Patients with diabetes have a 10-fold higher risk of lower-extremity amputation (LEA) than people without diabetes (1). LEA is associated with the greatest reduction in quality of life and the greatest increase in mortality and medical costs in all diabetes complications. Previous studies suggested that the mean hemoglobin A1c (HbA1c) level was associated with an increased LEA risk among patients with type 2 diabetes (2). However, emerging evidence indicates that long-term glycemic variability evaluated by clinical visit-to-visit HbA1c variability may be a better predictor of diabetes complications (3). So far, the definitions of long-term HbA1c variability are inconsistent. In most studies, standard deviation of serial HbA1c measurements (HbA1c SD) and the intrapersonal coefficient of variation of HbA1c (HbA1c CV) are often used to represent HbA1c variability. In the current study, we added a new marker—HbA1c variability score (HVS) (4)—which is more easily applied to clinical practice.
We collected data from electronic health records for patients with type 2 diabetes between 2013 and 2019 in the Louisiana Experiment Assessing Diabetes outcomes (LEAD) cohort study (5). We excluded patients who had LEA diagnosis before entry and within 2 years after the first date of diabetes diagnosis, those with incomplete baseline data, those who did not have at least four HbA1c tests within 2 years after their first diagnosis of diabetes, and those who did not have at least five HbA1c measures between the date of diagnosis of diabetes and the date of diagnosis of the outcome. HbA1c SD was calculated within 2 years following the first date of type 2 diabetes diagnosis. HbA1c CV was calculated as the HbA1c SD divided by the mean value of HbA1c and then converted to a percentage. HVS was calculated as the percentage of the number of changes (increase or decrease) in HbA1c >0.5% (5.5 mmol/mol) from the value prior among all HbA1c measurements between the diagnosis of diabetes and LEA for each individual. We defined type 2 diabetes, LEA, and some other outcomes according to codes from ICD-9 or ICD-10, Clinical Modification, and Current Procedural Terminology (CPT) codes.
The present analysis included 30,039 patients after excluding ineligible patients. During a mean follow-up of 5.64 years, 286 participants had LEA. Multivariable-adjusted (age, sex, race, BMI, systolic blood pressure, LDL cholesterol, estimated glomerular filtration rate, smoking, mean value of HbA1c, peripheral arterial disease, foot deformity, and use of antihypertensive drugs, diabetes medications, lipid-lowering agents, and aspirin) hazard ratios (HRs) for LEA based on different levels of HVS (≤20%, >20% to ≤40%, >40% to ≤60%, >60% to ≤80%, and >80%) were 1.00, 1.00, 1.54, 1.70, and 3.31 (Ptrend < 0.001), respectively (Table 1). Multivariable-adjusted HRs for LEA events were 1.00, 1.35, 1.81, and 2.15 across quartiles of HbA1c SD (Ptrend = 0.012) and 1.00, 1.21, 1.35, and 1.88 across quartiles of HbA1c CV (Ptrend = 0.012). After additional adjustment for foot ulcers, the positive association with LEA risk was still significant for HVS but was no longer significant for either HbA1c SD or HbA1c CV. This can be explained as a history of foot ulcers and LEA being extremely relevant. A total of 266 of 286 incident cases of LEA had a history of foot ulcers.
HRs of LEA according to three different indicators of visit-to-visit HbA1c variability as categorical or continuous variables among patients with type 2 diabetes
| . | No. of participants . | No. of cases . | HRs (95% CI) . | |||
|---|---|---|---|---|---|---|
| Model 1 . | Model 2 . | Model 3 . | Model 4 . | |||
| HVS (%) | ||||||
| ≤20 | 11,189 | 42 | 1 | 1 | 1 | 1 |
| >20 to ≤40 | 7,189 | 41 | 1.48 (0.96–2.28) | 1.10 (0.71–1.70) | 1.00 (0.64–1.54) | 0.95 (0.61–1.48) |
| >40 to ≤60 | 5,817 | 71 | 3.04 (2.07–4.46) | 1.94 (1.30–2.90) | 1.54 (1.01–2.34) | 1.33 (0.88–2.00) |
| >60 to ≤80 | 4,289 | 76 | 4.54 (3.09–6.67) | 2.50 (1.67–3.76) | 1.70 (1.08–2.67) | 1.37 (0.90–2.13) |
| >80 | 1,555 | 56 | 10.0 (6.65–15.1) | 5.41 (3.50–8.34) | 3.31 (2.02–5.42) | 2.21 (1.36–3.59) |
| Ptrend | <0.001 | <0.001 | <0.001 | 0.004 | ||
| As a continuous variable | 1.03 (1.02–1.03) | 1.02 (1.02–1.03) | 1.01 (1.01–1.02) | 1.01 (1.00–1.02) | ||
| HbA1c SD | ||||||
| Quartile 1 | 7,588 | 27 | 1 | 1 | 1 | 1 |
| Quartile 2 | 7,531 | 43 | 1.81 (1.11–2.97) | 1.42 (0.86–2.32) | 1.35 (0.82–2.21) | 1.33 (0.81–2.19) |
| Quartile 3 | 7,543 | 87 | 3.48 (2.22–5.46) | 2.07 (1.30–3.28) | 1.81 (1.13–2.89) | 1.60 (1.00–2.56) |
| Quartile 4 | 7,377 | 129 | 5.42 (3.50–8.40) | 2.94 (1.86–4.63) | 2.15 (1.31–3.52) | 1.85 (1.13–3.02) |
| Ptrend | <0.001 | <0.001 | 0.012 | 0.084 | ||
| As a continuous variable | 1.64 (1.45–1.85) | 1.39 (1.21–1.60) | 1.18 (1.00–1.39) | 1.08 (0.92–1.28) | ||
| HbA1c CV | ||||||
| Quartile 1 | 7,561 | 32 | 1 | 1 | 1 | 1 |
| Quartile 2 | 7,575 | 50 | 1.67 (1.06–2.64) | 1.28 (0.81–2.02) | 1.21 (0.76–1.92) | 1.14 (0.72–1.81) |
| Quartile 3 | 7,556 | 81 | 2.68 (1.75–4.09) | 1.55 (1.01–2.40) | 1.35 (0.87–2.10) | 1.25 (0.80–1.93) |
| Quartile 4 | 7,347 | 123 | 4.45 (2.96–6.70) | 2.40 (1.58–3.67) | 1.88 (1.21–2.92) | 1.55 (1.00–2.39) |
| Ptrend | <0.001 | <0.001 | 0.012 | 0.157 | ||
| As a continuous variablea | 1.52 (1.33–1.74) | 1.32 (1.12–1.55) | 1.17 (1.00–1.40) | 1.08 (0.91–1.28) | ||
| . | No. of participants . | No. of cases . | HRs (95% CI) . | |||
|---|---|---|---|---|---|---|
| Model 1 . | Model 2 . | Model 3 . | Model 4 . | |||
| HVS (%) | ||||||
| ≤20 | 11,189 | 42 | 1 | 1 | 1 | 1 |
| >20 to ≤40 | 7,189 | 41 | 1.48 (0.96–2.28) | 1.10 (0.71–1.70) | 1.00 (0.64–1.54) | 0.95 (0.61–1.48) |
| >40 to ≤60 | 5,817 | 71 | 3.04 (2.07–4.46) | 1.94 (1.30–2.90) | 1.54 (1.01–2.34) | 1.33 (0.88–2.00) |
| >60 to ≤80 | 4,289 | 76 | 4.54 (3.09–6.67) | 2.50 (1.67–3.76) | 1.70 (1.08–2.67) | 1.37 (0.90–2.13) |
| >80 | 1,555 | 56 | 10.0 (6.65–15.1) | 5.41 (3.50–8.34) | 3.31 (2.02–5.42) | 2.21 (1.36–3.59) |
| Ptrend | <0.001 | <0.001 | <0.001 | 0.004 | ||
| As a continuous variable | 1.03 (1.02–1.03) | 1.02 (1.02–1.03) | 1.01 (1.01–1.02) | 1.01 (1.00–1.02) | ||
| HbA1c SD | ||||||
| Quartile 1 | 7,588 | 27 | 1 | 1 | 1 | 1 |
| Quartile 2 | 7,531 | 43 | 1.81 (1.11–2.97) | 1.42 (0.86–2.32) | 1.35 (0.82–2.21) | 1.33 (0.81–2.19) |
| Quartile 3 | 7,543 | 87 | 3.48 (2.22–5.46) | 2.07 (1.30–3.28) | 1.81 (1.13–2.89) | 1.60 (1.00–2.56) |
| Quartile 4 | 7,377 | 129 | 5.42 (3.50–8.40) | 2.94 (1.86–4.63) | 2.15 (1.31–3.52) | 1.85 (1.13–3.02) |
| Ptrend | <0.001 | <0.001 | 0.012 | 0.084 | ||
| As a continuous variable | 1.64 (1.45–1.85) | 1.39 (1.21–1.60) | 1.18 (1.00–1.39) | 1.08 (0.92–1.28) | ||
| HbA1c CV | ||||||
| Quartile 1 | 7,561 | 32 | 1 | 1 | 1 | 1 |
| Quartile 2 | 7,575 | 50 | 1.67 (1.06–2.64) | 1.28 (0.81–2.02) | 1.21 (0.76–1.92) | 1.14 (0.72–1.81) |
| Quartile 3 | 7,556 | 81 | 2.68 (1.75–4.09) | 1.55 (1.01–2.40) | 1.35 (0.87–2.10) | 1.25 (0.80–1.93) |
| Quartile 4 | 7,347 | 123 | 4.45 (2.96–6.70) | 2.40 (1.58–3.67) | 1.88 (1.21–2.92) | 1.55 (1.00–2.39) |
| Ptrend | <0.001 | <0.001 | 0.012 | 0.157 | ||
| As a continuous variablea | 1.52 (1.33–1.74) | 1.32 (1.12–1.55) | 1.17 (1.00–1.40) | 1.08 (0.91–1.28) | ||
Model 1 adjusted for age, sex, and race. Model 2 adjusted for covariates in model 1 plus baseline BMI, systolic blood pressure, LDL cholesterol, HDL cholesterol, triglycerides, estimated glomerular filtration rate, smoking, insurance type, use of antihypertensive drugs, use of antidiabetes medications, use of lipid-lowering agents, use of aspirin, peripheral arterial disease, and foot deformity. Model 3 adjusted for covariates in model 2 plus mean value of HbA1c. Model 4 adjusted for covariates in model 3 plus foot ulcers.
Per 10 units increase for HbA1c CV.
The current guidelines from the American Diabetes Association recommend HbA1c <7% (53 mmol/mol) as the treatment goal for patients with diabetes to prevent diabetes complications and that HbA1c tests should be performed approximately every 3 months in all patients. Poor glucose control can lead to a higher risk of LEA. However, the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial did not confirm the beneficial effect of intensive glycemic treatment compared with the standard therapy. Moreover, HbA1c does not reflect glucose fluctuations over a long period, which may have better ability to predict diabetes complications. Oscillating glucose can have more deleterious effects on endothelial function and oxidative stress than constantly high glucose exposure. The rigorous inclusion and exclusion criteria of the current study were similar to several post hoc analyses of clinical trials, which would enhance the accuracy of our analysis to a large extent. To our knowledge, the study is the first to assess the association between HbA1c variability, defined as HVS, HbA1c SD, and HbA1c CV, and LEA risk using electronic record data to generate real-world evidence.
In conclusion, we found long-term glycemic fluctuation was an independent indicator of LEA risk among patients with type 2 diabetes. Our findings indicated that HbA1c variability could be considered as a supplementary glycemic control target in preventing LEA among patients with type 2 diabetes.
Y.F. and Y.S. contributed equally to the work.
Article Information
Acknowledgments. The LEAD study would like to acknowledge the contributions of its partners. The success of this study depended on their ongoing support and expertise. These partners include Ochsner Health System and the Ochsner Patient Research Advisory Board, Tulane Medical Center, Research Action for Health Network (REACHnet, a PCORnet Clinical Research Network) and their multistakeholder Diabetes Advisory Groups, Pennington Biomedical Research Center, and our patient and community partners Patricia Dominick (Metairie, Louisiana), Catherine Glover (Baton Rouge, Louisiana), and Peggy Malone (New Orleans, Louisiana).
Funding. This work was supported by a Patient-Centered Outcomes Research Institute (PCORI) cooperative agreement (NEN-1508-32257) as part of Natural Experiments for Translation in Diabetes 2.0 (NEXT-D2) and a grant from the National Institute of General Medical Sciences (U54GM104940) of the National Institutes of Health.
All statements in this manuscript, including findings and conclusions, are solely those of the authors and do not necessarily represent the views of PCORI, its Board of Governors, or Methodology Committee.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. Y.F., Y.S., and J.Z. cleaned the data. Y.F. and Y.S. performed statistical analysis. Y.F., Y.S., J.Z., and G.H. wrote the manuscript. L.S., E.N., P.T.K., E.G.P.-H., R.H., S.C., and A.N.B. reviewed and edited the manuscript. All authors read and approved the final manuscript. G.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Data Availability. Restrictions apply to the availability of data generated or analyzed during this study to preserve patient confidentiality or because they were used under license. The corresponding author will, on request, detail the restrictions and any conditions under which access to some data may be provided.
