1 Epidemiology/Health Services Research O R I G I N A L A R T I C L E. Low Hemoglobin A1c in Nondiabetic Adults An elevated risk state? VIKAS AGGARWAL, MD1 RESEARCH DESIGN AND. ANDREA SCHNEIDER, PHD2,3 METHODS. ELIZABETH SELVIN, PHD, MPH2,3. Study population The Atherosclerosis Risk in Communities OBJECTIVEdTo identify predictors of low Hemoglobin A1c (HbA1c) (, ) and to inves- (ARIC) Study is an ongoing community- tigate the association of low HbA1c with cause-speci c mortality and risk of liver disease hospi- based prospective cohort study of 15,792.
2 Talization. middle-aged Adults from four com- RESEARCH DESIGN AND METHODSdProspective cohort study of 13,288 partici- munities: Washington County, MD; sub- pants in the Atherosclerosis Risk in Communities Study. Logistic regression was used to identify urban Minneapolis, MN; Jackson, MS;. cross-sectional correlates of low HbA1c and Cox proportional hazards models were used to and Forsyth County, NC. The rst study estimate the association of low HbA1c with cause-speci c mortality. visit occurred between 1987 and 1989. with three follow-up visits that occurred RESULTSdCompared with participants with HbA1c in the normal range ( to , ), participants with low HbA1c were younger, less likely to smoke, had lower BMI, lower white cell approximately every 3 years (13,14).
3 Visit count and brinogen levels, and lower prevalence of hypercholesterolemia and history of cor- 2 (1990 1992) was attended by 14,348. onary heart disease. However, this group was more likely to have anemia and had a higher mean participants and is the baseline for the corpuscular volume. In adjusted Cox models with HbA1c of to , as the reference group, present analysis. Participants were in- HbA1c , was associated with a signi cantly increased risk of all-cause mortality (hazard cluded in our analyses irrespective of ratio [HR]: , 95% CI: ) and of cancer death ( , 95% CI: ).
4 We also the previous occurrence of nonfatal noted nonsigni cant trends toward increased risk of death from cardiovascular causes ( ; events. We excluded participants who 95% CI, ) and respiratory causes ( , 95% CI: ). There was a J-shaped self-identi ed as other than white or black association between HbA1c and risk of liver disease hospitalization. race (n 5 48) or who were missing data on CONCLUSIONSdNo single cause of death appeared to drive the association between low HbA1c or other covariates of interest (n 5. HbA1c and total mortality.)
5 These results add to evidence that low HbA1c values may be a gen- 970), leaving a nal sample size of 13,288. eralized marker of mortality risk in the general population. participants in this analysis. Institutional review boards at each clinical site ap- proved the study protocol, and written in- formed consent was obtained from all H. emoglobin A1c (HbA1c) is the stan- (3,7 9). Previous studies in Nondiabetic participants. dard measure of glucose control in populations have also reported a J-shaped persons with diagnosed diabetes association of HbA1c with all-cause mor- Measurement of HbA1c mellitus and is now recommended for tality (3,7,10,11).
6 The objectives of this Frozen whole-blood samples collected use as a diagnostic test for diabetes study were to examine predictors of at ARIC visit 2 were thawed and assayed (1,2). The 2010 American Diabetes Asso- low HbA1c ( , , ) and investigate for the measurement of HbA 1c using ciation recommendations for use of the association of low HbA 1c with all- high-performance liquid chromatogra- HbA1c as a diagnostic test will likely in- cause and cause-speci c mortality in phy (Tosoh A1c Plus Glycohemoglo- crease its use in persons without a prior community-based population.)
7 Because bin Analyzer method in 2003 to 2004. diagnosis of diabetes. A number of studies recent studies have shown a high preva- and the Tosoh G7 method in 2007 to have demonstrated that HbA1c values, lence of liver disease among persons with 2008; Tosoh Corporation) (15). Both in- even below the diagnostic threshold of low HbA1c (10,12), we also examined the struments were standardized to the Dia- , are associated with clinical outcomes association between low HbA1c and risk betes Control and Complications Trial including cardiovascular events (3 5), of liver disease hospitalization in this assay (16).
8 Kidney disease (6), and total mortality cohort. c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c Outcomes 1. ARIC Study investigators conduct contin- From the Department of Medicine, Division of Cardiology, University of Colorado (Anschutz Medical Campus), Aurora, Colorado; the2 Department of Epidemiology, Bloomberg School of Public Health, Johns uous surveillance for all hospitalizations Hopkins University, Baltimore, Maryland; and the3 Welch Center for Prevention, Epidemiology, and and deaths among participants via an- Clinical Research, Johns Hopkins University, Baltimore, Maryland.
9 Nual phone calls to participants or prox- Corresponding author: Elizabeth Selvin, ies, and detailed information on deaths is Received 27 December 2011 and accepted 2 May 2012. obtained from family members, coroner DOI: 2012 by the American Diabetes Association. Readers may use this article as long as the work is properly reports, or health department death cer- cited, the use is educational and not for pro t, and the work is not altered. See ti cates. Methods for the ascertainment of licenses/by-nc- for details. death and its causes in ARIC have been DIABETES CARE 1.
10 Diabetes Care Publish Ahead of Print, published online August 1, 2012. Low HbA1c and mortality published previously (13). We classi ed , , to , , and $ or diag- The odds ratios and 95% CIs for deaths according to underlying cause, on nosed diabetes) (2). With the HbA1c cat- predictors of low HbA 1c are shown in the basis of coding from the ICD-9 and -10. egory of to , as reference, we Table 2. Compared with those with HbA1c We divided causes of death into the fol- used logistic regression models to identify values to , , participants with lowing major diagnosis categories de- predictors of low HbA1c (, ) at base- HbA 1c values , were younger, ned by the ICD codes: 1) cancers line among persons with HbA1c , had a lower BMI, were less likely to (ICD-10 codes of C00-D48, ICD-9 codes (n 5 9,254).