Correspondence to Dr Paul R Conlin; [email protected]
WHAT IS ALREADY KNOWN ON THIS TOPIC
Older adults with diabetes should have hemoglobin A1c (A1c) treatment targets that reflect their comorbidities, complications, and life expectancy.
A1c variability is also a risk factor for adverse outcomes.
It is unclear if A1c stability over time within individualized target ranges with upper and lower bounds is a risk predictor for adverse outcomes.
WHAT THIS STUDY ADDS
In this observational study, older veterans with diabetes had lower risk of mortality when A1c levels during a 3-year baseline period were mostly within individualized target ranges.
Patients with A1c levels mostly above or below target ranges had increased risk of mortality and macrovascular complications.
Greater A1c time below range was associated with somewhat lower risk of microvascular complications.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Maintaining A1c stability over time within patient-specific target ranges is associated with reduced risk of mortality and cardiovascular outcomes in older adults with diabetes.
Introduction
The relationship between chronic hyperglycemia and diabetes complications is complex. Several lines of evidence suggest that both the level of hemoglobin A1c (A1c) and its variability over time have clinical implications. Targeting A1c to <7.0% in type 2 diabetes reduces the risk of microvascular complications but does not impact cardiovascular disease (CVD) or mortality,1–3 and there is a non-linear association between average A1c and mortality.4–6 Increased A1c variability is also associated with higher risk of CVD and mortality independent of A1c levels.7–9
Many guidelines for treating older adults with diabetes recommend adjusting A1c goals based on life expectancy, comorbidities, and complications.10–12 In particular, the 2017 VA Department of Defense clinical practice guideline identifies specific A1c target ranges based on these clinical factors.11 However, clinicians and patients often focus on avoiding A1c upper limits without considering the risks and benefits of lower A1c levels. In addition, treatment decisions may be based on the most recent A1c level with less attention paid to its stability over time. This leads to many older adults being potentially overtreated,13 exposing them to polypharmacy, hypoglycemia risk, and increased A1c variability.
We developed a clinical measure that incorporates A1c levels and their stability over time, A1c time in range (TIR). A1c TIR represents the percentage of time that A1c levels fall within guideline-directed target ranges11 over a 3-year period. We showed that higher A1c TIR is associated with lower risk of mortality and macrovascular and microvascular complications independent of A1c mean and SD.14 15 However, patients with low A1c TIR may have patterns composed of varying times when A1c levels are mostly above, below, or fluctuating around their target ranges. Therefore, we evaluated the risks of mortality and macrovascular and microvascular complications among older adults with diabetes based on having A1c levels mostly within, above, or below their patient-specific target ranges.
Research design and methods
Study population
We used administrative data from VA and Medicare from 2004 to 2016. The sample included VA–Medicare dual enrollees who were ≥65 years old to ensure access to more complete diagnosis codes for risk adjustment and outcome measures. A diabetes diagnosis, including both type 1 and type 2 diabetes, was established in patients with at least two outpatient or one inpatient diagnosis code (ie, Internationational Classification of Diseases [ICD]-9 and ICD-10) or a prescription for diabetes medications.16 We further filtered the sample by requiring patients to have at least four A1c tests during a 3-year baseline period that started between January 1, 2005, and December 31, 2012. This latter criterion provided the means to calculate patient-level A1c TIR within the corresponding baseline period. The overall study sample included 397 634 patients (online supplemental appendix figure 1).
Main independent variables
Target range categories and A1c TIR
Individuals were assigned to A1c target ranges based on the VA Department of Defense Diabetes Clinical Practice Guideline.11 There are four different A1c ranges (6.0%–7.0%, 7.0%–8.0%, 7.5%–8.5%, and 8.0%–9.0%) which are based on life expectancy and the presence of comorbidities and diabetes complications. Higher ranges are recommended for patients with lower life expectancy and/or moderate or advanced diabetes complications. To assess life expectancy, we used a prediction model that categorized patients into groups with mortality risks of either <5, 5–10, or ≥10 years using clinical and administrative data.17 Diabetes complications were measured using the Diabetes Complications Severity Index (DCSI).18 This index identifies major diabetes complications (ie, retinopathy, nephropathy, neuropathy, cardiovascular, cerebrovascular, and peripheral vascular) using diagnosis codes and laboratory data. We used DCSI scores of 0–1, 2–3, and ≥4 to represent absent or mild, moderate, or advanced diabetes complications, respectively.
A 1-year period preceding the 3-year baseline was used to assess predictors of mortality and diabetes complications and to set the initial A1c target range. These factors were updated annually during the baseline, and A1c target ranges were adjusted each year based on new diagnoses.
Patient-level A1c TIR was calculated as the percentage of days during baseline that a patient’s A1c was in the unique target range using A1c levels and linear interpolation between levels and test dates (online supplemental appendix figure 2). Time below range (TBR) and time above range (TAR) represented the percentage of days that A1c was either below or above the assigned range, respectively. For each patient, A1c TIR, TBR, and TAR summed to 100%.
To identify an association between A1c TIR, TBR, and TAR and adverse outcomes, we compared patients who were grouped into four mutually exclusive categories: ≥60% A1c TIR, ≥60% A1c TBR, ≥60% A1c TAR, and a mixed group composed of patients whose TIR, TBR, and TAR were all below 60%. A threshold level of ≥60% was used because this is approximately the upper quartile for A1c TIR in the sample.
Covariates
Baseline covariate variables included several patient-level characteristics including age, race/ethnicity (self-reported), sex (self-reported), marital status, and eligibility constraints for VA services (ie, copayment requirements and service-connected disabilities). Time trends and facility factors were controlled with indicator variables for the calendar quarter patients entered the outcome period and the VA Medical Center, where the patient received most of their care. The Elixhauser Comorbidity Index19 was used to identify major comorbidities during the baseline period, and each comorbidity was recorded as a binary covariate. Diabetes complications and severity were measured by the DCSI and recorded as the highest score during baseline. Diabetes medications included insulin (categorized as any use, whether prandial, basal, or a combination), metformin, sulfonylureas, thiazolidinediones, and all others used for diabetes treatment during baseline. Medication adherence was expressed as a dichotomous measure based on the proportion of days covered and whether it was ≥80% for all prescribed diabetes medications. Laboratory measures included serum creatinine, serum albumin, urine albumin-to-creatinine ratio, and blood lipids (high-density lipoprotein, low-density lipoprotein (LDL), and triglycerides). Clinical measures included body mass index and blood pressure. Each was categorized using clinical criteria (eg, low, normal, and high), and a separate category was used for missing measures. We also included the number of A1c tests during baseline and A1c SD.
Patients were associated with a treating clinician based on the individual who ordered the most A1c tests during baseline. Process quality variables were computed at the clinician level during baseline and were used to account for other factors that may impact a patient’s diabetes care. These included the percent of each clinician’s patients with diabetes with A1c of >9%, LDL cholesterol of >100 mg/day, and blood pressure of >140/90 mm Hg.20 We also included clinician type (eg, physician, nurse practitioner/physician assistant, or other) and whether they were a primary care clinician.
Outcomes
Outcomes included mortality and incident diabetes complications from the DCSI. The latter was grouped into macrovascular and microvascular composites. The macrovascular composite included cardiovascular, cerebrovascular, and peripheral vascular diseases, and the microvascular composite included retinopathy, neuropathy, and nephropathy complications. For the composite measures, patients were excluded if they had an existing complication for any of the three underlying measures during baseline. For the individual complications, patients were excluded if they had that specific complication during baseline. The VA Vital Status File was used to determine all-cause mortality.21
Analyses
We performed descriptive analyses and report the numbers and proportions for selected demographic characteristics and key categorical variables. We examined if predictor variables differed by A1c categories. Cox proportional hazards models were used to estimate the effects of ≥60% A1c TIR, ≥60% TBR, ≥60% TAR, and the mixed group on mortality, macrovascular, and microvascular complications, and the individual components of each composite. In the mortality model, individuals were censored on death or at the end of the study period. In the diabetes complications models, individuals were censored on death, on experiencing an incident complication, or at the end of the study period. All models included ≥60% A1c TIR (reference group), ≥60% TBR, ≥60% TAR, and the mixed group as the main explanatory variables, as well as all covariates.
We conducted several additional analyses. We examined a higher threshold (≥80%) for A1c TIR, TBR, TAR, and the mixed group, and also tested results using a shorter outcome period of 24 months. Since A1c TIR, TBR, and TAR are defined by A1c levels within, below, or above specific ranges, respectively, each category was highly correlated with baseline A1c levels. To address a potential role of A1c levels, we performed separate analyses in which the four A1c target ranges were included as covariates. We tested findings when A1c SD was removed from study models to assess if findings were independent of A1c variability. Patients using insulin may comprise a group associated with higher risk, so we performed analyses among insulin users and non-insulin users. Finally, because individuals are at risk for both diabetes complications and mortality at the same time, we used Fine and Gray’s competing risk Cox proportional hazard model for the macrovascular and microvascular outcomes.
Role of the funding sources
The funders had no influence over the study design, conduct, or reporting of results.
Results
The study sample included 397 634 patients who had a mean age of 76.9 years (SD 5.7) and were predominantly white (86%) and male (99%) (table 1). The mean baseline A1c was 7.0% (SD 1.0), and 26% of the patients had DCSI scores of >5. During an average of 5.5 years (SD 2.4) of follow-up, 28% of the patients died, and a similar number experienced a new diabetes complication (13.4% microvascular and 14.4% macrovascular). The percentages of patients assigned to each of the four A1c target ranges at the end of baseline were A1c 6.0%–7.0% (18%), A1c 7.0%–8.0% (24%), A1c 7.5%–8.5% (45%), and A1c 8.0%–9.0% (13%). Patients were then grouped into four unique categories based on the proportion of days during baseline in which A1c was within, above, or below their target range. Percentages of patients in each category were ≥60% A1c TIR (19%), ≥60% A1c TBR (50%), ≥60% A1c TAR (12%), and mixed group (19%). The actual time spent in each respective category was ≥80% for TIR, TAR, and TBR, while the mixed group had smaller amounts of time within, below, and above range (online supplemental appendix figure 3).
Table 1Baseline characteristics of the study population (N=397 634)
| All participants (N=397 634) | Time in range ≥60% (n=76 760) | Time below range ≥60% (n=197 810) | Time above range ≥60% (n=47 448) | Mixed group (n=75 616) | |
| Demographics | |||||
| Age at end of baseline (years) | 76.9 (5.7) | 75.1 (4.7) | 78.6 (6.0) | 74.3 (4.4) | 76.0 (5.2) |
| Sex, n (%) | |||||
| Male | 392 643 (98.7) | 75 718 (98.6) | 195 267 (98.7) | 46 889 (98.8) | 74 769 (98.9) |
| Female | 4991 (1.3) | 1042 (1.4) | 2543 (1.3) | 559 (1.2) | 847 (1.1) |
| Race/ethnicity, n (%) | |||||
| Asian | 1422 (0.4) | 300 (0.4) | 715 (0.4) | 163 (0.3) | 244 (0.3) |
| Black | 42 476 (10.7) | 7677 (10.0) | 18 844 (9.6) | 7,006 (14.8) | 8949 (11.8) |
| Hispanic | 6119 (1.5) | 872 (1.1) | 2415 (1.2) | 1325 (2.8) | 1507 (2.0) |
| Other | 4416 (1.1) | 963 (1.3) | 1784 (0.9) | 778 (1.7) | 891 (1.2) |
| White | 343 201 (86.3) | 66 948 (87.2) | 174 052 (88.0) | 38 176 (80.5) | 64 025 (84.7) |
| Clinical parameters* | |||||
| Baseline A1c (%) | 7.0 (1.0) | 7.0 (0.7) | 6.4 (0.5) | 8.6 (1.1) | 7.5 (0.7) |
| Number of A1c tests, median (IQR) | 6 (5–7) | 6 (5–7) | 6 (6–7) | 7 (5–9) | 6 (5–8) |
| Body mass index (kg/m2) | 30.2 (5.2) | 30.5 (5.0) | 29.8 (5.2) | 30.7 (5.3) | 30.8 (5.4) |
| Diabetes Complications Severity Index (highest baseline score), n (%) | |||||
| 0 | 28 762 (7.2) | 13 195 (17.2) | 4843 (2.5) | 6473 (13.7) | 4251 (5.7) |
| 1–2 | 102 031 (25.7) | 30 740 (40.1) | 33 330 (16.9) | 18 990 (40.0) | 18 971 (25.1) |
| 3–5 | 161 987 (40.7) | 22 731 (29.6) | 93 838 (46.9) | 14 490 (30.5) | 31 928 (42.2) |
| 6–8 | 86 636 (21.8) | 8311 (10.8) | 55 847 (28.2) | 5979 (12.6) | 16 499 (21.8) |
| ≥9 | 18 218 (4.6) | 1783 (2.3) | 10 952 (5.5) | 1516 (3.2) | 3967 (5.3) |
| Cardiovascular comorbidities, n (%)† | |||||
| Cardiovascular | 274 554 (70.0) | 39 466 (51.41) | 158 849 (80.3) | 23 784 (50.1) | 52 455 (69.4) |
| Cerebrovascular | 120 400 (30.3) | 13 450 (17.5) | 76 962 (38.9) | 8143 (17.2) | 21 845 (28.9) |
| Congestive heart failure | 119 723 (30.1) | 12 742 (16.6) | 74 873 (37.8) | 8697 (18.3) | 23 411 (31.0) |
| Hypertension | 381 357 (4.1) | 72 051 (93.9) | 191 819 (97.0) | 44 278 (94.3) | 72 759 (96.2) |
| Medications, n (%)‡ | |||||
| Sulfonylureas | 213 116 (53.6) | 37 284 (48.6%) | 94 946 (48.0%) | 32 049 (67.6) | 48 837 (64.6) |
| Metformin | 196 329 (49.4) | 43 891 (57.2) | 76 536 (38.7) | 32 220 (67.9) | 43 682 (57.8) |
| Insulin | 97 183 (24.4) | 15 272 (19.9) | 33 025 (16.7) | 21 776 (45.9) | 27 110 (35.9) |
| Thiazolidinediones | 63 084 (15.9) | 11 008 (14.3) | 23 247 (11.8) | 12 241 (25.8) | 16 588 (21.9) |
| Alpha-glucosidase inhibitors | 7672 (1.9) | 1380 (1.8) | 2286 (1.2) | 1775 (3.7) | 2231 (3.0) |
| Other | 6370 (1.6) | 1162 (1.5) | 2588 (1.3) | 1123 (2.4) | 1497 (2.0) |
| Adherence to diabetes medications, n (%) | |||||
| Proportion of days covered ≥80% | 226 599 (57.0) | 46 182 (60.2) | 99 521 (50.3) | 31 245 (65.9) | 49 651 (65.6) |
Other medications: amylin analog; bile acid sequestrant, dipeptidyl peptidase inhibitor, dopamine receptor agonist, glucagon-like peptide agonist, meglitinides, sodium–glucose cotransporter inhibitors.
*Mean (SD) or median (IQR).
†Comorbidities observed in the baseline period.
‡Medications may sum up to >100% due to combination usage.
A1c, hemoglobin A1c.
Descriptive statistics of demographics, clinical parameters, DCSI categories, and major comorbidities (table 1) had statistically significant differences across the four A1c categories for each patient-level profile (p<0.001). A1c levels also differed among the categories. Those in the ≥60% A1c TBR category and the mixed group were older, had higher DCSI scores, and had greater prevalence of cardiovascular comorbidities. The ≥60% A1c TAR category included more black and Hispanic patients, and the mixed group had greater use of insulin and sulfonylureas.
≥60% A1c TBR and adverse outcomes
Higher A1c TBR was associated with increased risk of mortality (HR 1.12, 95% CI 1.10 to 1.13) and macrovascular complications (HR 1.04, 95% CI 1.01 to 1.06) when compared with ≥60% A1c TIR (table 2). Within the macrovascular composite, ≥60% A1c TBR was associated with incident CVD (HR 1.05, 95% CI 1.02 to 1.07) and cerebrovascular disease (HR 1.03, 95% CI 1.01 to 1.05). For the microvascular composite, ≥60% A1c TBR was associated with lower risk (HR 0.97, 95% CI 0.95 to 1.00), with retinopathy (HR 0.92, 95% CI 0.90 to 0.94) and neuropathy (HR 0.95, 95% CI 0.93 to 0.97) carrying lower HRs, and nephropathy was unaffected (HR 1.00, 95% CI 0.98 to 1.02).
Table 2Adjusted HRs (95% CI) predicting mortality and incident diabetes complications by≥60% A1c time above or below range categories
| ≥60% Time below range* | ≥60% Time above range* | Mixed group* | |
| Mortality (n=397 634) | 1.12 (1.11 to 1.14) | 1.10 (1.08 to 1.12) | 1.06 (1.04 to 1.07) |
| Macrovascular Composite (n=89 625) | 1.04 (1.01 to 1.06) | 1.06 (1.03 to 1.09) | 1.02 (0.99 to 1.04) |
| Cardiovascular (n=122 135) | 1.06 (1.03 to 1.08) | 1.08 (1.06 to 1.11) | 1.03 (1.01 to 1.05) |
| Cerebrovascular (n=277 234) | 1.03 (1.01 to 1.05) | 1.09 (1.07 to 1.12) | 1.04 (1.02 to 1.07) |
| Peripheral vascular (n=228 583) | 1.12 (1.09 to 1.03) | 1.12 (1.09 to 1.15) | 1.03 (1.01 to 1.06) |
| Microvascular Composite (n=74 016) | 0.97 (0.95 to 1.00) | 1.11 (1.08 to 1.14) | 1.00 (0.97 to 1.03) |
| Retinopathy (n=235 580) | 0.92 (0.90 to 0.94) | 1.21 (1.18 to 1.25) | 1.02 (1.00 to 1.05) |
| Neuropathy (n=222 274) | 0.95 (0.93 to 0.97) | 1.09 (1.06 to 1.11) | 1.01 (0.99 to 1.03) |
| Nephropathy (n=168 616) | 1.00 (0.98 to 1.02) | 1.11 (1.08 to 1.13) | 1.11 (1.08 to 1.13) |
Models included all covariates.
*Referent: ≥60% A1c time in range.
A1c, hemoglobin A1c.
≥60% A1c TAR and adverse outcomes
Higher A1c TAR was associated with increased risk of mortality (HR 1.10, 95% CI 1.08 to 1.12) and macrovascular (HR 1.06, 95% CI 1.03 to 1.09) and microvascular complications (HR 1.11, 95% CI 1.08 to 1.14) compared with ≥60% A1c TIR (table 2). In addition, ≥60% A1c TAR was associated with increased risk of each component of the macrovascular (ie, cardiovascular, cerebrovascular, and peripheral vascular diseases) and microvascular (ie, retinopathy, neuropathy, and nephropathy) composites.
Mixed group of A1c TIR, TBR, and TAR and adverse outcomes
Patients with a combination of A1c TIR, TBR, and TAR, with all being <60%, had increased risk of mortality (HR 1.05, 95% CI 1.03 to 1.07) when compared with ≥60% A1c TIR. The mixed group was not associated with increased risk of the macrovascular or microvascular composites but was associated with some of the individual components.
A plot of HRs and 95% CIs for the main outcomes (mortality, macrovascular, and microvascular composites) and key predictor variables showed that A1c SD, insulin, and sulfonylureas were each associated with increased risks of mortality and macrovascular and microvascular complications; female sex carried lower risk of each (figure 1). Complete results for mortality outcomes are presented in online supplemental appendix table 1.
Enlarge this image.Figure 1. Adjusted proportional hazard models predicting outcomes by A1c categories and major covariates. A1c, hemoglobin A1c.
Additional analyses
We assessed outcomes using a higher threshold (≥80%) for A1c TIR, TBR, and TAR. Risks were greater for each outcome when compared with the ≥60% threshold (table 3).
Table 3Adjusted HR (95% CI) predicting mortality and diabetes complications by ≥80% A1c time above or below range categories
| Time below range ≥80%* | Time above range ≥80%* | Mixed group* | |
| Mortality (n=397 634) | 1.20 (1.17 to 1.22) | 1.18 (1.15 to 1.21) | 1.18 (1.15 to 1.21) |
| Macrovascular complications (n=89 625) | 1.08 (1.05 to 1.11) | 1.10 (1.06 to 1.13) | 1.04 (1.01 to 1.16) |
| Cardiovascular (n=122 135) | 1.08 (1.05 to 1.11) | 1.10 (1.05 to 1.13) | 1.02 (1.00 to 1.15) |
| Cerebrovascular (n=277 234) | 1.06 (1.03 to 1.09) | 1.14 (1.10 to 1.18) | 1.06 (1.03 to 1.08) |
| Peripheral vascular (n=228 583) | 1.07 (1.04 to 1.10) | 1.20 (1.16 to 1.25) | 1.08 (1.05 to 1.11) |
| Microvascular complications (n=74 016) | 1.00 (0.97 to 1.03) | 1.14 (1.10 to 1.18) | 1.02 (1.00 to 1.05) |
| Retinopathy (n=235 580) | 0.93 (0.90 to 0.96) | 1.29 (1.24 to 1.34) | 1.03 (1.01 to 1.06) |
| Neuropathy (n=222 274) | 0.97 (0.95 to 0.99) | 1.16 (1.13 to 1.20) | 1.05 (1.03 to 1.07) |
| Nephropathy (n=168 616) | 1.03 (1.00 to 1.05) | 1.13 (1.10 to 1.17) | 1.03 (1.01 to 1.06) |
Models included all covariates.
*Referent: ≥80% A1c time in range.
A1c, hemoglobin A1c.
We explored the incidence of adverse outcomes over a shorter follow-up period of 24 months. Both ≥60% and ≥80% A1c TBR categories were associated with increased risk of mortality and macrovascular complications, and lower risk of retinopathy and neuropathy. Similarly, the ≥60% and ≥80% A1c TAR categories were associated with mortality, macrovascular, and microvascular complications (online supplemental appendix tables 2 and 3). We also performed analyses stratified by the four A1c target ranges with the lowest range group as the referent. Mortality risks were similar to the main results (online supplemental appendix table 4). Study models that excluded A1c SD showed results that paralleled the main results with slightly higher HRs (online supplemental appendix table 5). In analyses that stratified patients who were insulin users and non-insulin users, again mortality risks were similar in both groups (online supplemental appendix table 6). In a model that predicted macrovascular and microvascular complications with the competing risk of mortality, patients with ≥60% A1c TBR had increased risk of macrovascular complications and lower risk of microvascular complications, whereas those with ≥60% A1c TAR had increased risk of both outcomes (online supplemental appendix table 7).
Conclusions
We showed that in older adults with diabetes, A1c stability over time within individualized target ranges is associated with a lower risk of mortality and macrovascular complications. Higher A1c TBR had increased risk of mortality and macrovascular complications and lower risk of some microvascular complications. Increased A1c TAR was associated with higher risks of all the adverse outcomes. HRs were small but significant, and results were similar after controlling for several covariates and across additional tests that included higher A1c TIR thresholds, a shorter outcome period, and with mortality as a competing risk. These findings suggest that A1c stability over time within patient-specific ranges confers somewhat lower risk of adverse outcomes, and persistently out-of-range A1c levels are associated with distinct risks.
Clinical practice guidelines recommend individualized A1c goals for older adults with diabetes to balance risks and benefits and to limit hyperglycemia and hypoglycemia. Higher A1c levels are recommended for individuals with a greater burden of comorbidities or shorter life expectancy.10–12 However, there is no guidance about the risks associated with lower A1c levels or when A1c levels are unstable over time. This knowledge gap creates practice implications for our results. First, our findings suggest that A1c patterns over time may be relevant in glycemic management. Among older adults with diabetes, stable A1c within specific ranges is associated with lower risk or diabetes complications and mortality. This aligns with other findings that greater A1c variability increases risk of mortality and diabetes-related complications.7–9 22 Our results build on this construct by using patient-specific A1c ranges and additionally show the unique risks incurred when A1c levels over time deviate from these ranges in different directions. Our study data included a period when earlier guidelines may have influenced clinicians to target different A1c levels. Nonetheless, our analyses applied individualized target ranges with upper and lower bounds11 and confirmed that categorizing A1c levels identifies patients with unique risks based on their time and pattern of A1c stability over time. Second, greater A1c TBR is associated with mortality and macrovascular complications but has a somewhat lower risk of microvascular complications. These findings were similar in insulin users and non-insulin users. It is unclear if the risks of mortality and CVD relate to the occurrence of hypoglycemia or other unmeasured factors, so this warrants further study. Given the mixture of risks and benefits, these observations support the need to engage older adults in shared decision making when setting and sustaining lower A1c goals. Third, the TIR concept has been applied to intraday glucose variability, and some data suggest benefits with glycemic stability on short-term outcomes (eg, A1c and hypoglycemia).23 24 However, most older adults with diabetes still have glycemic goals tracked with less intensive monitoring such as periodic A1c levels. A1c TIR is calculated from data included in electronic health records, so it is possible to generate such a measure for clinicians and patients at the point of care. Thus, A1c TIR may be a practical measure to assess A1c stability over time.
A1c TIR integrates both A1c levels and their stability over time within patient-specific ranges that include upper and lower limits. This has relevance, given that our findings show a U-shaped relationship between A1c TIR, mortality, and macrovascular disease outcomes. A similar relationship between mean A1c and mortality was noted in the control group in the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial,25 and others have suggested a non-linear association between A1c levels and mortality but often at the extremes of A1c values (A1c <6.5% or A1c >10%).4–6 Our results differ in that we showed this relationship spanned a range of clinically relevant A1c levels, with target ranges varying between 6%–7% and 8%–9%. We also observed a continuous and increasing risk for microvascular complications, as was evident by the contrasts between A1c TBR, TIR, and TAR categories. These latter findings are consistent with clinical trials in younger patients showing a graded relationship between A1c levels and risk of microvascular complications.2 26 Older adults with diabetes appear to carry these same risks.
The associations of A1c TIR with diabetes complications and mortality are small but consistent, significant, and aligned with findings from intensive glucose management in ACCORD, Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation (ADVANCE), and the Veterans Affairs Diabetes Trial (VADT). Although these clinical trials showed no significant effects on CVD or mortality, there were some favorable effects on microvascular complications. In ACCORD, intensive treatment was associated with somewhat lower microvascular complications (HR 0.95, 95% CI 0.89 to 1.01),27 and in ADVANCE, the microvascular composite was reduced (HR 0.86, 95% CI 0.77 to 0.97), mostly due to fewer kidney events.2 A meta-analysis of ACCORD, ADVANCE, United Kingdom Prospective Diabetes Study (UKPDS) and VADT showed significant effects with intensive glucose management on kidney outcomes (HR 0.80, 95% CI 0.72 to 0.88) and retina complications (HR 0.87, 95% CI 0.76 to 1.00) but not neuropathy (HR 0.98, 95% CI 0.87 to 1.09).28 Our findings suggest that A1c TIR as a risk predictor is similar in magnitude to intensive glucose-lowering interventions but add new information about risks of CVD and mortality.
Our study has several strengths. We included a large nationwide sample of older veterans with diabetes and used VA and Medicare data to comprehensively assess diagnoses and outcomes. A1c TIR was calculated over a 3-year baseline, and outcome measures were assessed during a follow-up period to minimize risk of reverse causation. We controlled for patient-level demographics and many clinical characteristics that could be confounders. Nonetheless, the study has several limitations. The sample included older adults who were mostly male, which limits the generalizability of the findings. We studied veterans, who tend to have higher comorbidities than non-veterans,29 30 and their reliance on VA healthcare may influence coding of complications or differences in outcomes by treatment setting.31 A1c TIR as a clinically derived construct was calculated by linear interpolation between A1c tests obtained in usual care, which may introduce some measurement error. In addition, the requirement for four or more A1c tests during the baseline period may lead to selection bias since patients with fewer tests were not included. This is an observational study and there are some unmeasured factors that may affect A1c stability over time that are not coded in electronic health records, such as diabetes duration, food insecurity, engagement with self-care, and financial and social support. These factors may confound the observed associations. The study was conducted during a period when newer diabetes medications, such as sodium–glucose cotransporter-2 inhibitors or glucagon-like peptide-1 receptor agonists, were not available or were less often prescribed. These medications are associated with reduced cardiovascular and renal outcomes and less hypoglycemia. We did not have a sufficient number of patients using these medications to assess whether the risks associated with low A1c TIR are mitigated, leaving such explorations for future research. Finally, we cannot assert that prospectively maintaining higher A1c TIR over time confers protection against diabetes complications or mortality.
In summary, integrating A1c levels and measuring their stability within individualized target ranges over time is a predictor of risk of major adverse events. Higher A1c TIR is associated with somewhat lower risk of mortality and macrovascular complications, whereas higher TBR and TAR are associated with increased risk of both. Assessing A1c levels and their stability over time may help identify older adults with diabetes who are at increased risk of adverse outcomes.
Data availability statement
Data may be obtained from a third party and are not publicly available. The data that support the findings of this study are available from the VA and the Center for Medicare and Medicaid Services, but restrictions apply to the availability of these data, which were used under license for the current study and therefore are not publicly available. Data are, however, available from the authors upon reasonable request and with permission of the VA and the Center for Medicare and Medicaid Services.
Ethics statements
Patient consent for publication
Not applicable.
Ethics approval
This study involved human participants and was reviewed and approved by the institutional review board at the Department of Veterans Affairs (VA) Boston Healthcare System. The study was classified as exempt from most requirements as it posed no or minimal risk to participants and was granted a waiver from consent requirements by the human studies subcommittee of VA Boston Healthcare System (IRB-1577893-5).
Contributors All authors made substantial contributions to the intellectual content of the paper, including the design of the study (PRC, JCP, and DCM), acquisition of data (LZ and DL), statistical analyses (PRC, JCP, DCM, LZ, and RN), interpretation of the data (all authors), and the drafting and critical revision of the manuscript (all authors). PRC 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.
Funding This study was supported by the Department of Veterans Affairs, Veterans Health Administration, Office of Research and Development, Health Services Research and Development (IIR 15-116) and the National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases (R01 DK114098). Data were obtained with support from VA Information Resource Center, VA/CMS Data for Research Projects SDR 02-23 and 98-004.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.
1 Duckworth W, Abraira C, Moritz T, et al. Glucose control and vascular complications in veterans with type 2 diabetes. N Engl J Med 2009; 360: 129–39. doi:10.1056/NEJMoa0808431
2 ADVANCE Collaborative Group, Patel A, MacMahon S, et al. Intensive blood glucose control and vascular outcomes in patients with type 2 diabetes. N Engl J Med 2008; 358: 2560–72. doi:10.1056/NEJMoa0802987
3 Gerstein HC, Miller ME, Byington RP, et al. Effects of intensive glucose lowering in type 2 diabetes. N Engl J Med 2008; 358: 2545–59. doi:10.1056/NEJMoa0802743
4 Currie CJ, Peters JR, Tynan A, et al. Survival as a function of HbA (1C) in people with type 2 diabetes: a retrospective cohort study. Lancet 2010; 375: 481–9. doi:10.1016/S0140-6736(09)61969-3
5 Huang ES, Liu JY, Moffet HH, et al. Glycemic control, complications, and death in older diabetic patients: the diabetes and aging study. Diabetes Care 2011; 34: 1329–36. doi:10.2337/dc10-2377
6 Twito O, Ahron E, Jaffe A, et al. New-Onset diabetes in elderly subjects: association between HbA1c levels, mortality, and coronary revascularization. Diabetes Care 2013; 36: 3425–9. doi:10.2337/dc12-2503
7 Forbes A, Murrells T, Mulnier H, et al. Mean hba1c, hba1c variability, and mortality in people with diabetes aged 70 years and older: a retrospective cohort study. Lancet Diabetes Endocrinol 2018; 6: 476–86. doi:10.1016/S2213-8587(18)30048-2
8 Prentice JC, Pizer SD, Conlin PR. Identifying the independent effect of HbA1c variability on adverse health outcomes in patients with type 2 diabetes. Diabet Med 2016; 33: 1640–8. doi:10.1111/dme.13166
9 Critchley JA, Carey IM, Harris T, et al. 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: 2237–46. doi:10.2337/dc19-0848
10 Committee A, Draznin B, Aroda VR. 13. older adults: standards of medical care in diabetes-2022. Diabetes Care 2022; 45: S195–207. doi:10.2337/dc22-S013
11 Conlin PR, Colburn J, Aron D, et al. Synopsis of the 2017 U.S. department of Veterans affairs/U.S. Department of defense clinical practice guideline: management of type 2 diabetes mellitus. Ann Intern Med 2017; 167: 655–63. doi:10.7326/M17-1362
12 Qaseem A, Wilt TJ, Kansagara D, et al. Hemoglobin A1c targets for glycemic control with pharmacologic therapy for nonpregnant adults with type 2 diabetes mellitus: a guidance statement update from the American College of physicians. Ann Intern Med 2018; 168: 569–76. doi:10.7326/M17-0939
13 Tseng CL, Soroka O, Maney M, et al. Assessing potential glycemic overtreatment in persons at hypoglycemic risk. JAMA Intern Med 2014; 174: 259–68. doi:10.1001/jamainternmed.2013.12963
14 Prentice JC, Mohr DC, Zhang L, et al. Increased hemoglobin A1c time in range reduces adverse health outcomes in older adults with diabetes. Diabetes Care 2021; 44: 1750–6. doi:10.2337/dc21-0292
15 Mohr DC, Zhang L, Prentice JC, et al. Association of hemoglobin A1c time in range with risk for diabetes complications. BMJ Open Diabetes Res Care 2022; 10: e002738. doi:10.1136/bmjdrc-2021-002738
16 Miller DR, Safford MM, Pogach LM. Who has diabetes? Best estimates of diabetes prevalence in the Department of Veterans Affairs based on computerized patient data. Diabetes Care 2004; 27 Suppl 2: B10–21. doi:10.2337/diacare.27.suppl_2.b10
17 Griffith KN, Prentice JC, Mohr DC, et al. Predicting 5- and 10-year mortality risk in older adults with diabetes. Diabetes Care 2020; 43: 1724–31. doi:10.2337/dc19-1870
18 Glasheen WP, Renda A, Dong Y. Diabetes complications severity index (DCSI)-update and ICD-10 translation. J Diabetes Complications 2017; 31: 1007–13. doi:10.1016/j.jdiacomp.2017.02.018
19 Elixhauser A, Steiner C, Harris DR, et al. Comorbidity measures for use with administrative data. Med Care 1998; 36: 8–27. doi:10.1097/00005650-199801000-00004
20 Prentice JC, Conlin PR, Gellad WF, et al. Capitalizing on prescribing pattern variation to compare medications for type 2 diabetes. Value Health 2014; 17: 854–62. doi:10.1016/j.jval.2014.08.2674
21 Arnold SM, Maynard C, Hynes DM. VIReC technical report 2: VA-NDI mortality data merge project. Hines, IL: Edward Hines, Jr. VA Hospital, 2006.
22 Laiteerapong N, Karter AJ, Moffet HH, et al. Ten-year hemoglobin A1c trajectories and outcomes in type 2 diabetes mellitus: the diabetes & aging study. J Diabetes Complications 2017; 31: 94–100. doi:10.1016/j.jdiacomp.2016.07.023
23 Martens T, Beck RW, Bailey R, et al. Effect of continuous glucose monitoring on glycemic control in patients with type 2 diabetes treated with basal insulin: a randomized clinical trial. JAMA 2021; 325: 2262–72. doi:10.1001/jama.2021.7444
24 Karter AJ, Parker MM, Moffet HH, et al. Association of real-time continuous glucose monitoring with glycemic control and acute metabolic events among patients with insulin-treated diabetes. JAMA 2021; 325: 2273–84. doi:10.1001/jama.2021.6530
25 Riddle MC, Ambrosius WT, Brillon DJ, et al. Epidemiologic relationships between A1c and all-cause mortality during a median 3.4-year follow-up of glycemic treatment in the Accord trial. Diabetes Care 2010; 33: 983–90. doi:10.2337/dc09-1278
26 Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK prospective diabetes study (UKPDS) group. Lancet 1998; 352: 837–53.
27 Ismail-Beigi F, Craven T, Banerji MA, et al. Effect of intensive treatment of hyperglycaemia on microvascular outcomes in type 2 diabetes: an analysis of the Accord randomised trial. Lancet 2010; 376: 419–30. doi:10.1016/S0140-6736(10)60576-4
28 Zoungas S, Arima H, Gerstein HC, et al. Effects of intensive glucose control on microvascular outcomes in patients with type 2 diabetes: a meta-analysis of individual participant data from randomised controlled trials. Lancet Diabetes Endocrinol 2017; 5: 431–7. doi:10.1016/S2213-8587(17)30104-3
29 Betancourt JA, Granados PS, Pacheco GJ, et al. Exploring health outcomes for U.S. veterans compared to non-veterans from 2003 to 2019. Healthcare (Basel) 2021; 9: 604. doi:10.3390/healthcare9050604
30 Bloeser K, Lipkowitz-Eaton J. Disproportionate multimorbidity among veterans in middle age. J Public Health (Oxf) 2022; 44: 28–35. doi:10.1093/pubmed/fdab149
31 Cashion W, Gellad WF, Sileanu FE, et al. Source of post-transplant care and mortality among kidney transplant recipients dually enrolled in Va and Medicare. Clin J Am Soc Nephrol 2021; 16: 437–45. doi:10.2215/CJN.10020620
© 2023 Author(s) (or their employer(s)) 2023. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ. http://creativecommons.org/licenses/by-nc/4.0/ This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/ . Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Details
; Zhang, Libin 2 ; Li, Donglin 2 ; Nelson, Richard E 3 ; Prentice, Julia C 4 ; Mohr, David C 5
1 Medical Service (111), VA Boston Health Care System West Roxbury Campus, West Roxbury, Massachusetts, USA; Harvard Medical School, Boston, MA, USA
2 VA Center for Healthcare Organization and Implementation Research Boston Campus, Boston, Massachusetts, USA
3 University of Utah Health, Salt Lake City, Utah, USA; VA Salt Lake City Healthcare System, Salt Lake City, UT, USA
4 Boston University School of Medicine, Boston, Massachusetts, USA; VA Boston Heatlhcare System, Boston, MA, USA
5 VA Center for Healthcare Organization and Implementation Research Boston Campus, Boston, Massachusetts, USA; Boston University School of Public Health, Boston, MA, USA