Variability in hemoglobin A1c predicts all-cause mortality in patients with type 2 diabetes.
ABSTRACT To evaluate the relationship between hemoglobin A1c variability and all-cause mortality in type 2 diabetic patients.
This was a retrospective cohort study in type 2 diabetic patients followed for at least 2 years between 2003 and 2009. A1C variability was determined from the standard deviation or coefficient of variation of serial A1C values (A1C(SD) or A1C(CV)). Subjects were categorized into either the high or low A1C variability group according to their A1C(CV) median. Hazard ratios (HRs) of various factors for all-cause mortality were determined from Cox's proportional hazard models.
A total of 881 subjects (422 men, 459 women) were included and 73 (8.3%) died during follow-up. The follow-up period was 4.7 ± 2.3 years. All-cause mortality was higher in subjects with high A1C(CV) (11.0% vs. 5.4%, p=0.002). In the Kaplan-Meier failure curve, subjects with higher A1C(CV) demonstrated a trend of higher mortality (p=0.1). In multivariate Cox's proportional hazards models, A1C(SD) and A1C(CV) significantly predicted all-cause mortality with an HR of 1.987 (p=0.02) and 1.062 (p=0.013), respectively, after adjusting for age, gender, body mass index, duration of diabetes, mean systolic blood pressure, use of antihypertensives and statins, mean LDL-cholesterol, smoking status, chronic kidney disease, and mean A1C values (A1C(MEAN)). The ability of A1C(SD) and A1C(CV) to predict all-cause mortality was more evident in subjects with relatively low A1C(MEAN.)
A1C variability is an important risk factor for all-cause mortality in type 2 diabetic patients.
- SourceAvailable from: PubMed Central[Show abstract] [Hide abstract]
ABSTRACT: The effects of glucose on cardiovascular events or mortality in nondiabetic patients has been recently reported. However, since atherosclerosis can be formed over a long period of time, it is necessary to devote several years to unveil the relationship between the two factors. Here, we attempted to find out the relationship between the mean hemoglobin A1c (HbA1c) level and HbA1c variability for 5 years and coronary artery disease (CAD) by using coronary angiography (CAG) to assess nondiabetic patients. We reviewed patients who performed CAG who were followed up for at least 5 years after the initial diagnosis. The fasting blood test was performed annually for glucose and HbA1c level. CAD was defined as more than 50% of luminal narrowing. The severity of CAD was divided into two groups depending on whether no vessels were involved or one more vessel were involved (CAD(-) or CAD(+), respectively). The patients in CAD(+) group had higher mean HbA1c level for 5 years than CAD(-) group (5.71±0.40 vs. 5.86±0.68; P=0.04). Mean HbA1c was a significant predictor for CAD in multiple regression (odds ratio, 2.224; P=0.028). The percentage of patients with CAD was significantly higher in patients with >6.2% of mean HbA1c levels compared to patients with <6.2% of mean HbA1c levels (P<0.019). When the mean HbA1c levels were above 6.2%, the risk of CAD was higher. Also this study shows that HbA1c level can be one of the predictors for CAD even if the patients do not have diabetes.Diabetes & metabolism journal 02/2014; 38(1):58-63.
- [Show abstract] [Hide abstract]
ABSTRACT: Aims We aimed to evaluate the association between HbA1c variability and mortality due to all causes, cancer, and non-cancer in patients with type 2 diabetes independently of mean HbA1c levels. Methods We enrolled 754 patients with type 2 diabetes who first visited our hospital between 1995 and 1996, had been followed for at least 2 years, and had undergone four or more HbA1c determinations. Patients were followed through June 2012. The standard deviation (SD) or coefficient of variation (CV) was used as a measure of HbA1c variability. Risk of death was evaluated by multivariate Cox proportional hazard models. Results Through June 2012, 63 patients died. Hazard ratios (HRs) for all-cause mortality and non-cancer mortality including cardiovascular diseases (CVD) increased across tertiles of both HbA1cSD and HbA1cCV. HRs for cancer mortality did not increase across tertiles of either HbA1cSD or HbA1cCV. Using a stepwise regression method, both HbA1cSD and HbA1cCV predicted all-cause mortality, especially non-cancer mortality. In contrast, mean HbA1c predicted cancer mortality. Conclusions HbA1c variability is a predictor of all-cause mortality, especially non-cancer mortality including CVD, in patients with type 2 diabetes, independent of mean HbA1c level. In contrast, mean HbA1c, but not HbA1c variability, might predict cancer mortality.Journal of diabetes and its complications 01/2014; · 2.11 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: AIMS: We hypothesised that serum fibroblast growth factor 21 (FGF-21), a novel adipokine with postulated insulin-sensitizing effects, may be predictive of cardiovascular (CV) events in patients with type 2 diabetes (DM2) at a relatively short-term follow-up. METHODS: Serum FGF-21 levels were assessed in 87 DM2 patients, aged 57-66 years, with the median duration of diabetes of 10 years, who were referred to the Department of Endocrinology for routine annual metabolic assessment. During a follow-up of 24 months, overall mortality, CV mortality and CV nonfatal events were registered. Cox proportional hazards regression assessed adjusted differences in CV morbidity and mortality risk. RESULTS: Patients stratified according to serum FGF-21 levels ≤ and > the median value of 240.7pg/mL showed no significant differences at baseline in gender distribution, diabetes duration, insulin therapy, BMI, biochemical profiles and previous CV events. At 24-month follow-up, 21 (24.1%) patients experienced a nonfatal CV event. A significantly (P=0.0013) higher incidence of the combined end point of CV morbidity and mortality was observed in the FGF-21>240.7pg/mL group. In the multivariate Cox proportional hazards regression model, the presence of FGF-21>the median value was associated with a significant increase in the risk of the combined end point of CV morbidity and mortality (HR: 4.7, 95% CI 1.67-13.24). CONCLUSIONS: The obtained results support the prognostic value of FGF-21 in DM2 and may provide a useful tool for stratification of CV prognosis in DM2 patients.Diabetes research and clinical practice 06/2013; · 2.74 Impact Factor
Variability in hemoglobin A1c predicts all-cause mortality in patients with
type 2 diabetes☆,☆☆
Wen-Ya Maa,1, Hung-Yuan Lib,1, Dee Peia, Te-Lin Hsiaa, Kuo-Cheng Luc, Li-Yu Tsaid,
Jung-Nan Weie, Ching-Chieh Suf,⁎
aDivision of Endocrinology, Department of Internal Medicine, Cardinal Tien Hospital, Xindian; School of Medicine, Fu-Jen Catholic University, New Taipei City, Taiwan
bDepartment of Internal Medicine, National Taiwan University Hospital and College of Medicine, National Taiwan University, Taipei, Taiwan
cDivision of Nephrology, Department of Internal Medicine, Cardinal Tien Hospital, Xindian; School of Medicine, Fu-Jen Catholic University, New Taipei City, Taiwan
dDivision of Internal medicine, Cardinal Tien Hospital, Yung Ho Branch, Taiwan
eChia Nan University of Pharmacy and Science, Tainan, Taiwan
fDepartments of Medicine, Cardinal Tien Hospital, and School of Medicine and the Graduate Institute of Applied Science and Engineering, Fu-Jen Catholic University,
New Taipei City, Taiwan
a b s t r a c t a r t i c l ei n f o
Received 26 October 2011
Received in revised form 28 February 2012
Accepted 22 March 2012
Available online 23 May 2012
Type 2 diabetes
Background: To evaluate the relationship between hemoglobin A1c variability and all-cause mortality in type
2 diabetic patients.
Methods: This was a retrospective cohort study in type 2 diabetic patients followed for at least 2 years
between 2003 and 2009. A1C variability was determined from the standard deviation or coefficient of
variation of serial A1C values (A1CSDor A1CCV). Subjects were categorized into either the high or low A1C
variability group according to their A1CCVmedian. Hazard ratios (HRs) of various factors for all-cause
mortality were determined from Cox's proportional hazard models.
Results: A total of 881 subjects (422 men, 459 women) were included and 73 (8.3%) died during follow-up.
The follow-up period was 4.7±2.3 years. All-cause mortality was higher in subjects with high A1CCV(11.0%
vs. 5.4%, p=0.002). In the Kaplan–Meier failure curve, subjects with higher A1CCVdemonstrated a trend of
higher mortality (p=0.1). In multivariate Cox's proportional hazards models, A1CSDand A1CCVsignificantly
predicted all-cause mortality with an HR of 1.987 (p=0.02) and 1.062 (p=0.013), respectively, after
adjusting for age, gender, body mass index, duration of diabetes, mean systolic blood pressure, use of
antihypertensives and statins, mean LDL-cholesterol, smoking status, chronic kidney disease, and mean A1C
values (A1CMEAN). The ability of A1CSDand A1CCVto predict all-cause mortality was more evident in subjects
with relatively low A1CMEAN.
Conclusions: A1C variability is an important risk factor for all-cause mortality in type 2 diabetic patients.
© 2012 Elsevier Inc. All rights reserved.
Type 2 diabetes results in numerous complications and is a major
cause of death worldwide. Hemoglobin A1c (A1C), a marker of
glycation, may be a therapeutic target for type 2 diabetes, since a
reduction in A1C levels lowers the risk of diabetes-related complica-
tions and mortality (DCCT, 1993; DCCT, 1995; Turner, 1998;
Tuomilehto, 1999). However, recent large clinical trials failed to
lower all-cause mortality by lowering glucose to near-normal levels
(Gerstein et al., 2008; Patel et al., 2008; Duckworth et al., 2009;
Gerstein et al., 2011). Although hypoglycemia may, in part, explain
the unfavorable results of intensive glycemic control, other factors,
such as glucose variability, may also play an important role.
Mean glycemic control can be determined via continuous
glucose monitoring (CGM), as well as from fasting plasma glucose
(FPG), plasma fructosamine, plasma glycated albumin, and plasma
A1C concentrations (Suhonen, Stenman, Koivisto, & Teramo, 1989;
Koskinen, Erkkola, Viikari, Mattila, & Irjala, 1992; Monnier, Colette,
& Owens, 2008). Conversely, glucose variability can be determined
from the standard deviation (SD) or coefficient of variation (CV) of
serial glycemic indices. One study has previously shown that a
greater variation in FPG is associated with increased mortality,
independent of mean FPG (Muggeo et al., 2000). Additionally, in
Journal of Diabetes and Its Complications 26 (2012) 296–300
☆ Author Contributions: WY Ma, HY Li, and CC Su analyzed and interpreted the data,
contributed to the discussion, wrote the manuscript, and reviewed and edited the
manuscript. D Pei, TL Hsia, KC Lu, LY Tsai and JN Wei contributed to the discussion, and
reviewed and edited the manuscript.
☆☆ Disclosure: There were no conflicts of interest.
⁎ Corresponding author. Department of Internal Medicine, Cardinal Tien Hospital,
No. 362. Zhongzheng Rd. Xindian Dist, New Taipei city, 231 Taiwan, R.O.C. Tel.: +886
918050479, fax: +886 2 22193923.
E-mail address: email@example.com (C.-C. Su).
1Authors contributed equally to this study.
1056-8727/$ – see front matter © 2012 Elsevier Inc. All rights reserved.
Contents lists available at SciVerse ScienceDirect
Journal of Diabetes and Its Complications
journal homepage: WWW.JDCJOURNAL.COM
type 1 diabetes, A1C variability predicts the incidence of cardio-
vascular events, microalbuminuria, and overt diabetic nephropathy
(Waden et al., 2009).
To our best knowledge, the effects of A1C variability on all-cause
mortality have not yet been studied in subjects with type 2 diabetes.
Therefore, this issue was explored in the present study.
2. Materials and methods
Retrospective clinical data from patients with type 2 diabetes, who
attended the Diabetes Shared Care Program between 2003 and 2009
at the Cardinal Tien Hospital were used in the present study. These
patients regularly visited the endocrinology and metabolism outpa-
tient department on average every 3 months. Baseline patient
characteristics, including age, gender, time of disease onset, family
history, and smoking status, were obtained by trained diabetes
educators and uploaded into a computer database. Blood pressure,
height, and body weights were measured by an outpatient depart-
ment nurse during every visit. Body mass index (BMI) was also
calculated as follows: BMI=body weight/(body length)2. A1C levels
were assessed every 2–4 months, and other biochemical data,
including total cholesterol, triglycerides, low-density lipoprotein
(LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, alanine
aminotransferase (ALT), aspartate aminotransferase (Gikas et al.,
2009), serum creatinine, and spot urine albumin–creatinine ratio
(ACR) were determined annually. The study was reviewed and
approved by the Institutional Review Board.
2.2. Data management
Subjects with a follow-up period of less than 2 years or those that
did not have a follow-up in over 13 months were excluded. Patients
werefollowed-upuntil November31, 2010. Vitalstatus, dateof death,
and cause of death of all subjects were obtained from computerized
death certificates maintained by the Department of Health, Executive
Yuan in Taiwan (Li et al., 2011). Based on medical records, the use of
antihypertensives or statins was defined as taking the medication for
at least 6 months. The use of insulin was recorded in the baseline and
the end of the study. The estimated glomerular filtration rate (eGFR)
was calculated by using the abbreviated Modification of Diet in Renal
Disease (MDRD) equation, as follows: eGFR=(186×serum creati-
nine)−(1.154×Age)−0.203 (multiply by −0.742, if female) (Levey
et al., 1999; Levey, Greene, Kusek, & Beck, 2000). The degree of
proteinuria was categorized according to the absence of albuminuria
(spot urine ACRb30 mg/g) or presence of microalbuminuria (spot
urine ACR 30–300 mg/g) or macroalbuminuria (spot urine ACR
≧300 mg/g). Progression of proteinuria was defined if absence of
albuminuria in the baseline with microalbuminuria or macroalbumi-
nuria (spot urine ACR ≧300 mg/g) in the end of study or the presence
of the study.
The mean, minimum, maximum, SD, and CV of serial A1C values
for each patient during the follow-up were calculated and
presented as A1CMEAN, A1CMIN, A1CMAX, A1CSD, and A1CCV, respec-
tively. The CV was calculated, as follows: CV=SD/mean×100%.
A1CSDand A1CCVwere utilized as indices of A1C variability. Subjects
were divided into either the high or low A1C variability group
based on their median A1CCV.
Statistical differences in the clinical characteristics between
subjects with high or low A1C variability were determined via the
Student's t or chi-squared tests. The Kaplan–Meier failure curve for
all-cause mortality was also constructed and analyzed in subjects
with low and high A1C variability. Hazard ratios (HR) of various
factors for all-cause mortality were estimated from univariate Cox's
proportional hazard models. Factors with a p-value of b0.2 were
included in the multivariate Cox's proportional hazard models.
Smoking status, duration of diabetes, and mean systolic blood
pressure were considered as potential confounders, and thereby
were adjusted. Proportional hazard assumptions were tested by
goodness of fit based on Schoenfeld residuals. The predictive ability of
Cox's proportional hazard model was tested by Harrell's C concor-
dance statistic, which is similar to the area under the receiver-
operating characteristic (ROC) curve and ranged from 0.5 (no
predictive ability) to 1 (perfect predictive ability). A two tailed p-
value below 0.05 was considered significant. Stata/SE 11.0 (StataCorp
LP, TX) was used for statistical analyses.
A total of 881 subjects (422 men, 459 women) were included in
this study. Seventy-three (8.3%) subjects died during the follow-up
period. The average follow-up period was 4.7±2.3 years. The average
number of A1C measurements conducted was 12±7, and the
Subject characteristics by A1C variability during the follow-up period.
Variations in glycemic controlsA1CCV
Follow-up period (years)
Male, N (%)
Current smoker, N (%)
Waist circumference (cm)
Mean systolic BP (mmHg)
Mean diastolic BP (mmHg)
Use of antihypertensives, N (%)
Duration of diabetes (years)
Use of insulin in the baseline, N (%)
Use of insulin in the study end, N (%)
Mean total cholesterol (mmol/L)
Mean triglyceride (mmol/L)
Mean LDL-cholesterol (mmol/L)
Mean HDL-cholesterol (mmol/L)
Use of statins, N (%)
Baseline serum creatinine (μmol/L)
Spot urine ACR (mg/g) in the baseline
Spot urine ACR (mg/g) at study end
Progression of proteinuria
All-cause mortality (N, %)
Data presented as mean±standard deviation.
A1CMEAN, mean of intrapersonal serial A1C values; A1CSD, standard deviation of
intrapersonal serial A1C values; A1CCV, coefficient of variation of intrapersonal serial
A1C values; A1CMAX, maximum of intrapersonal serial A1C values; A1CMIN, minimum of
intrapersonal serial A1C values, ALT, alanine aminotransferase; AST, aspartate
aminotransferase; ACR, albumin–creatinine ratio.
Progression of proteinuria was defined if absence of albuminuria (spot urine
ACRb30 mg/g) in the baseline with microalbuminuria (spot urine ACR 30–300 mg/g)
or macroalbuminuria (spot urine ACR≧300 mg/g) in the end of study or the presence of
microalbuminuria in the baseline with macroalbuminuria in the end of the study.
W.-Y. Ma et al. / Journal of Diabetes and Its Complications 26 (2012) 296–300
A1CMEANwas 7.6%±1.3%. The details of A1C data were summarizedin
Supplementary table 1. There were 333 (37.8%) subjects that had an
A1CMEAN within the therapeutic goal (A1C≤7%), and 43 (4.9%)
subjects that had an A1CMEANhigher than 10%. The major causes of
death were: cancer (n=22; 30%), diabetes (n=17; 23%), pneumonia
or respiratory disease (n=12; 16%), cardiovascular disease (n=7;
10%), uremia (n=2; 3%), liver disease (n=3; 4%), gastrointestinal
disease (n=2; 3%), cerebrovascular disease (n=2; 3%), and other
Subjects were divided into two groups: low A1C variability
(A1CCVb8%) and high A1C variability (A1CCV≥8%). The characteris-
tics of these two groups were summarized in Table 1. Both the follow-
up period and disease duration were longer in subjects with greater
A1C variability. Subjects with lower A1C variability had more
favorable metabolic profiles, including lower BMI, waist circumfer-
ence, systolic and diastolic blood pressure, total cholesterol, triglyc-
eride, LDL-cholesterol and ALT, than those withhigherA1C variability.
More subjects were using insulin in high A1C variability group both in
the baseline and the end of the study. Subjects with high A1C
variability group were more likely to develop proteinuria during
follow up, but the trend did not reach statistical significance.
Mortality rate was also higher among subjects with high A1C
variability (11.0% vs. 5.4%, p=0.002). Fig. 1 demonstrates the
Kaplan–Meier failure curve of all-cause mortality according to A1C
variability (i.e. A1CCVb8% vs. A1CCV≥8%). There was a trend of
increased mortality among subjects with higher A1C variability
(p=0.1 by log-rank test).
The HR and 95% CI of various risk factors for all-cause mortality
are presented in Fig. 2. The unadjusted HR (95% CI) of A1CMEAN,
A1CMAX, A1CMIN, A1CSD and A1CCV were 1.05 (95% CI, 0.87–1.26,
p=0.6), 1.02 (95% CI, 0.91–1.15, p=0.7), 0.99 (95% CI, 0.76–1.28,
p=0.9), 1.54 (95% CI, 1.08–2.20, p=0.02), and 1.05 (95% CI, 1.01–
1.08, P=0.009), respectively. Aging, male gender, elevated LDL-
cholesterol, A1CSDand A1CCVwere associated with an increased risk
of death (age: HR=1.08 [95% CI, 1.06–1.11], pb0.001; male gender:
HR=1.96 [95% CI, 1.22–3.13], p=0.005; LDL-cholesterol: HR=1.65
[95% CI, 1.17–2.33], p=0.005). However, subjects taking antihy-
pertensives and/or statins had a lower risk of mortality (use of
antihypertensives: 0.41 [95% CI, 0.30–0.54], pb0.001; use of statins:
HR=0.10 [95% CI 0.03–0.33]).
Based on multivariate Cox's proportional hazard models
(Table 2), A1C variability, defined by A1CSD (model 1;
HR=1.987, p=0.02) or A1CCV(model 2; HR=1.062, p=0.013),
independently predicted all-cause mortality after adjusting for age,
gender, BMI, duration of diabetes, current smoker, mean systolic
blood pressure, use of antihypertensives and/or statins, mean LDL-
cholesterol, presence of chronic kidney disease, and A1CMEAN.
These findings were similar if subjects with less A1C data (Nb4 or
Nb5) were excluded (Supplementary table 2). In patients with an
A1CMEANb7.3% (the median of A1CMEAN), the adjusted HRs of
A1CSD and A1CCV were 3.31 (95% CI, 0.94–11.61, p=0.061) and
1.09 (95% CI, 1.00–1.18, p=0.056), respectively, while in those with
an A1CMEAN≥7.3%, the adjusted HRs of A1CSDand A1CCVwere 1.76
(95% CI, 0.87–3.54, p=0.12) and 1.05 (95% CI, 0.99–1.12, p=0.12),
respectively. Cardiovascular mortality cannot be predicted by A1C
Fig. 1. A Kaplan–Meier failure curve of all-cause mortality of subjects with low and high
A1C variability. Solid line, A1CCV≥8%; dash line, A1CCVb8%; p-values by log-rank
Fig. 2. Hazard ratios of various risk factors for all-cause mortality in type 2 diabetes (N=881), as determined by univariate Cox's proportional regression models. An estimated
glomerular filtration rate of b60 mL.min−1.1.73 m−2indicated the presence of chronic kidney disease; SD, standard deviation; CV, coefficient of variation. *pb0.05.
W.-Y. Ma et al. / Journal of Diabetes and Its Complications 26 (2012) 296–300
variability, either A1CSDor A1CCV. The unadjusted HRs of A1CSDand
A1CCVwere 0.98 (95% CI, 0.87–1.10, p=0.74) and 0.70 (95% CI, 0.18–
2.81, p=0.62), respectively. After adjusting for age, gender, BMI,
current smoker, mean systolic blood pressure, use of antihyperten-
sives, use of statin, mean LDL-cholesterol, and A1CMEAN,the adjusted
HRs for A1CSDand A1CCVwere 1.19 (95% CI, 0.18–7.72, p=0.87) and
1.02 (95% CI, 0.89–1.18, p=0.75), respectively.
The Harrell's C indexes of concordance statistics for models 1
and 2 were 0.8536 and 0.8542, respectively, indicating good
predictability in each model (Table 2). The increments in concor-
dance statistics for A1CSD and A1CCV were 0.0016 and 0.0022,
respectively. Age, gender, use of antihypertensives and/or statins,
and the presence of chronic kidney disease increased the concor-
dance statistics in both models.
To our best knowledge, this is the first study to evaluate the
relationship between the A1C variability and all-cause mortality. In
the present study, it was found that type 2 diabetic patients with
greater A1C variability had an increased risk of mortality. In adjusted
all-cause mortality, particularly in subjects with near-normal average
A1C levels. In a large cohort of type 2 diabetes, increased mortality
was associated with higher baseline A1C values and the positive
trajectory of multiple A1C measurements (Gebregziabher, Egede,
Lynch, Echols, & Zhao, 2010). Our findings demonstrated the
importance of stable glycemic control on mortality in patients with
type 2 diabetes.
Fluctuations in glucose levels may induce endothelial dysfunc-
tion and are associated with higher oxidative stress (Ceriello et al.,
2008). Numerous studies have shown that intermittent hypergly-
cemia can enhance cell proliferation, induce the release of
cytokines, impair endothelial function, and increase oxidative
stress, all of which contribute to the mechanisms of diabetes
complications (Jones, Saunders, Qi, & Pollock, 1999; Brownlee,
2001; Quagliaro et al., 2003). Using CGM to monitor glucose
fluctuation, Monnier et al. reported that oxidative stress correlates
with the mean amplitude of glycemic excursion (MAGE) and
postprandial hyperglycemia, but not with A1C or mean glucose
concentrations (Monnier et al., 2006). In a recent analysis from the
diabetes control and complication trial (DCCT), A1C variability was
found to predict the development of diabetic nephropathy and
retinopathy (Kilpatrick, Rigby, & Atkin, 2008). In our result, the
variability of A1C could predict all-cause mortality but not
cardiovascular mortality. Factors affecting A1C variability other
than glucose excursion should be also responsible for the finding.
In addition to intermittent hyperglycemia, risk of hypoglycemia
may also contribute to increased mortality (Skyler et al., 2009).
Evidence from large clinical trials previously implicated the poten-
tially lethal effect of severe hypoglycemia accompanied by intensive
glucose lowering, especially in subjects with a long duration of
diabetes or established cardiovascular diseases (Gerstein et al., 2008;
Patel et al., 2008; Duckworth et al., 2009; Skyler et al., 2009). Indeed,
corroborating previous reports, the present study showed that the HR
of A1C variability was greater among patients with near-normal
average A1C levels.
Moreover, A1C variability can be contributed by seasons and
culture (Gikas et al., 2009). Gikas et al found that A1C is
significantly higher in colder months, especially after Christmas
and Easter months (Gikas et al., 2009). Chen et al also reported the
effect of winter holiday on A1C elevation and emphasize the
importance of education before holidays (Chen, Jap, Chen, & Lin,
2004; Chen, Wu, Jap, Chen, & Lin, 2008). Higher A1C variability may
also be the consequence of poor treatment adherence (Asche,
Lafleur, & Conner, 2011), use of glucocorticoids or anti-psychotics,
which may impair glycemic control (Newcomer, 2007), complex
underlying disorders, poor quality of life, financial instability
(Maciejewski, Dowd, Call, & Feldman, 2001), low socioeconomic
status (Welch, Sharp, Gottlieb, Skinner, & Wennberg, 2011), and the
lack of physiological and/or psychological support systems (Hamer,
Stamatakis, Kivimaki, Pascal Kengne, & Batty, 2011), all of which
may also lead to higher mortality. In our result, the Kaplan–Meier
curves of total mortality were parallel rather than diverting
between those with high and low A1C variability. Taken together,
higher A1C variability might be a marker of less responsible
behavior and unhealthy habits and not a detrimental factor itself.
The strength of this study was the 100% follow-up rate,
accompanied by the accurate records of the vital status. However,
this study was limited in that an analysis of the relationship between
A1C variability and disease-specific causes of death was not possible
due to the low numbers of subjects that died from the various causes
of death. Moreover, the subjects in this study were only slightly
overweight by European or American standards, the study results
cannot be readily applied to the western population or should be
interpreted with caution.
In conclusion, A1C variability, as defined by A1CSDor A1CCV, can
predict all-cause mortality in type 2 diabetic patients, independent
of A1CMEAN and other confounders. Thus, maintaining a stable
glycemic control in type 2 diabetic patients may be important for
The authors would like to thank the Department of Health,
Executive Yuan in Taiwan for maintaining the computerized death
certificates and the medical editor of MedCom Asia, Inc. for help in the
preparation of this manuscript.
Appendix A. Supplementary data
Supplementary data to this article can be found online at doi:10.
Asche, C., Lafleur, J., & Conner, C. (2011). A review of diabetes treatment adherence and
the association with clinical and economic outcomes. Clinical Therapeutics, 33(1),
Brownlee, M. (2001). Biochemistry and molecular cell biology of diabetic complica-
tions. Nature, 414(6865), 813–820.
Hazard ratios (95% CI) of risk factors for all-cause mortality, as determined by
multivariate Cox's proportional regression hazard models.
Model 1Model 2
Duration of diabetes (years)
Mean systolic blood pressure (mmHg)
Use of antihypertensives
Mean LDL-cholesterol (mmol/L)
Use of statins
Baseline eGFRb60 mL.min−1.1.73 m−2
HR, hazard ratio; A1CMEAN, mean of intrapersonal serial A1C values; A1CSD, standard
deviation of intrapersonal serial A1C values; A1CCV, coefficient of variation of
intrapersonal serial A1C values.
W.-Y. Ma et al. / Journal of Diabetes and Its Complications 26 (2012) 296–300
glucose is more deleterious to endothelial function and oxidative stress than mean
glucose in normal and type 2 diabetic patients. Diabetes, 57(5), 1349–1354.
Chen, H. -S., Jap, T. -S., Chen, R. -L., & Lin, H. -D. (2004). A prospective study of glycemic
control during holiday time in type 2 diabetic patients. Diabetes Care, 27(2),
Chen, H. S., Wu, T. E., Jap, T. S., Chen, R. L., & Lin, H. D. (2008). Effects of health education
on glycemic control during holiday time in patients with type 2 diabetes mellitus.
American Journal of Managed Care, 14(1), 45–51.
DCCT. (1993). The effect of intensive treatment of diabetes on the development and
progression of long-term complications in insulin-dependent diabetes mellitus.
The Diabetes Control and Complications Trial Research Group. New England Journal
of Medicine, 329(14), 977–986.
DCCT. (1995). The relationship of glycemic exposure (HbA1c) to the risk of
development and progression of retinopathy in the diabetes control and
complications trial. Diabetes, 44(8), 968–983.
Duckworth, W.,Abraira, C., Moritz, T., Reda, D., Emanuele, N., Reaven, P. D., et al. (2009).
Glucose control and vascular complications in veterans with type 2 diabetes. New
England Journal of Medicine, 360(2), 129–139.
Gebregziabher, M., Egede, L. E., Lynch, C. P., Echols, C., & Zhao, Y. (2010). Effect of
trajectories of glycemic control on mortality in type 2 diabetes: a semiparametric
joint modeling approach. American Journal of Epidemiology, 171(10), 1090–1098.
Gerstein, H. C., Miller, M. E., Byington, R. P., Goff, D. C., Jr., Bigger, J. T., Buse, J. B., et al.
(2008). Effects of intensive glucose lowering in type 2 diabetes. New England
Journal of Medicine, 358(24), 2545–2559.
Gerstein, H. C., Miller, M. E., Genuth, S., Ismail-Beigi, F., Buse, J. B., Goff, D. C., Jr., et al.
(2011). Long-term effects of intensive glucose lowering on cardiovascular
outcomes. New England Journal of Medicine, 364(9), 818–828.
Gikas, A., Sotiropoulos, A., Pastromas, V., Papazafiropoulou, A., Apostolou, O., & Pappas,
S. (2009). Seasonal variation in fasting glucose and HbA1c in patients with type 2
diabetes. Primary Care Diabetes, 3(2), 111–114.
Hamer, M., Stamatakis, E., Kivimaki, M., Pascal Kengne, A., & Batty, G. D. (2011).
Psychological distress, glycated hemoglobin, and mortality in adults with and
without diabetes. Psychosomatic Medicine, 72(9), 882–886.
Jones, S. C., Saunders, H. J., Qi, W., & Pollock, C. A. (1999). Intermittent high glucose
enhances cell growth and collagen synthesis in cultured human tubulointerstitial
cells. Diabetologia, 42(9), 1113–1119.
Kilpatrick, E. S., Rigby, A. S., & Atkin, S. L. (2008). A1C variability and the risk of
microvascular complications in type 1 diabetes: data from the Diabetes Control and
Complications Trial. Diabetes Care, 31(11), 2198–2202.
Koskinen, P., Erkkola, R., Viikari, J., Mattila, K., & Irjala, K. (1992). Blood glycated
haemoglobin, serum fructosamine, serum glycated albumin and serum glycated
total protein as measures of glycaemia in diabetes mellitus. Scandinavian Journal of
Clinical and Laboratory Investigation, 52(8), 863–869.
Levey, A. S., Bosch, J. P., Lewis, J. B., Greene, T., Rogers, N., & Roth, D. (1999). A more
accurate method to estimate glomerular filtration rate from serum creatinine: a
new prediction equation. Annals of Internal Medicine, 130(6), 461–470.
Levey, A. S., Greene, T., Kusek, J., & Beck, G. (2000). A simplified equation to predict
glomerular filtration rate from serum creatinine [Abstract]. Journal of the American
Society of Nephrology, 11, 155A.
Li, H. -Y., Jiang, Y. -D., Chang, T. -J., Wei, J. -N., Lin, M. -S., Lin, C. -H., et al. (2011). Serum
vascular adhesion protein-1 predicts 10-year cardiovascular and cancer mortality
in individuals with type 2 diabetes. Diabetes, 60(3), 993–999.
Maciejewski, M. L., Dowd, B., Call, K. T., & Feldman, R. (2001). Comparing mortality and
time until death for medicare HMO and FFS beneficiaries. Health Services Research,
Monnier, L., Colette, C., & Owens, D. R. (2008). Glycemic variability: the third
component of the dysglycemia in diabetes. Is it important? How to measure it?
Journal of Diabetes Science and Technology, 2(6), 1094–1100.
Monnier, L., Mas, E., Ginet, C., Michel, F., Villon, L., Cristol, J. P., et al. (2006). Activation of
oxidative stress by acute glucose fluctuations compared with sustained chronic
hyperglycemia in patients with type 2 diabetes. Journal of the American Medical
Association, 295(14), 1681–1687.
Muggeo, M., Zoppini, G., Bonora, E., Brun, E., Bonadonna, R. C., Moghetti, P., et al. (2000).
Fasting plasma glucose variability predicts 10-year survival of type 2 diabetic
patients: the Verona Diabetes Study. Diabetes Care, 23(1), 45–50.
Newcomer, J. W. (2007). Antipsychotic medications: metabolic and cardiovascular risk.
Journal of Clinical Psychiatry, 68(Suppl. 4), 8–13.
Patel, A., MacMahon, S., Chalmers, J., Neal, B., Billot, L., Woodward, M., et al. (2008).
Intensive blood glucose control and vascular outcomes in patients with type 2
diabetes. New England Journal of Medicine, 358(24), 2560–2572.
Quagliaro, L., Piconi, L., Assaloni, R., Martinelli, L., Motz, E., & Ceriello, A. (2003).
Intermittent high glucose enhances apoptosis related to oxidative stress in human
umbilical vein endothelial cells: the role of protein kinase C and NAD(P)H-oxidase
activation. Diabetes, 52(11), 2795–2804.
Skyler, J. S., Bergenstal, R., Bonow, R. O., Buse, J., Deedwania, P., Gale, E. A., et al. (2009).
Intensive glycemic control and the prevention of cardiovascular events: implica-
tions of the ACCORD, ADVANCE, and VA diabetes trials: a position statement of the
American Diabetes Association and a scientific statement of the American College
of Cardiology Foundation and the American Heart Association. Diabetes Care, 32(1),
Suhonen, L., Stenman, U., Koivisto, V., & Teramo, K. (1989). Correlation of HbA1C,
glycated serum proteins and albumin, and fructosamine with the 24-h glucose
profile of insulin-dependent pregnant diabetics. Clinical Chemistry, 35(6),
Tuomilehto, J. (1999). Glucose tolerance and mortality: comparison of WHO and
American Diabetes Association diagnostic criteria. The DECODE study group.
European Diabetes Epidemiology Group. Diabetes Epidemiology: Collaborative
analysis Of Diagnostic criteria in Europe. Lancet, 354(9179), 617–621.
Turner, R. (1998). 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,
Waden, J., Forsblom, C., Thorn, L. M., Gordin, D., Saraheimo, M., & Groop, P. H. (2009).
A1C variability predicts incident cardiovascular events, microalbuminuria, and
overt diabetic nephropathy in patients with type 1 diabetes. Diabetes, 58(11),
Welch, H. G., Sharp, S. M., Gottlieb, D. J., Skinner, J. S., & Wennberg, J. E. (2011).
Geographic variation in diagnosis frequency and risk of death among Medicare
beneficiaries. Journal of the American Medical Association, 305(11), 1113–1118.
W.-Y. Ma et al. / Journal of Diabetes and Its Complications 26 (2012) 296–300