Approximately one-third (n = 18 216; 32%) of the patients had an average HbA1c level of ≥8% during the run-in period. The observed association between ARV and the risk of MALEs, LLA, and all-cause mortality was consistent in patients with mean HbA1c ≥ 8% and those with mean HbA1c 6% to 8%. In contrast, patients with mean HbA1c <6% did not exhibit associations between HbA1c variation and outcomes. The plausible explanation for this observation is the minimal HbA1c variations in patients who maintain tight glycemic control. Specifically, in patients with a mean HbA1c <6%, the highest quartile of ARV was ≥0.317; whereas the lowest quartile of ARV was ≤0.331 in the entire cohort. Additionally, this finding may also be attributed to the limited number of patients with mean HbA1c <6% (n = 2068). Figure 3 shows the cumulative incidence curves of MALEs against mean HbA1c (<8% or ≥8%) and visit to visit HbA1c variation by ARV (<0.8 or >0.8). Patients with mean HbA1c ≥8% plus ARV >0.8 had the highest risk of MALEs; and patients with mean HbA1c <8% plus ARV <0.8 had the lowest risk. The risk of MALEs was comparable between those who had mean HbA1c ≥8% plus ARV <0.8 and those with mean HbA1c <8% plus ARV >0.8.
Figure 3.
Cumulative incidence curves of major adverse limb events (MALEs), among patients with (1) high mean HbA1c and high variations (visit to visit HbA1c variations by ARV); (2) high mean HbA1c and low variation; (3) low mean HbA1c and high variations; (4) low mean HbA1c and low variations. Patients with mean HbA1c ≥8% plus ARV >0.8 had the highest risk of MALEs; and patients with mean HbA1c <8% plus ARV <0.8 had the lowest risk of MALE. The risk of MALEs was comparable between those who had mean HbA1c ≥8% plus ARB <0.8 and those with mean HbA1c <8% plus ARV >0.8. ARV, average real variability; HbA1c, glycated hemoglobin A1c; MALE, major adverse limb event.
Discussion
Using a multi-institutional database in Taiwan, we observed that a higher initial visit to visit HbA1c variation was independently associated with long-term risks of incident MALEs, LLA, and all-cause mortality in more than 56 000 patients with type 2 diabetes during a mean follow-up period of 6.1 years. This finding remained consistent for all multivariate adjustments of the covariates and metrics used to calculate the HbA1c variations, thereby demonstrating the robustness of our results. Our findings suggest that keeping a stable visit to visit HbA1c variation is as least as important as lowering the mean HbA1c to the target level.
Few studies have investigated the association of various degrees of GV with the risk of LEAD in patients with type 2 diabetes. Yang et al identified 30 932 patients with type 2 diabetes from the National Diabetes Care Management Program in Taiwan and discovered an independent association between FPG variations by CV in the third tertile and LEAD after multivariate adjustment (HR 1.24; 95% CI 1.04-1.47) (18). A major limitation of the study is that the clinical presentations of LEAD were not clearly reported. In another nationwide retrospective cohort study that included 27 574 patients with type 2 diabetes, Li et al demonstrated that the patients in the third tertiles of HbA1c variation by CV were independently associated with the risk of minor LLA (HR 1.34; 95% CI 1.02-1.77) (19). The risk of LLA was also independently associated with long-term visit to visit HbA1c variation in the present study, and we included only major amputation above the ankle. The clinical presentations of LEAD are within the range of asymptomatic, claudication, and critical limb ischemia. In our study, only those with symptomatic LEAD were included, the majority of whom were assumed to have severe claudication, rest pain, or tissue loss. Our study comprised a larger number of patients, more HbA1c measurements for GV calculation, and 4 metrics representing GV compared with the 2 abovementioned studies, thereby elucidating the association between GV and all presentations within the spectrum of LEAD.
In addition to LEAD, peripheral neuropathy plays a crucial role in the pathophysiology of diabetic foot ulcers and subsequent revascularization or amputation (20). Moreover, peripheral neuropathy has been reported to be a predictor of MALEs (21-23). Several studies have demonstrated that higher long-term visit to visit HbA1c or FPG variations were associated with the risk and severity of peripheral neuropathy in patients with type 2 diabetes (24). In a cohort study, Yang et al indicated that FPG variation by CV was a potent predictor of diabetic polyneuropathy (25). In a case–control study, Pai et al reported that long-term FPG variation by CV was associated with the risk of painful peripheral neuropathy, evaluated using standardized questionnaires (26). In a cross-sectional study, visit to visit HbA1c variations by CV increased the risk of clinical manifestations of neuropathy and abnormalities in nerve conduction tests (27). The role of GV in the pathogenesis of MALEs could be explained by the association between GV, peripheral neuropathy, and LEAD.
Infection is another important factor in the pathogenesis of diabetic foot ulcers (20). Several studies have demonstrated that high short-term GV increased the risk of complications and mortality in diabetic patients admitted for various infectious diseases. Atamna et al reported that patients hospitalized for infectious diseases with high short-term GV had increased risks for bacteremia and 30-day and 5-year mortality (28). In a retrospective study, Takeishi et al also discovered that after multivariate adjustment, high short-term GV was associated with in-hospital mortality in patients with diabetes and infectious diseases hospitalized in nonintensive care units (29). In a large cross-sectional study, long-term visit to visit HbA1c variations by SD were independently associated with lower limb vascular events and ulceration/gangrene (30). In a single-center retrospective study of patients from a multidisciplinary diabetic foot clinic, the mean HbA1c and visit to visit HbA1c variations by SD were associated with the healing time of diabetic foot ulcers (31).
Oxidative stress is a major mechanism of diabetic complications (32). All 3 glycemic indices (hyperglycemia, GV, and hypoglycemia) have been reported to be associated with oxidative stress. Some data suggest that GV is associated with greater reactive oxygen species production than is hyperglycemia (33). Oxidative stress plays a major role in the biology of diabetic wound healing, peripheral neuropathy, and LEAD, all of which are considered major risk factors for and predictors of MALEs (34). Accumulating evidence has suggested that reactive oxygen species are crucial regulators of wound healing. Excessive or impaired detoxification of reactive oxygen species causes oxidative damage, which is the main cause of nonhealing chronic wounds (35). Oxidative stress also negatively affects the blood supply, structure, and metabolism of the peripheral nerve, which is crucial for the development of peripheral neuropathy in patients with type 2 diabetes (36, 37). Moreover, oxidative stress increases inflammatory cytokines, causes endothelial cell hyperplasia, reduces nitrogen oxide, impairs vasodilation, and promotes procoagulant biomarkers, all of which are involved in the development of diabetic microangiopathy and macroangiopathy (34, 38). In sum, GV increases oxidative stress, a crucial factor in the pathogenesis of microvasculopathy and macrovasculopathy, peripheral neuropathy, and delayed wound healing, thereby leading to MALEs in patients with diabetes.
In our study, GV by visit to visit HbA1c was strongly associated with mortality, which is consistent with the literature. Critchley et al reported that HbA1c variability (by CV) had a consistent dose–response relationship with all-cause mortality by using English primary care data on 58 832 patients with type 2 diabetes. The risk of mortality almost doubled in individuals with the most unstable HbA1c compared with those with the most stable HbA1c (HR 1.93; 95% CI 1.72-2.16) (39). A dose–response relationship between long-term HbA1c variation and all-cause mortality was also observed in this study, even after robust adjustment for confounding factors. In addition, the risk of mortality in patients in the highest quartile was 1.7 to 1.8 times higher than that in those in the lowest quartile; this result was consistent for all metrics used to calculate GV.
This study has some limitations. First, this was a retrospective observational study. The baseline characteristics differed among patients in different GV quartiles. Despite the meticulous identification of all possible predictors for MALEs or LLA, with adjustment through propensity score weighting, residual unmeasured confounders may exist. The database did not contain some crucial confounding factors, including patient adherence and patient socioeconomic status (40, 41). In addition, acute illness or major procedure/surgery during the run-in period was not identified, which may affect GV. Second, procedures and clinical events outside the Chang Gung Memorial Hospitals are not included in our database, which may have led to a loss to follow-up and an underestimation of the actual event numbers. Third, GV can be assessed through several methods. Although we used 4 of the most commonly reported metrics to assess long-term GV, we could not determine which metric has the optimal predictive ability for diabetic complications including MALEs and amputation. Fourth, most of the hospitals were referral centers, which may limit the generalizability of our findings to universal patients with diabetes. Finally, this study only enrolled Asian patients, and whether our results can be extrapolated to patients of other ethnicities remains unclear.
Conclusions
Initial visit to visit Hb1Ac variations were independently associated with long-term risks of incident MALEs, LLA, and all-cause mortality in patients with type 2 diabetes.
Acknowledgments
We would like to thank Alfred Hsing-Fen Lin for their assistance in statistical analysis during the completion of this article.
Abbreviations
- ARV
average real variability
- CV
coefficient of variation
- FPG
fasting plasma glucose
- GV
glycemic variability
- HbA1c
glycated hemoglobin
- HDL
high-density lipoprotein
- LDL
low-density lipoprotein
- LEAD
lower extremity arterial disease
- LLA
lower limb amputation
- MALE
major adverse limb event
- VIM
variability independent of the mean
Contributor Information
Fu-Chih Hsiao, Cardiovascular Department, Chang Gung Memorial Hospital, Linkou, Taoyuan 333, Taiwan.
Yi-Hsin Chan, Cardiovascular Department, Chang Gung Memorial Hospital, Linkou, Taoyuan 333, Taiwan; College of Medicine, Chang Gung University, Taoyuan 333, Taiwan; School of Traditional Chinese Medicine, College of Medicine, Chang Gung University, Taoyuan 333, Taiwan; Microscopy Core Laboratory, Chang Gung Memorial Hospital, Linkou, Taoyuan 333, Taiwan.
Ying-Chang Tung, Cardiovascular Department, Chang Gung Memorial Hospital, Linkou, Taoyuan 333, Taiwan.
Chia-Pin Lin, Cardiovascular Department, Chang Gung Memorial Hospital, Linkou, Taoyuan 333, Taiwan.
Ting-Hein Lee, Cardiovascular Department, Chang Gung Memorial Hospital, Linkou, Taoyuan 333, Taiwan; Department of Anatomy, College of Medicine, Chang Gung University, Taoyuan 333, Taiwan; Graduate Institute of Biomedical Sciences, College of Medicine, Chang Gung University, Taoyuan 333, Taiwan.
Yu-Chiang Wang, Department of Medicine, Brigham and Women's Hospital, Boston, MA 02115, USA; School of Medicine, Harvard University, Boston, MA 02115, USA.
Pao-Hsien Chu, Cardiovascular Department, Chang Gung Memorial Hospital, Linkou, Taoyuan 333, Taiwan; College of Medicine, Chang Gung University, Taoyuan 333, Taiwan; Institute of Stem Cell and Translational Cancer Research, Chang Gung Memorial Hospital, Linkou, Taoyuan 333, Taiwan.
Funding
This study was supported by grant NMRP 110-2314-B-182A-120 and 111-2314-B-182A-013-MY3 from the Ministry of Science and Technology and grant CIRPG3L0021 and CMRPG3L1051 from the Chang Gung Memorial Hospital of Linkou, Taiwan. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Disclosures
The authors declare that they have no competing interests.
Data Availability
The data that support the findings of this study are available from Chang Gung Memorial Hospital but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of Chang Gung Memorial Hospital.
References
- 1. Criqui MH, Matsushita K, Aboyans V, et al. Lower extremity peripheral artery disease. Circulation. 2021;144(9):e171‐e191. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Nativel M, Potier L, Alexandre L, et al. Lower extremity arterial disease in patients with diabetes: a contemporary narrative review. Cardiovasc Diabetol. 2018;17(1):138. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Criqui MH, Aboyans V. Epidemiology of peripheral artery disease. Circ Res. 2015;116(9):1509‐1526. [DOI] [PubMed] [Google Scholar]
- 4. Selvin E, Erlinger TP. Prevalence of and risk factors for peripheral arterial disease in the United States: results from the National Health and Nutrition Examination Survey, 1999-2000. Circulation. 2004;110(6):738‐743. [DOI] [PubMed] [Google Scholar]
- 5. Jude EB, Oyibo SO, Chalmers N, Boulton AJ. Peripheral arterial disease in diabetic and nondiabetic patients: a comparison of severity and outcome. Diabetes Care. 2001;24(8):1433‐1437. [DOI] [PubMed] [Google Scholar]
- 6. Anand SS, Caron F, Eikelboom JW, et al. Major adverse limb events and mortality in patients with peripheral artery disease: the COMPASS trial. J Am Coll Cardiol. 2018;71(20):2306‐2315. [DOI] [PubMed] [Google Scholar]
- 7. American Diabetes Association . Microvascular complications and foot care. Sec. 10. In Standards of Medical Care in Diabetes-2017. Diabetes Care. 2017;40(Suppl 1):S88‐S98. Erratum: Diabetes Care. 2017; 40(7):986. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Weissler EH, Clare RM, Lokhnygina Y, et al. Predicting major adverse limb events in individuals with type 2 diabetes: insights from the EXSCEL trial. Diabet Med. 2021;38(10):e14552. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Zhou Z, Sun B, Huang S, Zhu C, Bian M. Glycemic variability: adverse clinical outcomes and how to improve it? Cardiovasc Diabetol. 2020;19(1):102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Hirakawa Y, Arima H, Zoungas S, et al. Impact of visit-to-visit glycemic variability on the risks of macrovascular and microvascular events and all-cause mortality in type 2 diabetes: the ADVANCE trial. Diabetes Care. 2014;37(8):2359‐2365. [DOI] [PubMed] [Google Scholar]
- 11. Zhou JJ, Schwenke DC, Bahn G, Reaven P, Investigators V. Glycemic variation and cardiovascular risk in the veterans affairs diabetes trial. Diabetes Care. 2018;41(10):2187‐2194. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Shao SC, Chan YY, Kao Yang YH, et al. The Chang Gung Research Database—a multi-institutional electronic medical records database for real-world epidemiological studies in Taiwan. Pharmacoepidemiol Drug Saf. 2019;28(5):593‐600. [DOI] [PubMed] [Google Scholar]
- 13. Gorst C, Kwok CS, Aslam S, et al. Long-term glycemic variability and risk of adverse outcomes: a systematic review and meta-analysis. Diabetes Care. 2015;38(12):2354‐2369. [DOI] [PubMed] [Google Scholar]
- 14. Forbes A, Murrells T, Mulnier H, Sinclair AJ. Mean HbA(1c), HbA(1c) variability, and mortality in people with diabetes aged 70 years and older: a retrospective cohort study. Lancet Diabetes Endocrinol. 2018;6(6):476‐486. [DOI] [PubMed] [Google Scholar]
- 15. Hsiao FC, Chan YH, Tung YC, et al. Supplementary data for “Visit to visit hemoglobin A1c variation and long-term risk of major adverse limb events in patients with type 2 diabetes.” Figshare. Deposited March 29, 2023. 10.6084/m9.figshare.22354117.v1 [DOI] [PMC free article] [PubMed]
- 16. Mena L, Pintos S, Queipo NV, Aizpurua JA, Maestre G, Sulbaran T. A reliable index for the prognostic significance of blood pressure variability. J Hypertens. 2005;23(3):505‐511. [DOI] [PubMed] [Google Scholar]
- 17. Rothwell PM, Howard SC, Dolan E, et al. Prognostic significance of visit-to-visit variability, maximum systolic blood pressure, and episodic hypertension. Lancet. 2010;375(9718):895‐905. [DOI] [PubMed] [Google Scholar]
- 18. Yang CP, Lin CC, Li CI, et al. Fasting plasma glucose variability and HbA1c are associated with peripheral artery disease risk in type 2 diabetes. Cardiovasc Diabetol. 2020;19(1):4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Li CI, Cheng HM, Liu CS, et al. Association between glucose variation and lower extremity amputation incidence in individuals with type 2 diabetes: a nationwide retrospective cohort study. Diabetologia. 2020;63(1):194‐205. [DOI] [PubMed] [Google Scholar]
- 20. Bandyk DF. The diabetic foot: pathophysiology, evaluation, and treatment. Semin Vasc Surg. 2018;31(2-4):43‐48. [DOI] [PubMed] [Google Scholar]
- 21. Selvarajah D, Kar D, Khunti K, et al. Diabetic peripheral neuropathy: advances in diagnosis and strategies for screening and early intervention. Lancet Diabetes Endocrinol. 2019;7(12):938‐948. [DOI] [PubMed] [Google Scholar]
- 22. Schreuder SM, Nieuwdorp M, Koelemay MJW, Bipat S, Reekers JA. Testing the sympathetic nervous system of the foot has a high predictive value for early amputation in patients with diabetes with a neuroischemic ulcer. BMJ Open Diabetes Res Care. 2018;6(1):e000592. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Volmer-Thole M, Lobmann R. Neuropathy and diabetic foot syndrome. Int J Mol Sci. 2016;17(6):917. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Zhang X, Yang X, Sun B, Zhu C. Perspectives of glycemic variability in diabetic neuropathy: a comprehensive review. Commun Biol. 2021;4(1):1366. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Yang CP, Li CI, Liu CS, et al. Variability of fasting plasma glucose increased risks of diabetic polyneuropathy in T2DM. Neurology. 2017;88(10):944‐951. [DOI] [PubMed] [Google Scholar]
- 26. Pai YW, Lin CH, Lee IT, Chang MH. Variability of fasting plasma glucose and the risk of painful diabetic peripheral neuropathy in patients with type 2 diabetes. Diabetes Metab. 2018;44(2):129‐134. [DOI] [PubMed] [Google Scholar]
- 27. Su JB, Zhao LH, Zhang XL, et al. HbA1c variability and diabetic peripheral neuropathy in type 2 diabetic patients. Cardiovasc Diabetol. 2018;17(1):47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Atamna A, Ayada G, Akirov A, Shochat T, Bishara J, Elis A. High blood glucose variability is associated with bacteremia and mortality in patients hospitalized with acute infection. QJM. 2019;112(2):101‐106. [DOI] [PubMed] [Google Scholar]
- 29. Takeishi S, Mori A, Hachiya H, et al. Hypoglycemia and glycemic variability are associated with mortality in non-intensive care unit hospitalized infectious disease patients with diabetes mellitus. J Diabetes Investig. 2016;7(3):429‐435. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Penno G, Solini A, Zoppini G, et al. Hemoglobin A1c variability as an independent correlate of cardiovascular disease in patients with type 2 diabetes: a cross-sectional analysis of the renal insufficiency and cardiovascular events (RIACE) Italian multicenter study. Cardiovasc Diabetol. 2013;12(1):98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Dhatariya KK, Li Ping Wah-Pun Sin E, Cheng JOS, et al. The impact of glycaemic variability on wound healing in the diabetic foot—a retrospective study of new ulcers presenting to a specialist multidisciplinary foot clinic. Diabetes Res Clin Pract. 2018;135:23‐29. [DOI] [PubMed] [Google Scholar]
- 32. Brownlee M. Biochemistry and molecular cell biology of diabetic complications. Nature. 2001;414(6865):813‐820. [DOI] [PubMed] [Google Scholar]
- 33. Papachristoforou E, Lambadiari V, Maratou E, Makrilakis K. Association of glycemic indices (hyperglycemia, glucose variability, and hypoglycemia) with oxidative stress and diabetic complications. J Diabetes Res. 2020;2020:7489795. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Deng L, Du C, Song P, et al. The role of oxidative stress and antioxidants in diabetic wound healing. Oxid Med Cell Longev. 2021;2021:8852759. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Sanchez M C, Lancel S, Boulanger E, Neviere R. Targeting oxidative stress and mitochondrial dysfunction in the treatment of impaired wound healing: a systematic review. Antioxidants (Basel). 2018;7(8):98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Vincent AM, Russell JW, Low P, Feldman EL. Oxidative stress in the pathogenesis of diabetic neuropathy. Endocr Rev. 2004;25(4):612‐628. [DOI] [PubMed] [Google Scholar]
- 37. Obrosova IG. Update on the pathogenesis of diabetic neuropathy. Curr Diab Rep. 2003;3(6):439‐445. [DOI] [PubMed] [Google Scholar]
- 38. Steven S, Daiber A, Dopheide JF, Münzel T, Espinola-Klein C. Peripheral artery disease, redox signaling, oxidative stress—basic and clinical aspects. Redox Biol. 2017;12:787‐797. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Critchley JA, Carey IM, Harris T, DeWilde S, Cook DG. Variability in glycated hemoglobin and risk of poor outcomes among people with type 2 diabetes in a large primary care cohort study. Diabetes Care. 2019;42(12):2237‐2246. [DOI] [PubMed] [Google Scholar]
- 40. Venermo M, Manderbacka K, Ikonen T, Keskimäki I, Winell K, Sund R. Amputations and socioeconomic position among persons with diabetes mellitus, a population-based register study. BMJ Open. 2013;3(4):e002395. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Ha JH, Jin H, Park JU. Association between socioeconomic position and diabetic foot ulcer outcomes: a population-based cohort study in South Korea. BMC Public Health. 2021;21(1):1395. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data that support the findings of this study are available from Chang Gung Memorial Hospital but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of Chang Gung Memorial Hospital.