Abstract
Objective:
The objective of this study was to examine the sex-specific associations of mutually exclusive iron and anemia status categories with HbA1c levels among US Hispanics/Latinos without self-reported diabetes mellitus.
Methods:
Baseline cross-sectional data (7,247 women and 4,904 men without self-reported diabetes mellitus) from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL) were analyzed. Per ADA-defined criteria, HbA1c levels were categorized as normoglycemia, prediabetes, and probable diabetes. Iron-anemia status categories were the following: no anemia and no iron deficiency (NANID; reference), iron deficiency (ID), iron deficiency anemia (IDA), and non-iron deficiency anemia (non-IDA). Multinomial logistic regression models were used to examine the sex-specific associations of iron-anemia status with HbA1c levels adjusting for sociodemographic, lifestyle, and clinical factors.
Results:
The age-standardized prevalence of iron-anemia status categories differed by sex. Compared to those with NANID and normoglycemia, women with IDA had higher odds of prediabetes (OR= 2.18, 95% CI= 1.64, 2.89) and probable diabetes (OR= 3.59, 95% CI= 1.62, 7.99); men with non-IDA had higher odds of probable diabetes (OR= 2.97, 95% CI= 1.13, 7.78). All other associations did not reach statistical significance.
Conclusion:
In US Hispanics/Latinos without self-reported diabetes mellitus, the age-standardized prevalence of ID, IDA, and non-IDA is high and varies by sex. Women with IDA had higher odds of prediabetes and probable diabetes defined based on HbA1c levels. Men with non-IDA had higher odds of probable diabetes defined based on HbA1c levels. Iron-anemia status should be considered when interpreting elevated HbA1c levels among US Hispanics/Latinos without self-reported diabetes mellitus.
Keywords: Hispanics, Latinos, hemoglobin A1c, probable diabetes mellitus, iron deficiency, anemia, iron and anemia status
INTRODUCTION
Previous studies1–4 in adults without diabetes mellitus have documented differences in Hemoglobin A1c (HbA1c) levels across race/ethnicity despite adjustment for sociodemographic, lifestyle, and clinical factors. For instance, we previously described5 higher adjusted mean HbA1c levels in Hispanics/Latinos compared to non-Hispanic Whites. Therefore, clinicians and public health experts need to be aware of potential interferences that can influence the interpretation of HbA1c levels.6 One clinical factor that may alter HbA1c levels are conditions that shorten erythrocyte survival or decrease mean erythrocyte age, such as iron deficiency (ID), iron deficiency anemia (IDA), and non-iron deficiency anemia (non-IDA), in part, due to impairment of innate and cell-mediated immunity.7,8 This presents a challenge because there could be misclassification of pre-diabetes and diabetes using HbA1c among people with iron and anemia deficiencies.9
To date, studies on the associations of ID, IDA, and non-IDA with HbA1c among individuals without diabetes mellitus have yielded inconsistent findings.9–12 Some studies have found that both ID and IDA are associated with higher HbA1c9,10,12–15 but other studies have shown opposite or null results.14–16 A meta-analysis reported no differences in HbA1c levels in the presence of ID or IDA among individuals without diabetes mellitus.15 However, a systematic review concluded that the presence of ID, with or without anemia, is likely to contribute to an increase in HbA1c, whereas, non-IDA may contribute to a decrease in HbA1c in individuals without diabetes mellitus.14 To date, evidence suggests that the sole use of HbA1c without considering other tests such as the oral glucose tolerance test (OGTT; the gold standard) may lead to misclassification of prediabetes and diabetes mellitus among individuals with ID, IDA, or non-IDA, particularly, when HbA1c is elevated but OGTT is normal. Such misclassification between HbA1c and OGTT, also known as discordant classification, could lead to inappropriate clinical decisions and poor planning of disease prevention and treatment plans.
Most studies14,15,17,18 on the association between iron and anemia status with HbA1c have been limited by the inability to evaluate mutually exclusive categories of iron-anemia status, have been conducted in small, clinical, and non-representative samples, and have not examined these relations by sex (despite well-documented sex differences in HbA1c levels and anemia prevalence19). Also, most studies have been conducted on samples of predominantly non-Hispanic White individuals, despite the more significant burden of iron-anemia deficiencies20 and diabetes mellitus prevalence21 in US Hispanics/Latinos. Further, few epidemiological studies have examined the association of iron and anemia status with the classification of prediabetes and diabetes mellitus using HbA1c versus OGTT. Therefore, research is needed to identify clinical factors associated with differences in Hb1Ac levels among adults without self-reported diabetes mellitus including studies in Hispanics/Latinos, a population with a higher risk of prediabetes and diabetes mellitus than non-Hispanic Whites.21
We examined the sex-specific adjusted associations of iron and anemia status categories (NANID, ID, IDA, and non-IDA) with HbA1c levels (normoglycemia, prediabetes, and probable diabetes mellitus per American Diabetes Association [ADA]-criteria). In secondary analyses, we a) estimated the sex-specific proportion of prediabetes and probable diabetes mellitus classification based on HbA1c and OGTT (including discordant and concordant classification between HbA1C and OGTT tests) across iron and anemia status categories; and b) examined the sex-specific association of iron and anemia status categories with the proportion of prediabetes and probable diabetes mellitus classification based on HbA1c and OGTT. Our findings could have implications for identifying high-risk groups and informing clinical decision-making processes.
RESEARCH DESIGN AND METHODS
Study Population
We used baseline data from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL), a cohort study of 16,415 Hispanic/Latino adults. Detailed descriptions of the HCHS/SOL sampling design and methods have been described elsewhere.5,22,23 Protocol is available at the study website (http://sites.cscc.unc.edu/hchs). Blood samples were collected after a fasting period prior to the study visit and measured at a central laboratory. The variability (including within-individual, between-individual, and methodological variability) of the laboratory assays used on the HCHS/SOL study has been previously reported.24 HCHS/SOL was approved by the IRB at each participating site. Participants provided written informed consent.
The inclusion criteria for the present study were defined as participants aged 18–74 years, with ≥8 hours of fasting before the examination, and without self-reported diabetes. We excluded participants with self-reported diabetes to avoid the possible confounding effects of antihyperglycemic treatment on HbA1c. Of the 13,723 participants that met the inclusion criteria, we excluded participants with missing data on HbA1c (n= 155), iron or total hemoglobin values (n=. 328), or any of the covariates (n=. 1,092). Since renal diseases may shorten erythrocyte survival and are associated with both lower HbA1c and ID,19 we also excluded participants with estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m2 or missing data on eGFR (n=450). The final analytic sample of this study was 12,151 (7,247 women and 4,904 men).
Glycemic Levels
HbA1c was measured in EDTA whole blood using a Tosoh G7 Automated HPLC Analyzer. A 75-g OGTT was performed to obtain 2-hour post-load plasma glucose. Per ADA criteria,25 HbA1c was classified as normoglycemia (<5.7% [<39 mmol/mol]), prediabetes (5.7%−6.4% [39–46 mmol/mol]), or probable diabetes (≥6.5% [≥ 48 mmol/mol]); OGTT was classified as normoglycemia (<140 mg/dL), prediabetes (140–199 mg/dL), or probable diabetes (≥200 mg/dL). FPG was measured in EDTA plasma on a Roche Modular P Chemistry Analyzer using a hexokinase enzymatic method. According to our previous work,4 we use the term “probable diabetes” (also known as “suspected diabetes”) because results should be confirmed with repeat testing to diagnose diabetes mellitus.26
Iron and Anemia Status
Hemoglobin, serum iron, ferritin, transferrin saturation, and total iron-binding capacity (TIBC) were used to create the iron and anemia status categories. A hemogram and a platelet count were measured in EDTA whole blood using a Sysmex XE-2100 instrument. Iron and unsaturated iron-binding capacity were measured in serum on a Roche Modular P chemistry analyzer using a ferrozine reagent (Roche Diagnostics, Indianapolis, IN). Ferritin was measured in serum with Roche reagents on a Cobas 6000 Analyzer (Roche Diagnostics) using a particle enhanced immunoturbidimetric assay. TIBC is the total serum iron and unsaturated iron-binding capacity. Transferrin saturation level is calculated by dividing the serum iron by TBIC and, then, multiplying by 100.
Among women ages 18–69 and ≥70 years, low hemoglobin was defined as <12.0 g/dL; and <11.8 g/dL, respectively. Among men ages 18–49, 50–69, and ≥70 years, low hemoglobin was defined as <13.7 g/dL, <13.3 g/dL, and <12.4 g/dL, respectively (higher levels were classified as normal).11,27,28 Serum iron levels were categorized as low (<10.7 μmol/L) or normal (≥10.7 μmol/L).11,27 Serum ferritin levels were categorized as low (<27 pmol/Lng/mL) or normal (≥27 pmol/Lng/mL).29,30 Transferrin saturation was defined as low (<15%) or normal (≥15%) (14,15). TIBC was defined as low (<41 μmol/L), normal (41–77 μmol/L), or elevated (>77 μmol/L).29,30
Per previous research,28,31 participants were classified into one of the following mutually-exclusive categories: (1) no anemia and no iron deficiency (NANID): normal hemoglobin, serum iron, serum ferritin, transferrin saturation, and TIBC; (2) no anemia and iron deficiency (ID): normal hemoglobin and the presence of at least one of the following: low serum iron, low serum ferritin, low transferrin saturation, or elevated TIBC; (3) iron deficiency and anemia (IDA) defined as low hemoglobin and the presence of at least one of the following: low serum iron, low serum ferritin, low transferrin saturation, or elevated TIBC; and (4) non-iron deficiency and anemia (non-IDA) defined as low hemoglobin, normal serum iron, normal or low transferrin saturation, and normal or low TIBC. Figure 1 is a Venn diagram that depicts the iron and anemia status categories used in the present study.
Figure 1. Venn diagram of iron and anemia status categories.
This figure is a Venn diagram depicting the categories of iron and anemia status that were used in the present study. These categories, created based on previous research, are: NANID, ID, IDA, and Non-IDA.
Covariates
Covariates were selected a priori based on their biological plausibility and documented relationship with iron-anemia status or HbA1c.14,15,32 Socio-demographic and lifestyle characteristics were collected via interviewer-administered questionnaires.5 Weight and height were used to compute body mass index (BMI, kg/m2). Hypercholesterolemia was defined as total cholesterol ≥240 mg/dL, or high-density lipoprotein cholesterol ≥160 mg/dL, or low-density lipoprotein cholesterol <40 mg/dL, or antihyperlipidemic medication use. Highly sensitive C reactive protein (hs-CRP), alanine aminotransferase (ALT), and aspartate aminotransferase (AST) were measured on a Roche Modular P Chemistry Analyzer. Dietary iron, folate, vitamin C, and vitamin B12 were collected via two 24-hour dietary recalls.22 Overall dietary quality (2010-AHEI) was computed as previously described.33
Statistical Analysis
Analyses were stratified by sex due to well-established sex differences in HbA1c levels and iron-anemia deficiencies prevalence.19 First, the age-standardized prevalence (to the 2010 U.S. population Census) of iron and anemia status categories was computed for the overall target population and according to sex and Hispanic/Latino background using survey logistic regression predicted marginals. F-tests were used to assess differences in the prevalence of iron and anemia status categories by sex and background. Second, descriptive statistics of key study variables were estimated for women and men (separately) according to iron and anemia status; differences across iron and anemia status categories were assessed using F-tests for continuous variables and chi-square tests for categorical variables.
Third, in our main analyses, odds ratios (ORs) and their 95% confidence intervals (CIs) were calculated using survey-weighted adjusted multinomial logistic regression models on the sex-specific association between iron and anemia status (with NANID defined as the referent) and HbA1c categories (with normoglycemia defined as the referent). Model 1 adjusted for age only. Model 2 additionally included socio-demographic information (Hispanic/Latino background, annual household income, and education), lifestyle (smoking status and alcohol intake), and clinical (BMI, hypercholesterolemia, hs-CRP, ALT, AST, dietary iron, dietary folate, vitamin C, and vitamin B12) covariates. Further adjustment for hypertension or diet quality did not change our results (data not shown); thus, were not included as covariates in our models.
Secondary analyses were conducted to estimate the sex-specific proportion of prediabetes and probable diabetes mellitus classification based on HbA1c and OGTT. Prediabetes classification categories were defined as (1) normoglycemia by both HbA1C and OGTT (i.e., concordant classification); (2) prediabetes by HbA1c only (i.e., discordant classification), (3) prediabetes by OGTT only (i.e., discordant classification), and (4) prediabetes by both HbA1C and OGTT (i.e., concordant classification); participants with probable diabetes by HbA1c or OGTT were excluded from these analyses. Probable diabetes mellitus classification categories were defined as (1) non-diabetic by both HbA1c and OGTT (i.e., concordant classification, including normoglycemia or prediabetes by both HbA1c and OGTT); (2) probable diabetes by HbA1c only (i.e., discordant classification); (3) probable diabetes by OGTT only (i.e., discordant classification); and (4) probable diabetes by both HbA1C and OGTT (i.e., concordant classification). Finally, analogous adjusted multinomial logistic regression models were used to estimate the sex-specific association between iron and anemia status (with NANID as referent) and proportion of each prediabetes and probable diabetes classification based on HbA1c and OGTT (with normoglycemia/non-diabetic as referent).
Reported values were weighted, except sample size which is unweighted, to account for the disproportionate sample selection and adjust for any bias effects due to differential nonresponse.23 Weights were also trimmed to limit precision losses and calibrated to the 2010 US Census characteristics by age, sex, and Hispanic/Latino background in each field site’s target population. Analyses were performed using Stata Statistical Software Release 15 (Stata Corp LP, College Station, Texas).
RESULTS
Iron and Anemia Status Prevalence and Descriptive Characteristics of Target Population
The overall age-standardized (to the 2010 Census population) weighted prevalence of ID was 12.3%, IDA was 4.8%, and non-IDA was 4.3% (Table 1). Women had a higher age-standardized prevalence of ID and IDA and a lower prevalence of non-IDA than men. Among women, the mean age was 40 years (Table 2). Compared to NANID, women with ID were younger, women with IDA tended to have lower education, and women with non-IDA had lower annual household income levels. Women with IDA were more likely to have prediabetes and probable diabetes (per HbA1c) than women with NANID. Among men, the mean age was 38 years (Table 3). Compared with NANID, men with ID were younger, men with IDA tended to have lower education, and men with non-IDA had lower income levels. Men with IDA were more likely to have prediabetes and men with non-IDA were more likely to have probable diabetes (per HbA1c) than men with NANID.
Table 1.
Age-standardized survey-weighted prevalence of iron and anemia status categories among Hispanics/Latinos without self-reported diabetes mellitus, HCHS/SOL
| Iron and Anemia Status | |||||
|---|---|---|---|---|---|
| Category | Unweighted | NANID | ID | IDA | Non-IDA |
| n | % (SE) | % (SE) | % (SE) | % (SE) | |
| Overall | 12,151 | 78.6 (0.8) | 12.3 (0.8) | 4.8 (0.4) | 4.3 (0.4) |
| Sex | |||||
| Women | 7,247 | 72.7 (1.1) | 16.2 (1.1) | 7.2 (0.5) | 4.0 (0.4) |
| Men | 4,904 | 84.9 (1.3) | 7.8 (1.0) | 2.3 (0.6) | 5.1 (0.7) |
| Hispanic/Latino Background | |||||
| Cuban | 1,807 | 79.1 (1.8) | 13.4 (1.6) | 4.5 (0.9) | 3.0 (0.6) |
| Central American | 1,378 | 75.3 (2.9) | 13.6 (1.6) | 4.5 (1.1) | 6.7 (2.5) |
| Dominican | 1,083 | 77.0 (1.9) | 11.1 (1.4) | 4.9 (0.7) | 6.9 (1.3) |
| Mexican | 4,871 | 81.9 (1.4) | 10.2 (1.3) | 4.5 (0.4) | 3.5 (0.5) |
| Puerto Rican | 1,744 | 74.8 (2.5) | 13.7 (2.2) | 5.1 (0.7) | 6.4 (1.1) |
| South American | 870 | 80.4 (2.4) | 10.5 (1.9) | 4.7 (1.2) | 4.4 (1.2) |
| More than one | 398 | 82.5 (1.9) | 11.3 (1.8) | 4.1 (0.9) | 2.2 (0.8) |
Abbreviations: NANID= No anemia and no iron deficiency; ID= Iron deficiency; IDA= Iron deficiency anemia; Non-IDA= Non-iron deficiency anemia.
Notes: Prevalence estimates are weighted for the survey design and nonresponse and age-standardized to the 2010 U.S. Census population.
Table 2.
Characteristics of the target population of Hispanic/Latina women without self-reported diabetes mellitus: Overall and by iron and anemia status, HCHS/SOL
| Iron and Anemia Status | |||||
|---|---|---|---|---|---|
| Characteristics | Overall | NANID | ID | IDA | Non-IDA |
| n = 7,247 | n = 4,957 | n = 1,251 | n = 730 | n = 309 | |
| Demographic Characteristics | |||||
| Age, years*** | 39.6 ± 15.0 | 41.2 ± 15.8 | 35.9 ± 13.5 | 36.8 ± 11.3 | 39.2 ± 12.6 |
| Hispanic/Latino Background** | |||||
| Cuban | 18.5 | 20.3 | 16.3 | 13.8 | 11.0 |
| Central American | 7.9 | 7.8 | 9.1 | 6.9 | 7.4 |
| Dominican | 11.4 | 10.0 | 13.4 | 12.6 | 21.0 |
| Mexican | 39.4 | 38.9 | 39.4 | 42.4 | 39.0 |
| Puerto Rican | 13.2 | 13.0 | 12.7 | 14.7 | 15.7 |
| South American | 5.5 | 5.7 | 5.2 | 5.5 | 3.6 |
| More than one | 4.0 | 4.2 | 3.8 | 4.1 | 2.4 |
| Education*** | |||||
| Less than high school | 30.0 | 30.4 | 28.4 | 33.1 | 24.9 |
| High school graduate | 26.8 | 24.9 | 33.4 | 28.5 | 23.1 |
| More than high school | 43.2 | 44.8 | 38.3 | 38.3 | 51.9 |
| Annual Household Income | |||||
| <$20,000 | 43.9 | 43.3 | 42.8 | 47.1 | 50.6 |
| $20,000–$50,000 | 36.1 | 36.4 | 37.2 | 35.2 | 29.1 |
| >$50,000 | 9.5 | 9.7 | 8.4 | 7.8 | 13.9 |
| Not reported | 10.5 | 10.5 | 11.7 | 9.9 | 6.4 |
| Lifestyle and Clinical Characteristics | |||||
| Current smoking | 16.2 | 16.7 | 17.8 | 13.0 | 10.2 |
| Current alcohol use | 44.0 | 44.9 | 44.7 | 39.4 | 38.4 |
| Hypercholesterolemia | 30.5 | 31.6 | 29.3 | 26.7 | 28.2 |
| BMI, kg/m2* | 29.3 ± 6.7 | 29.0 ± 6.5 | 30.3 ± 7.1 | 30.1 ± 7.1 | 28.5 ± 6.1 |
| hs-CRP, mg/L* | 4.4 ± 8.4 | 3.6 ± 5.0 | 6.1 ± 8.4 | 4.6 ± 8.0 | 6.2 ± 24.5 |
| ALT, U/L*** | 21.8 ± 19.3 | 22.7 ± 20.9 | 21.2 ± 17.4 | 18.7 ± 13.9 | 18.5 ± 11.4 |
| AST, U/L*** | 21.6 ± 13.9 | 22.1 ± 15.0 | 20.8 ± 13.0 | 20.4 ± 9.6 | 20.4 ± 9.1 |
| Vitamin C intake, mg** | 88.3 ± 31.1 | 89.4 ± 31.2 | 86.8 ± 29.8 | 86.7 ± 32.9 | 84.1 ± 29.3 |
| Folate intake, mcg* | 357.0 ± 87.2 | 359.3 ± 88.7 | 358.2 ± 85.0 | 346.7 ± 74.4 | 344.4 ± 94.9 |
| Iron intake, mg* | 12.5 ± 3.1 | 12.6 ± 3.2 | 12.7 ± 3.0 | 12.1 ± 2.8 | 11.9 ± 3.1 |
| Vitamin B12 intake, mg* | 4.2 ± 1.4 | 4.2 ± 1.4 | 4.2 ± 1.3 | 4.1 ± 1.2 | 4.0 ± 1.4 |
| Alternate Healthy Eating Index* | 46.0 ± 7.4 | 46.4 ± 7.5 | 44.8 ± 6.9 | 45.6 ± 7.4 | 46.4 ± 7.3 |
| Glycemic Levels ** | |||||
| HbA1c, % | 5.4 ± 0.4 | 5.4 ± 0.5 | 5.4 ± 0.4 | 5.5 ± 0.4 | 5.5 ± 0.3 |
| HbA1c (Diabetes status categories)* | |||||
| Normoglycemia (<5.7%) | 73.1 | 72.9 | 75.4 | 67.4 | 78.6 |
| Prediabetes (5.7%−6.4%) | 24.9 | 25.1 | 22.7 | 29.7 | 20.7 |
| Probable diabetes (>6.4%) | 2.0 | 1.9 | 1.9 | 2.8 | 0.8 |
| Fasting plasma glucose, mg/dL** | 92.5 ± 9.9 | 92.7 ± 10.2 | 92.6 ± 9.4 | 91.9 ± 10.1 | 90.0 ± 7.7 |
| OGTT, mg/dL*** | 120.8 ± 41.7 | 122.7 ± 43.7 | 118.1 ± 38.9 | 117.3 ± 36.5 | 113.7 ± 34.1 |
| Iron and Anemia Status Indicators | |||||
| Hemoglobin, g/dL*** | 13.0 ± 1.3 | 13.4 ± 0.8 | 13.0 ± 0.9 | 10.7 ± 1.2 | 11.5 ± 0.6 |
| Serum iron, μg/dL*** | 81.7 ± 35.7 | 96.4 ± 29.7 | 53.3 ± 21.4 | 37.7 ± 19.8 | 84.4 ± 22.8 |
| Serum ferritin, μg/L*** | 66.8 ± 67.3 | 78.0 ± 69.9 | 47.2 ± 43.1 | 19.9 ± 31.4 | 65.6 ± 92.8 |
| Transferrin saturation, %*** | 25.5 ± 11.9 | 30.5 ± 10.0 | 15.8 ± 5.6 | 10.1 ± 5.5 | 27.1 ± 8.4 |
| Total iron binding capacity, μg/dL*** | 331.1 ± 53.6 | 320.8 ± 43.1 | 340.3 ± 58.2 | 386.9 ± 60.8 | 318.0 ± 48.3 |
Abbreviations: ALT= Alanine aminotransferase; AST= Aspartate aminotransferase; OGTT= Oral glucose tolerance test.
Notes: All values are weighted for the study design and nonresponse, except sample size which is unweighted. Values for categorical variables are given as percentage; values for continuous variables are expressed as mean ± standard deviation. Participants who declined to report their household income were included as a category to avoid deleting those observations.
p <0.05,
p <0.01,
p <0.001; overall differences across iron and anemia status categories were assessed using F-tests for continuous variables and chi-square tests for categorical variables.
Table 3.
Characteristics of the target population of Hispanic/Latino men without self-reported diabetes mellitus: Overall and by iron and anemia status, HCHS/SOL
| Iron and Anemia Status | |||||
|---|---|---|---|---|---|
| Characteristics | Overall | NANID | ID | IDA | Non-IDA |
| n = 4,904 | n = 4,195 | n = 377 | n = 92 | n = 240 | |
| Demographic Characteristics | |||||
| Age, years* | 38.1 ± 12.4 | 38.0 ± 12.4 | 36.4 ± 11.9 | 41.0 ± 13.8 | 41.0 ± 12.9 |
| Hispanic/Latino Background*** | |||||
| Cuban | 21.9 | 21.4 | 26.6 | 30.2 | 19.6 |
| Central American | 7.9 | 7.6 | 11.6 | 4.9 | 8.4 |
| Dominican | 8.0 | 7.8 | 8.2 | 10.8 | 10.3 |
| Mexican | 37.1 | 39.4 | 21.2 | 26.9 | 25.6 |
| Puerto Rican | 15.6 | 14.4 | 21.8 | 20.6 | 25.7 |
| South American | 5.2 | 5.2 | 3.2 | 2.1 | 7.6 |
| More than one | 4.4 | 4.2 | 7.3 | 4.4 | 2.9 |
| Education | |||||
| Less than high school | 29.9 | 29.9 | 24.3 | 37.7 | 36.8 |
| High school graduate | 31.4 | 31.6 | 32.2 | 31.7 | 27.4 |
| More than high school | 38.7 | 38.5 | 43.5 | 30.5 | 35.8 |
| Annual Household Income | |||||
| <$20,000 | 36.2 | 35.4 | 37.3 | 44.6 | 45.1 |
| $20,000–$50,000 | 40.3 | 40.8 | 37.1 | 30.1 | 38.4 |
| >$50,000 | 15.7 | 16.0 | 17.2 | 7.3 | 10.0 |
| Not reported | 7.8 | 7.7 | 8.4 | 12.0 | 6.5 |
| Lifestyle and Clinical Characteristics | |||||
| Current smoking** | 26.8 | 25.6 | 33.2 | 30.8 | 36.0 |
| Current alcohol use** | 65.5 | 66.3 | 66.2 | 46.8 | 55.8 |
| Hypercholesterolemia | 47.5 | 47.6 | 51.1 | 37.6 | 41.7 |
| BMI, kg/m2 | 28.7 ± 4.8 | 28.6 ± 4.6 | 28.8 ± 5.3 | 31.0 ± 7.8 | 28.7 ± 5.6 |
| hs-CRP, mg/L*** | 2.8 ± 4.0 | 2.4 ± 3.0 | 5.1 ± 7.8 | 7.5 ± 10.1 | 3.2 ± 3.9 |
| ALT, U/L*** | 34.5 ± 26.2 | 35.3 ± 27.1 | 30.9 ± 17.9 | 26.3 ± 15.4 | 30.4 ± 22.7 |
| AST, U/L | 27.6 ± 18.6 | 27.7 ± 18.8 | 25.5 ± 10.9 | 26.5 ± 13.2 | 29.2 ± 25.7 |
| Vitamin C intake, mg | 104.2 ± 32.1 | 104.6 ± 32.3 | 102.4 ± 28.6 | 94.6 ± 27.4 | 100.9 ± 33.5 |
| Folate intake, mcg | 469.4 ± 92.4 | 469.5 ± 92.3 | 475.6 ± 86.7 | 449.7 ± 95.1 | 464.1 ± 102.0 |
| Iron intake, mg | 16.9 ± 3.4 | 16.9 ± 3.4 | 17.2 ± 3.3 | 16.7 ± 3.1 | 16.6 ± 3.6 |
| Vitamin B12 intake, mg* | 5.7 ± 1.8 | 5.7 ± 1.8 | 5.8 ± 1.8 | 5.5 ± 1.0 | 5.3 ± 1.3 |
| Alternate Healthy Eating Index** | 48.2 ± 6.4 | 48.4 ± 6.4 | 46.5 ± 6.0 | 47.2 ± 5.8 | 47.3 ± 6.9 |
| Glycemic Levels | |||||
| HbA1c, %** | 5.4 ± 0.4 | 5.4 ± 0.4 | 5.4 ± 0.4 | 5.6 ± 0.4 | 5.5 ± 0.5 |
| HbA1c (Diabetes status categories)* | |||||
| Normoglycemia (<5.7%) | 73.4 | 73.8 | 77.5 | 57.1 | 64.3 |
| Prediabetes (5.7%−6.4%) | 25.0 | 24.8 | 21.3 | 39.8 | 31.8 |
| Probable diabetes (>6.4%) | 1.6 | 1.5 | 1.3 | 3.1 | 3.9 |
| Fasting plasma glucose, mg/dL | 96.4 ± 9.9 | 96.4 ± 8.6 | 96.2 ± 8.1 | 95.6 ± 8.3 | 96.2 ± 9.6 |
| OGTT, mg/dL | 112.5 ± 34.5 | 113.2 ± 34.4 | 104.7 ± 30.3 | 114.7 ± 38.9 | 114.1 ± 42.2 |
| Iron and Anemia Status Indicators | |||||
| Hemoglobin, g/dL*** | 15.1 ± 0.9 | 15.2 ± 0.8 | 15.0 ± 0.8 | 12.6 ± 1.0 | 13.0 ± 0.6 |
| Serum iron, μg/dL*** | 102.3 ± 30.5 | 107.9 ± 27.9 | 58.9 ± 21.5 | 46.6 ± 15.6 | 96.2 ± 25.5 |
| Serum ferritin, μg/L*** | 178.5 ± 129.8 | 184.7 ± 130.9 | 126.7 ± 80.5 | 85.3 ± 80.0 | 179.8 ± 165.0 |
| Transferrin saturation, %*** | 33.3 ± 10.3 | 35.1 ± 9.5 | 19.1 ± 6.3 | 14.6 ± 5.2 | 32.6 ± 10.6 |
| Total iron binding capacity, μg/dL | 311.6 ± 36.4 | 311.6 ± 33.9 | 310.6 ± 46.9 | 336.7 ± 61.4 | 303.2 ± 41.8 |
Abbreviations: ALT= Alanine aminotransferase; AST= Aspartate aminotransferase; OGTT= Oral glucose tolerance test.
Notes: All values are weighted for the study design and nonresponse, except sample size which is unweighted. Values for categorical variables are given as percentage; values for continuous variables are expressed as mean ± standard deviation. Participants who declined to report their household income were included as a category to avoid deleting those observations.
p <0.05,
p <0.01,
p <0.001; overall differences across iron and anemia status categories were assessed using F-tests for continuous variables and chi-square tests for categorical variables.
Sex-specific Association between Iron-Anemia Status and HbA1c
Women.
Compared to women with NANID and normoglycemia, women with ID had significantly higher odds of prediabetes adjusting for age (Model 1; OR= 1.39, 95% CI: 1.08, 1.79); however, this association was no longer significant upon additional adjustment for other demographic, lifestyle, and clinical covariates (Model 2) (Table 4). There was no association of ID with probable diabetes in Model 1 or Model 2. Contrastingly, women with IDA (versus NANID and normoglycemia) had significantly higher odds of prediabetes (OR= 2.16, 95% CI: 1.65, 2.84) and probable diabetes (OR= 3.37, 95% CI: 1.63, 6.94) in Model 1. These associations persisted upon additional adjustments in Model 2; namely, women with IDA had higher odds of prediabetes (OR= 2.18, 95% CI: 1.64, 2.89) and probable diabetes (OR= 3.59, 95% CI: 1.62, 7.99). Finally, there was no association of non-IDA with prediabetes or probable diabetes, regardless of model-based adjustments.
Table 4.
Sex-specific adjusted odds ratios (OR) and 95% confidence intervals (CI) for the association between iron and anemia status categories and HbA1c categories among Hispanics/Latinos without self-reported diabetes mellitus, HCHS/SOL
| Iron and Anemia Status | Women (n = 7,247) | Men (n = 4,904) | ||||
|---|---|---|---|---|---|---|
| Normoglycemia by HbA1c | Prediabetes by HbA1c OR (95% CI) |
Probable Diabetes by HbA1c OR (95% CI) |
Normoglycemia by HbA1c | Prediabetes by HbA1c OR (95% CI) |
Probable Diabetes by HbA1c OR (95% CI) |
|
| NANID | ||||||
| (Ref) | -- | -- | -- | -- | -- | -- |
| ID | ||||||
| Model 1 | 1.00 | 1.39 (1.08, 1.79) | 1.73 (0.98, 3.04) | 1.00 | 0.88 (0.65, 1.18) | 0.88 (0.37, 2.10) |
| Model 2 | 1.00 | 1.23 (0.95, 1.58) | 1.34 (0.80, 2.26) | 1.00 | 0.73 (0.53, 1.00) | 0.72 (0.25, 2.02) |
| IDA | ||||||
| Model 1 | 1.00 | 2.16 (1.65, 2.84) | 3.37 (1.63, 6.94) | 1.00 | 1.93 (1.01, 3.68) | 2.39 (0.85, 6.67) |
| Model 2 | 1.00 | 2.18 (1.64, 2.89) | 3.59 (1.62, 7.99) | 1.00 | 1.52 (0.74, 3.12) | 1.35 (0.29, 6.30) |
| Non-IDA | ||||||
| Model 1 | 1.00 | 0.90 (0.58, 1.39) | 0.47 (0.16, 1.35) | 1.00 | 1.32 (0.87, 2.00) | 2.64 (1.24, 5.62) |
| Model 2 | 1.00 | 0.93 (0.59, 1.48) | 0.40 (0.10, 1.60) | 1.00 | 1.39 (0.89, 2.19) | 2.97 (1.13, 7.78) |
Abbreviations: NANID= No anemia and no iron deficiency; ID= Iron deficiency; IDA= Iron deficiency anemia; Non-IDA= Non-iron deficiency anemia.
Notes: Statistically significant associations at p <0.05 are shown in boldface type. HbA1c levels were classified into the following groups: normoglycemia (<5.7%), prediabetes (5.7%−6.4%), or probable diabetes (≥6.5%). The term “probable diabetes” is used because results should be confirmed with repeated testing to diagnose diabetes.
Model 1 adjusted for age only.
Model 2 adjusted for age + Hispanic/Latino background, education, annual household income, smoking status, alcohol intake, hypercholesterolemia, BMI, hs-CRP, alanine aminotransferase, aspartate aminotransferase, vitamin C intake, dietary folate intake, dietary iron intake, and vitamin B12 intake.
Men.
In men, there was no association of ID with prediabetes or probable diabetes despite adjustments (Table 4). Compared to men with NANID and normoglycemia, men with IDA had higher odds of prediabetes in Model 1 (OR= 1.93, 95% CI: 1.0.1, 3.68) but this association was no longer significant in Model 2. There was no association between IDA and probable diabetes regardless of adjustments. Contrastingly, we observed that men with non-IDA (versus NANID and normoglycemia) had higher odds of probable diabetes (OR= 2.64, 95% CI: 1.24, 5.62) in Model 1 and this association persisted in Model 2 (OR= 3.08, 95% CI: 1.13, 7.78). There was no association of non-IDA with prediabetes regardless of adjustments.
Secondary Analyses
Sex-specific proportion of prediabetes and probable diabetes mellitus classification based on HbA1c and OGTT across iron and anemia status.
In women and men with ID we observed that a higher proportion was classified with prediabetes using HbA1c only compared to OGTT; similar trends were observed among women and men with IDA and non-IDA. Contrastingly, among women and men, there was no difference in probable diabetes mellitus classification using HbA1c versus OGTT across categories of iron and anemia status (Table 5). However, among women and men with IDA we observed that a higher proportion was classified with probable diabetes mellitus using HbA1c only versus OGTT only. A similar trend was observed among men with ID and non-IDA, namely, a higher proportion was classified with probable diabetes using HbA1c only compared to using OGTT only.
Table 5.
Sex-specific proportion of prediabetes and probable diabetes classification according to HbA1c and OGTT levels stratified by iron and anemia status among Hispanics/Latinos without self-reported diabetes, HCHS/SOL
| Prediabetes Classification * | Overall % | Normoglycemia by HbA1c + OGTT % | Prediabetes by HbA1c only % | Prediabetes by OGTT only % | Prediabetes by HbA1c + OGTT % | P-value |
|---|---|---|---|---|---|---|
| Women, % | n = 6,725 | n = 3,905 | n = 1,243 | n = 751 | n = 826 | 0.007 |
| NANID | 65.8 | 65.0 | 63.5 | 71.2 | 69.5 | |
| ID | 18.9 | 20.2 | 17.6 | 15.4 | 15.5 | |
| IDA | 10.4 | 9.4 | 14.2 | 10.0 | 11.3 | |
| Non-IDA | 4.9 | 5.4 | 4.7 | 3.4 | 3.7 | |
| Men, % | n = 4,615 | n = 2,878 | n = 933 | n = 387 | n = 425 | 0.002 |
| NANID | 85.6 | 85.5 | 82.6 | 90.2 | 88.1 | |
| ID | 8.4 | 9.2 | 7.8 | 4.7 | 6.2 | |
| IDA | 1.7 | 1.5 | 3.4 | 0.4 | 1.8 | |
| Non-IDA | 4.3 | 3.8 | 6.3 | 4.7 | 4.0 | |
| Probable Diabetes Classification † | Overall % | Non-diabetic by HbA1c + OGTT % | Probable Diabetes by HbA1c only % | Probable Diabetes by OGTT only % | Probable Diabetes by HbA1c + OGTT % | |
| Women, % | n = 7,247 | n = 6,725 | n = 72 | n = 320 | n = 130 | 0.103 |
| NANID | 66.1 | 65.8 | 65.2 | 74.1 | 64.8 | |
| ID | 18.9 | 18.9 | 12.5 | 18.9 | 21.2 | |
| IDA | 10.2 | 10.4 | 19.6 | 3.5 | 12.5 | |
| Non-IDA | 4.8 | 4.9 | 2.6 | 3.5 | 1.5 | |
| Men, % | n = 4,904 | n = 4,615 | n = 47 | n = 166 | n = 76 | 0.129 |
| NANID | 85.5 | 85.6 | 72.5 | 85.2 | 83.3 | |
| ID | 8.3 | 8.4 | 11.4 | 6.3 | 3.5 | |
| IDA | 1.8 | 1.7 | 4.1 | 2.1 | 3.0 | |
| Non-IDA | 4.5 | 4.3 | 12.0 | 6.3 | 10.3 |
Abbreviations: NANID= No anemia and no iron deficiency; ID= Iron deficiency; IDA= Iron deficiency anemia; Non-IDA= Non-iron deficiency anemia; OGTT= Oral glucose tolerance test.
Notes: HbA1c levels were classified into the following groups: normoglycemia (<5.7%), prediabetes (5.7%−6.4%), or probable diabetes (≥6.5%). OGTT levels were classified into the following groups: normoglycemia (<140 mg/dL); prediabetes (140–199 mg/dL), or probable diabetes (≥200 mg/dL). The term “probable diabetes” is used because results should be confirmed with repeated testing to diagnose diabetes.
Prediabetes classification categories were defined as: normoglycemia by both HbA1C and OGTT (i.e., concordant classification); prediabetes by HbA1c only (i.e., discordant classification), prediabetes by OGTT only (i.e., discordant classification), and prediabetes by both HbA1C and OGTT (i.e., concordant classification); participants with probable diabetes (according to HbA1c and/or OGTT) were excluded from these analyses.
Probable diabetes classification categories were defined as: non-diabetic including normoglycemia by both HbA1c and OGTT as well as prediabetes by both HbA1c and OGTT (i.e., concordant classification); probable diabetes by HbA1c only (i.e., discordant classification), probable diabetes by OGTT only (i.e., discordant classification); probable diabetes by both HbA1C and OGTT (i.e., concordant classification).
Sex-specific association between iron-anemia status and proportion of prediabetes and probable diabetes mellitus classification based on HbA1c and OGTT.
Women with IDA (versus NANID and normoglycemia) had higher odds of prediabetes classification according to HbA1c levels only (OR= 2.35, 95% CI: 1.68–3.28) in Model 1; this association persisted in Model 2 (OR= 2.42, 95% CI: 1.75–3.35). Moreover, women with IDA had higher odds of prediabetes classification according to both HbA1c and OGTT levels (OR= 2.29, 95% CI: 1.57–3.36) in Model 1 and Model 2 (OR= 2.20, 95% CI: 1.44–3.35). There was no association of ID or non-IDA with odds of prediabetes classification in women. Among men, no association was observed of iron and anemia status with the proportion of prediabetes classification (Table 6).
Table 6.
Sex-specific adjusted odds ratios (OR) and 95% confidence intervals (CI) for the association between iron-anemia status categories and proportion of prediabetes and probable diabetes classification according to HbA1c and OGTT levels among Hispanics/Latinos without self-reported diabetes, HCHS/SOL
| Women (n = 6,725) | Men (n = 4,615) | |||||||
|---|---|---|---|---|---|---|---|---|
| Prediabetes Classification * | Normoglycemia by HbA1c + OGTT | Prediabetes by HbA1c only OR (95% CI) |
Prediabetes by OGTT only OR (95% CI) |
Prediabetes by HbA1c + OGTT OR (95% CI) |
Normoglycemia by HbA1c + OGTT | Prediabetes by HbA1c only OR (95% CI) |
Prediabetes by OGTT only OR (95% CI) |
Prediabetes by HbA1c + OGTT OR (95% CI) |
| NANID | ||||||||
| (Ref) | -- | -- | -- | -- | -- | -- | -- | -- |
| ID | ||||||||
| Model 1 | 1.00 | 1.35 (0.97, 1.87) | 0.88 (0.66, 1.19) | 1.29 (0.94, 1.77) | 1.00 | 0.92 (0.65, 1.29) | 0.52 (0.25, 1.05) | 0.70 (0.39, 1.11) |
| Model 2 | 1.00 | 1.20 (0.87, 1.64) | 0.78 (0.58, 1.05) | 1.03 (0.75, 1.42) | 1.00 | 0.77 (0.53, 1.11) | 0.56 (0.30, 1.06) | 0.59 (0.32, 1.09) |
| IDA | ||||||||
| Model 1 | 1.00 | 2.35 (1.68, 3.28) | 1.18 (0.80, 1.75) | 2.29 (1.57, 3.36) | 1.00 | 2.18 (1.04, 4.57) | 0.21 (0.05, 0.93) | 1.01 (0.39, 2.59) |
| Model 2 | 1.00 | 2.42 (1.75, 3.35) | 1.15 (0.77, 1.72) | 2.20 (1.44, 3.35) | 1.00 | 1.64 (0.77, 3.48) | 0.22 (0.05, 1.01) | 0.84 (0.26, 2.71) |
| Non-IDA | ||||||||
| Model 1 | 1.00 | 0.98 (0.58, 1.66) | 0.60 (0.29, 1.23) | 0.78 (0.43, 1.43) | 1.00 | 1.50 (0.93, 2.42) | 1.02 (0.57, 1.83) | 0.84 (0.43, 1.67) |
| Model 2 | 1.00 | 1.09 (0.66, 1.80) | 0.60 (0.29, 1.25) | 0.75 (0.36, 1.56) | 1.00 | 1.58 (0.95, 2.63) | 1.23 (0.65, 2.33) | 0.90 (0.44, 1.84) |
| Women (n = 7,247) | Men (n = 4,904) | |||||||
| Probable Diabetes Classification † | Non-diabetic by HbA1c + OGTT | Probable Diabetes by HbA1c only OR (95% CI) |
Probable Diabetes by OGTT only OR (95% CI) |
Probable Diabetes by HbA1c + OGTT OR (95% CI) |
Non-diabetic by HbA1c + OGTT | Probable Diabetes by HbA1c only OR (95% CI) |
Probable Diabetes by OGTT only OR (95% CI) |
Probable Diabetes by HbA1c + OGTT OR (95% CI) |
| NANID | ||||||||
| (Ref) | -- | -- | -- | -- | -- | -- | -- | -- |
| ID | ||||||||
| Model 1 | 1.00 | 1.08 (0.40, 2.89) | 1.39 (0.87, 2.23) | 1.73 (0.90, 3.34) | 1.00 | 1.73 (0.54, 5.54) | 0.83 (0.34, 2.00) | 0.47 (0.12, 1.90) |
| Model 2 | 1.00 | 0.81 (0.29, 2.24) | 1.18 (0.74, 1.90) | 1.43 (0.78, 2.63) | 1.00 | 1.78 (0.50, 6.31) | 0.86 (0.33, 2.20) | 0.39 (0.08, 2.01) |
| IDA | ||||||||
| Model 1 | 1.00 | 3.60 (1.60, 8.08) | 0.53 (0.28, 1.01) | 2.06 (0.72, 5.93) | 1.00 | 2.34 (0.62, 8.89) | 0.96 (0.29, 3.17) | 1.40 (0.34, 5.81) |
| Model 2 | 1.00 | 3.07 (1.28, 7.35) | 0.51 (0.27, 0.98) | 2.09 (0.64, 6.82) | 1.00 | 1.74 (0.27, 11.33) | 1.01 (0.27, 3.76) | 0.83 (0.13, 5.34) |
| Non-IDA | ||||||||
| Model 1 | 1.00 | 0.73 (0.21, 2.54) | 0.82 (0.28, 2.39) | 0.39 (0.08, 1.77) | 1.00 | 2.90 (1.04, 8.07) | 1.27 (0.59, 2.73) | 2.12 (0.80, 5.63) |
| Model 2 | 1.00 | 0.52 (0.06, 4.31) | 0.80 (0.24, 2.67) | 0.35 (0.06, 1.93) | 1.00 | 2.87 (0.69, 11.94) | 1.51 (0.61, 3.71) | 2.39 (0.80, 7.13) |
NANID= No anemia and no iron deficiency; ID= Iron deficiency; IDA= Iron deficiency anemia; Non-Non-IDA= Non-iron deficiency anemia.
Notes: Statistically significant associations at p <0.05 are shown in boldface type. HbA1c levels were classified into the following groups: normoglycemia (<5.7%), prediabetes (5.7%−6.4%), or probable diabetes (≥6.5%). OGTT levels were classified into the following groups: normoglycemia (<140 mg/dL); prediabetes (140–199 mg/dL), or probable diabetes (≥200 mg/dL). The term “probable diabetes” is used because results should be confirmed with repeated testing to diagnose diabetes.
Prediabetes classification categories were defined as: normoglycemia by both HbA1C and OGTT (i.e., concordant classification); prediabetes by HbA1c only (i.e., discordant classification), prediabetes by OGTT only (i.e., discordant classification), and prediabetes by both HbA1C and OGTT (i.e., concordant classification); participants with probable diabetes (according to HbA1c and/or OGTT) were excluded from these analyses.
Probable diabetes classification categories were defined as: non-diabetic including normoglycemia by both HbA1c and OGTT as well as prediabetes by both HbA1c and OGTT (i.e., concordant classification); probable diabetes by HbA1c only (i.e., discordant classification), probable diabetes by OGTT only (i.e., discordant classification); probable diabetes by both HbA1C and OGTT (i.e., concordant classification).
Model 1 adjusted for age only.
Model 2 adjusted for age + Hispanic/Latino background, education, annual household income, smoking status, alcohol intake, hypercholesterolemia, BMI, hs-CRP, alanine aminotransferase, aspartate aminotransferase, vitamin C intake, dietary folate intake, dietary iron intake, and vitamin B12 intake.
A similar pattern was observed in the association of iron and anemia status categories with probable diabetes mellitus classification. Specifically, women with IDA (versus NANID and non-diabetic) had higher odds of probable diabetes mellitus classification using HbA1c levels only in Model 1 (OR= 3.60, 95% CI: 1.60–8.08); this association persisted in Model 2 after additional adjustments (OR= 3.07, 95% CI: 1.28–7.35). Women with IDA had higher odds of probable diabetes mellitus classification per OGTT only in Model 2. There was no association of ID or Non-IDA with odds of probable diabetes classification in women. Moreover, among men, no association was observed of iron and anemia status with odds of probable diabetes classification.
DISCUSSION
In our study of US Hispanics/Latinos without self-reported diabetes mellitus, the age-standardized prevalence of ID, IDA, and non-IDA (i.e., 12.3%, 4.8%, and 4.3%, respectively) were each higher than the previously reported prevalence for the US general population (i.e., 4.4%, 2.3%, and 3.2%, respectively).11 The sex-specific age-standardized prevalence of ID reported in our study (i.e., 16.2% among women and 7.8% among men) was also higher than previously reported for the US general population (i.e., 13.7% among women and 1.6% among men).19 The higher prevalence of ID, IDA, and non-IDA found in our study compared to previous studies11,19 could be explained, at least partially, by differences in the definitions of iron and anemia status and/or populations included across studies.
Our main findings support previous research9,14,15 suggesting that iron and anemia status is a crucial factor to consider when using HbA1c as the sole diagnostic test for prediabetes or diabetes mellitus. Indeed, we observed that women with IDA (versus NANID) had greater odds of prediabetes and probable diabetes mellitus using HbA1c, and a similar trend was observed in men with IDA (although not statistically significant). Contrastingly, non-IDA (versus NANID) was related with higher odds of probable diabetes using HbA1c in men but not in women (and estimates in women were in the opposite direction than in men). Post-hoc analyses with HbA1c defined as a continuous variable (Table 7) confirmed our main finding on the association between IDA and higher HbA1c levels in women. Further, we observed that non-IDA was related to lower HbA1c levels in women but higher HbA1c levels in men—although these relations did not reach statistical significance. Previous clinic-based studies suggest that the relations of ID, IDA, and non-IDA with HbA1c are mixed,14,15 partly because ID and non-IDA have been proposed to have two different disease mechanisms.34 ID may artificially increase HbA1c levels by producing changes in the shape conformation of hemoglobin, stimulating glycation of the terminal valine, or reducing erythrocyte turnover allowing more time for glycation to occur; whereas non-IDA could contribute to lower HbA1c levels due to an increased erythrocyte turnover.35 Additionally, a significant decrease in HbA1c levels after the administration of iron supplementation in patients with IDA has been reported.36 Differences in the relationship between IDA and non-IDA on HbA1c levels may also be explained by differences in hemoglobin-glucose affinity associated with changes in the three-dimensional configuration of hemoglobin, different hemoglobin variants or genetic determinants of hemoglobin glycation, medications, or comorbid factors37 that were not evaluated in the present study. The sex differences observed in our study on the associations of IDA and non-IDA with HbA1c could reflect different biological mechanisms, including true iron deficiency due to blood loss and/or nutrition deficiencies in women, fat distribution or other biochemical aspects of adipose tissue metabolism in men,38 or may reflect sex differences in the prevalence of iron and anemia deficiencies.
Table 7.
Sex-specific association between iron and anemia status categories and HbA1c as continuous among Hispanics/Latinos without self-reported diabetes, HCHS/SOL
| Iron and Anemia Status | Women (n = 7,247) | Men (n = 4,904) |
|---|---|---|
| HbA1c B (95% CI) |
HbA1c B (95% CI) |
|
| No anemia and no iron deficiency (Ref) | -- | -- |
| Iron deficiency | 0.03 (−0.01, 0.06) | −0.05 (−0.10, 0.00) |
| Iron deficiency anemia | 0.12 (0.06, 0.17) | 0.08 (−0.08, 0.23) |
| Non-iron deficiency anemia | −0.03 (−0.10, 0.04) | 0.08 (−0.01, 0.18) |
Notes: Statistically significant associations at p <0.05 are shown in boldface type.
Final model adjusted for age + Hispanic/Latino background, education, annual household income, smoking status, alcohol intake, hypercholesterolemia, BMI, hs-CRP, alanine aminotransferase, aspartate aminotransferase, vitamin C intake, dietary folate intake, dietary iron intake, and vitamin B12 intake.
Our findings have important implications for estimating the prevalence of prediabetes and probable diabetes mellitus among US Hispanics/Latinos, a population with a relatively high prevalence of ID, IDA, and non-IDA, and in settings where HbA1c is used for glycemic assessment. Results from our secondary analysis suggest that the sole use of HbA1c may be overestimating the prevalence of prediabetes in women and men with ID, IDA, and non-IDA and overestimating the prevalence of probable diabetes mellitus in women with IDA and men with ID, IDA, and non-IDA. Moreover, adjusted analyses showed that women with IDA had higher odds of prediabetes and diabetes mellitus classification per HbA1c levels (but not OGTT) ‒a relation that persisted despite model-based adjustments. Other studies in clinic-based samples similarly have shown higher mean HbA1c levels in individuals with IDA versus NANID.16
Strengths of this study include the large sample size with diverse Hispanic/Latino backgrounds, sex-specific analyses, mutually exclusive iron-anemia status categories, and the use of OGTT. Limitations of our study include the cross-sectional nature of the analysis (precluding us from distinguishing temporal sequence) and lower precision in some of estimates due to smaller sample size of probable diabetes in the HBA1c only category. Other limitations are that the HCHS/SOL did not collect data on events that may affect the bone marrow or formation of blood cell lineages (e.g., blood transfusion, hemoglobinopathy, and coagulation disorders), did not assess hemoglobin variants, and participants receiving cancer treatment were not excluded from the study. Finally, the generalizability of our findings to populations with diabetes mellitus may be limited.
CONCLUSION
In sum, this study confirms a public health problem, namely, the high prevalence of ID, IDA, and non-NIDA in Hispanics/Latinos without self-reported diabetes mellitus. Findings add to the growing body of knowledge proposing that iron and anemia status should be considered in the interpretation of HbA1c levels, mainly, when this test is used as a screening or clinical follow-up tool in the care of individuals at diabetes mellitus risk. Accounting for iron-anemia status may contribute to assess the prevalence of prediabetes and probable mellitus more accurately in US Hispanic/Latino populations. A holistic, or integrative care approach, would be appropriate and beneficial in identifying the potential cause and recommending treatment of ID, IDA, and non-NIDA while assessing prediabetes or probable diabetes mellitus status in US Hispanics/Latinos.
Highlights.
The age-standardized prevalence of iron deficiency (ID), iron deficiency anemia (IDA), and non-iron deficiency anemia (non-IDA) among US-based Hispanics/Latinos without self-reported diabetes mellitus is high and varies by sex.
Compared to women with normoglycemia as well as no anemia and no iron deficiency (NANID), women with IDA had higher odds of prediabetes and probable diabetes mellitus (defined based on HbA1c levels).
Compared to men with normoglycemia and NANID, men with non-IDA had higher odds of probable diabetes mellitus (defined based on HbA1c levels).
Findings suggest that iron and anemia status should be considered when interpreting elevated HbA1c levels among US Hispanics/Latinos without self-reported diabetes mellitus.
Clinical Relevance.
Iron-anemia status should be considered in the interpretation of HbA1c, mainly, when it is used as a screening or clinical follow-up tool in the care of Hispanics/Latinos at diabetes mellitus risk. Accounting for iron-anemia status may contribute to assess the prevalence of prediabetes and probable mellitus more accurately in Hispanics/Latinos.
Acknowledgements:
The authors would like to sincerely thank the HCHS/SOL participants and staff members. The authors thank the staff members across Field Centers, whose dedication and ceaseless energy made the recruitment and baseline examination a success; and the over 16,000 participants who believed in making a difference, ¡Gracias!
Sources of Support:
The baseline examination of the Hispanic Community Health Study/Study of Latinos was carried out as a collaborative study supported by contracts from the National Heart, Lung, and Blood Institute (NHLBI) to the University of North Carolina (N01-HC65233), University of Miami (N01-HC65234), Albert Einstein College of Medicine (N01-HC65235), Northwestern University (N01-HC65236), and San Diego State University (N01-HC65237). The following NIH Institutes/Offices collaborated and co-funded the first phase of the study: the National Institute on Minority Health and Health Disparities, the National Institute on Deafness and Other Communication Disorders, the National Institute of Dental and Craniofacial Research, the National Institute of Diabetes and Digestive and Kidney Diseases, the National Institute of Neurological Disorders and Stroke, and the NIH Office of Dietary Supplements. Dr. Estrella was partially funded by a NHBLI Research Supplement to Promote Diversity in Health-Related Research 75N92019D00012 to the Hispanic Community Health Study/Study of Latinos (HCHS/SOL) Chicago Field Center.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
REFERENCES
- 1.Saaddine JB, Fagot-Campagna A, Rolka D, et al. Distribution of HbA1c levels for children and young adults in the U.S.: Third National Health and Nutrition Examination Survey. Diabetes Care. 2002;25(8):1326–1330. doi: 10.2337/diacare.25.8.1326 [DOI] [PubMed] [Google Scholar]
- 2.Menke A, Rust KF, Savage PJ, Cowie CC. Hemoglobin A1c, fasting plasma glucose, and 2-hour plasma glucose distributions in U.S. population subgroups: NHANES 2005–2010. Ann Epidemiol. 2014;24(2):83–89. doi: 10.1016/j.annepidem.2013.10.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Davidson MB, Schriger DL. Effect of age and race/ethnicity on HbA1c levels in people without known diabetes mellitus: Implications for the diagnosis of diabetes. Diabetes Res Clin Pract. 2010;87(3):415–421. doi: 10.1016/j.diabres.2009.12.013 [DOI] [PubMed] [Google Scholar]
- 4.Larissa Avilés-Santa M, Schneiderman N, Savage PJ, et al. Identifying probable diabetes mellitus among hispanics/latinos from four U.S. cities: Findings from the hispanic community health study/study of latinos. Endocr Pract. 2016;22(10):1151–1160. doi: 10.4158/EP151144.OR [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Avilés-Santa ML, Hsu LL, Arredondo M, et al. Differences in hemoglobin a1cbetween hispanics/latinos and non-hispanic whites:An analysis of the hispanic community health study/study of latinos and the 2007–2012 national health and nutrition examination survey. Diabetes Care. 2016;39(6):1010–1017. doi: 10.2337/dc15-2579 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Little RR, Rohlfing C, Sacks DB. The National Glycohemoglobin Standardization Program: Over 20 Years of Improving Hemoglobin A1c Measurement. Clin Chem. 2019;65(7):839–848. doi: 10.1373/CLINCHEM.2018.296962 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Theurl I, Aigner E, Theurl M, et al. Regulation of iron homeostasis in anemia of chronic disease and iron deficiency anemia: Diagnostic and therapeutic implications. Blood. 2009;113(21):5277–5286. doi: 10.1182/blood-2008-12-195651 [DOI] [PubMed] [Google Scholar]
- 8.Shaw JG, Friedman JF. Iron deficiency anemia: Focus on infectious diseases in lesser developed countries. Anemia. 2011;2011. doi: 10.1155/2011/260380 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Attard SM, Herring AH, Wang H, et al. Implications of iron deficiency/anemia on the classification of diabetes using HbA1c. Nutr Diabetes. 2015;5(6). doi: 10.1038/nutd.2015.16 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Kim C, Bullard KMK, Herman WH, Beckles GL. Association between iron deficiency and A1C levels among adults without diabetes in the National Health And Nutrition Examination Survey, 1999–2006. Diabetes Care. 2010;33(4):780–785. doi: 10.2337/dc09-0836 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Ford ES, Cowie CC, Li C, Handelsman Y, Bloomgarden ZT. Iron-deficiency anemia, non-iron-deficiency anemia and HbA1c among adults in the US. J Diabetes. 2011;3(1):67–73. doi: 10.1111/j.1753-0407.2010.00100.x [DOI] [PubMed] [Google Scholar]
- 12.Hong JW, Ku CR, Noh JH, Ko KS, Rhee BD, Kim DJ. Association between the Presence of Iron Deficiency Anemia and Hemoglobin A1c in Korean Adults. Med (United States). 2015;94(20). doi: 10.1097/MD.0000000000000825 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Ford ES, Cowie CC, Li C, Handelsman Y, Bloomgarden ZT. Iron-deficiency anemia, non-iron-deficiency anemia and HbA1c among adults in the US. J Diabetes. 2011;3(1):67–73. doi: 10.1111/j.1753-0407.2010.00100.x [DOI] [PubMed] [Google Scholar]
- 14.English E, Idris I, Smith G, Dhatariya K, Kilpatrick ES, John WG. The effect of anaemia and abnormalities of erythrocyte indices on HbA1c analysis: a systematic review. Diabetologia. 2015;58(7):1409–1421. doi: 10.1007/s00125-015-3599-3 [DOI] [PubMed] [Google Scholar]
- 15.Cavagnolli G, Pimentel AL, Freitas PAC, Gross JL, Camargo JL. Factors affecting A1C in non-diabetic individuals: Review and meta-analysis. Clin Chim Acta. 2015;445:107–114. doi: 10.1016/j.cca.2015.03.024 [DOI] [PubMed] [Google Scholar]
- 16.Sinha N, Mishra TK, Singh T, Gupta N. Effect of iron deficiency anemia on hemoglobin A1c levels. Ann Lab Med. 2012;32(1):17–22. doi: 10.3343/alm.2012.32.1.17 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Gang L, Lifang W. Association of the elevated red blood cell distribution width with the risk of developing diabetes mellitus. Intern Med. 2016;55(15):1959–1965. doi: 10.2169/internalmedicine.55.5956 [DOI] [PubMed] [Google Scholar]
- 18.Simmons D, Hlaing T. Interpretation of HbA1c: Association with mean cell volume and haemoglobin concentration. Diabet Med. 2014;31(11):1387–1392. doi: 10.1111/dme.12518 [DOI] [PubMed] [Google Scholar]
- 19.Le CHH. The Prevalence of Anemia and Moderate-Severe Anemia in the US Population (NHANES 2003–2012). Collins JF, ed. PLoS One. 2016;11(11):e0166635. doi: 10.1371/journal.pone.0166635 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Balarajan Y, Ramakrishnan U, Özaltin E, Shankar AH, Subramanian SV. Anaemia in low-income and middle-income countries. Lancet. 2011;378(9809):2123–2135. doi: 10.1016/S0140-6736(10)62304-5 [DOI] [PubMed] [Google Scholar]
- 21.Avilés-Santa ML, Hsu LL, Arredondo M, et al. Differences in hemoglobin a1cbetween hispanics/latinos and non-hispanic whites:An analysis of the hispanic community health study/study of latinos and the 2007–2012 national health and nutrition examination survey. Diabetes Care. 2016;39(6):1010–1017. doi: 10.2337/dc15-2579 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Sorlie PD, Avilés-Santa LM, Wassertheil-Smoller S, et al. Design and implementation of the Hispanic Community Health Study/Study of Latinos. Ann Epidemiol. 2010;20(8):629–641. doi: 10.1016/j.annepidem.2010.03.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.LaVange LM, Kalsbeek WD, Sorlie PD, et al. Sample Design and Cohort Selection in the Hispanic Community Health Study/Study of Latinos. Ann Epidemiol. 2010;20(8):642–649. doi: 10.1016/j.annepidem.2010.05.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Thyagarajan B, Howard AG, Durazo-Arvizu R, et al. Analytical and biological variability in biomarker measurement in the Hispanic Community Health Study/Study of Latinos. Clin Chim Acta. 2016;463:129. doi: 10.1016/J.CCA.2016.10.019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Association AD. Classification and diagnosis of diabetes: Standards of medical care in diabetesd2019. Diabetes Care. 2019;42(Supplement 1):S13–S28. doi: 10.2337/dc19-S002 [DOI] [PubMed] [Google Scholar]
- 26.Saraco M, Hill MI, Petersen MP, Robinson S, Mandelson J, Gabbay RA. Professional practice committee: Standards of medical care in diabetes-2021. Diabetes Care. 2021;44(Supplement 1):S3. doi: 10.2337/dc21-SPPC [DOI] [PubMed] [Google Scholar]
- 27.Beutler E, Waalen J. The definition of anemia: What is the lower limit of normal of the blood hemoglobin concentration? Blood. 2006;107(5):1747–1750. doi: 10.1182/blood-2005-07-3046 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Hardikar PS, Joshi SM, Bhat DS, et al. Spuriously high prevalence of prediabetes diagnosed by HbA 1c in young Indians partly explained by hematological factors and iron deficiency anemia. Diabetes Care. 2012;35(4):797–802. doi: 10.2337/dc11-1321 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Cheung CL, Cheung TT, Lam KSL, Cheung BMY. High ferritin and low transferrin saturation are associated with pre-diabetes among a national representative sample of U.S. adults. Clin Nutr. 2013;32(6):1055–1060. doi: 10.1016/j.clnu.2012.11.024 [DOI] [PubMed] [Google Scholar]
- 30.Veeranna V, Zalawadiya SK, Panaich SS, Ramesh K, Afonso L. The association of red cell distribution width with glycated hemoglobin among healthy adults without diabetes mellitus. Cardiol. 2012;122(2):129–132. doi: 10.1159/000339225 [DOI] [PubMed] [Google Scholar]
- 31.Kim C, Bullard KMK, Herman WH, Beckles GL. Association between iron deficiency and A1C levels among adults without diabetes in the National Health And Nutrition Examination Survey, 1999–2006. Diabetes Care. 2010;33(4):780–785. doi: 10.2337/dc09-0836 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Han Ling-ling, Wang Yu-xia, Li Jia, Zhang Xiao-lei, Bian Che, Wang He, Shufa Du LS. Gender differences in associations of serum ferritin and diabetes, metabolic syndrome, and obesity in the China Health and Nutrition Survey. Mol Nutr Food Res. 2014;58(11):2189–2195. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Chiuve SE, Fung TT, Rimm EB, et al. Alternative dietary indices both strongly predict risk of chronic disease. J Nutr. 2012;142(6):1009–1018. doi: 10.3945/jn.111.157222 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Madhu SV, Raj A, Gupta S, Giri S, Rusia U. Effect of iron deficiency anemia and iron supplementation on HbA1c levels - Implications for diagnosis of prediabetes and diabetes mellitus in Asian Indians. Clin Chim Acta. 2017;468:225–229. doi: 10.1016/j.cca.2016.10.003 [DOI] [PubMed] [Google Scholar]
- 35.Ng JM, Cooke M, Bhandari S, Atkin SL, Kilpatrick ES. The effect of iron and erythropoietin treatment on the A1C of patients with diabetes and chronic kidney disease. Diabetes Care. 2010;33(11):2310–2313. doi: 10.2337/dc10-0917 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Rajagopal L Does Iron Deficiency Anaemia and its Severity Influence HbA1C Level in Non Diabetics? An Analysis of 150 Cases. J Clin DIAGNOSTIC Res. 2017;11(2). doi: 10.7860/jcdr/2017/25183.9464 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Urrechaga E Influence of iron deficiency on Hb A1c levels in type 2 diabetic patients. Diabetes Metab Syndr Clin Res Rev. 2018;12(6):1051–1055. doi: 10.1016/j.dsx.2018.06.024 [DOI] [PubMed] [Google Scholar]
- 38.Kalra S, Coetzee A, Kalra PA, Saldaña JR, Kilov G. Rubrometabolic syndrome. Minerva Endocrinol. December 2020. doi: 10.23736/S0391-1977.20.03353-2 [DOI] [PubMed] [Google Scholar]