Abstract
Higher and lower hemoglobin concentrations are associated with coronary heart disease (CHD), but whether this risk is consistent across age, sex, and race is unclear. The Reasons for Geographic And Racial Differences in Stroke (REGARDS) study is an observational cohort study of 30 239 black, and white, adults aged 45 and older recruited 2003–7. Participants were included if they had hemoglobin measures, were CHD-free at baseline, and had all baseline variables. The primary outcome was incident CHD. Multivariable Cox proportional hazards models were used to estimate the hazard ratios (HR) and 95% confidence intervals (CI) for incident CHD by hemoglobin concentration. This was expressed as a continuous variable and divided into age-, sex-, and race-specific quintiles. The 16 332 participants were included, contributing 114 362 person-years of follow-up and 915 incident CHD events. The mean age was 63 years, 35% were male, 41% were black, and the mean baseline hemoglobin was 13.6 g/dL (SD 1.4). A significant non-linear association between hemoglobin and CHD was identified (P < .001). This association differed significantly by race (P = .025) but not by sex or age. In whites, the risk for incident CHD was higher in the lowest (HR 2.28, 95% CI 1.61, 3.33) and highest (HR 1.94, 95% CI 1.35, 2.79) hemoglobin quintiles relative to the third quintile. For blacks, only those in the lowest hemoglobin quintile had an increased risk for incident CHD events (HR 1.70, 95% CI 1.20, 2.41). Hemoglobin is an independent risk factor for CHD in whites and blacks but with different hemoglobin concentrations conferring different risks.
1 |. INTRODUCTION
Both higher and lower hemoglobin concentrations have been associated with cardiovascular disease (CVD) risk1–4 and coronary heart disease (CHD) is the leading cause of death among cardiovascular diseases.5 It is not known if higher or lower hemoglobin concentrations are merely a marker for CHD, or if there is a causal relationship. Many disease states influence hemoglobin concentrations and contribute to CVD events. For example, smoking increases hemoglobin6,7 and chronic inflammatory conditions can reduce hemoglobin.8 Higher hemoglobin may have a pro-coagulant effect by enhancing platelet interactions9 and increasing serum viscosity10 which could contribute to CHD events. Lower hemoglobin is hypothesized to contribute to CVD risk by causing a compensatory increase in cardiac output leading to left ventricular hypertrophy and arterial remodeling.11–13
In the absence of disease or nutritional deficiencies, variations in hemoglobin and hematocrit, as well as other red blood cell indices, are associated with single nucleotide polymorphisms (SNPs) that influence the production, structure, and lifespan of red blood cells.14 For unknown reasons, hemoglobin concentrations decrease with advancing age and are lower in women and among blacks.4,15–20 Clonal hematopoiesis of indeterminate potential may affect hemoglobin concentrations21 and is a known risk factor for CHD and ischemia heart failure.21–23 Since there are differences in hemoglobin concentration by age, sex, and race, the impact of hemoglobin on CHD risk may differ by these factors. A better understanding of the association between hemoglobin and CHD might assist in risk stratification and help optimize treatments for anemia and erythrocytosis that could reduce CHD incidence. We sought to determine the impact of age, sex, and race on the association of hemoglobin concentration with CHD risk in a large, geographically and racially diverse population.
2 |. METHODS
2.1 |. Setting and study population
The Reasons for Geographic And Racial Differences in Stroke (REGARDS)24 study is a prospective longitudinal cohort study assessing health outcomes in 30 239 community-dwelling participants, with oversampling in the “stroke belt” region of the Southeastern United States. REGARDS participants were black and white adults age 45 years and older, enrolled in their homes across the contiguous United States in 2003–7 and followed every 6 months for health outcomes. Complete blood counts were performed at a central laboratory, which was initiated after the first 8608 participants were recruited. Details on participant recruitment and follow-up methods have been published.24 Participants were eligible for inclusion in this analysis if they were free of CHD at baseline and had baseline hemoglobin measured. All participants provided oral and written informed consent, the study was approved by the Institutional Review Boards of all participating institutions, and the study conformed with the principles of the Declaration of Helsinki.
2.2 |. Outcome
The primary outcome of interest was incident CHD, defined as definite or probable MI (myocardial infarction) and included definite or probable acute CHD death25,26 through December 31st, 2014. Definite MI was defined as diagnostic enzymes or electrocardiogram evidence of MI. Probable MI was defined as elevated but not diagnostic enzymes, with a positive but not diagnostic electrocardiogram, or positive electrocardiogram with missing enzymes in the setting of ischemic signs and symptoms. Participants were contacted every 6 months by telephone, and medical records of potential CHD events were obtained from local hospitals. Acute CHD death was defined as death due to CHD, and was determined using information from death certificates, the last hospitalization prior to death, and interviews with next of kin. Acute CHD death was included in the primary outcome and as a secondary outcome separately. Incident CHD outcomes were adjudicated by two investigators and any disagreements were adjudicated by a committee.
2.3 |. Risk factor assessment
Baseline risk factors were obtained using an initial computer assisted telephone interview followed by an in-home assessment for a medication inventory, fasting phlebotomy, and anthropomorphic measures. Baseline CHD was identified by a history of self-reported myocardial infarction (MI), coronary artery bypass surgery, percutaneous coronary intervention, or evidence of a prior MI on enrollment electrocardiogram. Sex (male or female) and race (black, white) were based on participant self-report. Weight and height were measured at study enrollment, and body mass index (BMI) was calculated as the weight in kilograms divided by the height in meters squared. The stroke belt was defined geographically (North Carolina, South Carolina, Georgia, Tennessee, Alabama, Mississippi, Arkansas, Louisiana). Likewise, the stroke buckle was defined geographically (coastal plains of North Carolina, South Carolina, and Georgia, and the rest of the contiguous United States). Annual income (<$20 000, $20 000-$34 000, $35 000-$74 000, >$75 000, refused), smoking status (current, former, never), alcohol intake (drinks/week), and history of stroke were defined during the computer assisted telephone interview. Medication use (including anti-hypertensive medication and cholesterol lowering medications) was assessed during the interview.
2.4 |. Laboratory studies
Phlebotomy was performed during a morning in-home visit after a 10–12 hour fast, and samples were centrifuged, refrigerated, and shipped to a laboratory at the University of Vermont. Complete blood counts (including hemoglobin and hematocrit) were performed on whole blood in the central laboratory from ethylenediaminetetraacetic acid tubes using automated cell counting on a Beckman Coulter LH 755 Hematology Workcell (Beckman Coulter, Incorporated, Fullerton, California). The coefficient of variation for hemoglobin was 3%, and measurements were successful for 90.3% of samples. Methods and definitions for other laboratory analytes have been previously described.18
2.5 |. Statistical analysis
Hemoglobin was divided into age-, sex, and race-specific quintiles, and modeled as a continuous variable. Associations between all risk factors and incident CHD and acute CHD death were evaluated by using χ2 tests for categorical variables and two-tailed t tests for continuous variables. Cox proportional hazards models were then used to estimate the hazard ratios (HR) and 95% confidence intervals (CI) for incident CHD by hemoglobin measures. Two models were used; Model 1 adjusted for demographic variables (age, sex, race, and region of residence) and traditional CHD risk factors (systolic blood pressure, use of antihypertensive medications, diabetes (yes/no), current smoking (yes/no), total cholesterol (mg/dL), HDL cholesterol (mg/dL), and use of lipid-lowering medications). Model 2, an extended CHD risk factor model, included all risk factors in Model 1, and the following additional variables: annual income (<$25 000, $25 000 - $75 000, >$75 000, and refused), pre-baseline stroke, estimated glomerular filtration rate (eGFR,27 continuous), log transformed C-reactive protein (CRP), BMI, and number of alcoholic drinks per week. When hemoglobin was examined as a continuous variable, a non-linear association was assessed by adding a hemoglobin-squared term into the model. For the quintile analysis, interactions between hemoglobin and age, race, and sex were evaluated by adding the cross-product terms of the variables to the model, with a significant interaction defined as a P value of <.1. Stratified results were reported when significant interaction was identified. Sensitivity analyses were performed by duplicating the above analyses using hematocrit and excluding participants with a reduced mean corpuscular volume (MCV). All analyses were performed with SAS version 9.1 software (SAS Institute, Inc., Cary, North Carolina).
3 |. RESULTS
Of 30 239 REGARDS participants, 16 332 had hemoglobin measures and were CHD-free at baseline (Figure S1). The mean age was 63 years (IQR 56, 70), 35% were male, and 41% were black. The mean baseline hemoglobin was 13.6 g/dL (SD 1.4, IQR 12.7, 14.6). During 114 362 person-years of follow up (median 8.1 years) there were 915 incident CHD events (694 definite or probable MIs and 221 acute CHD deaths). Compared to those without CHD, subjects with definite or probable MI were older and were more likely to have diabetes mellitus (32% vs 17%), report current smoking (19% vs 14%), and have a history of prior stroke (18% vs 14%) (Table 1). Race (black vs white) was not associated with MI but blacks were more likely to have acute CHD death (53% vs 47%). Cholesterol-lowering medications and anti-hypertensive medications were more frequent in those with incident MI compared to those with no CHD. Mean hemoglobin did not differ by MI status but was lower for those with acute CHD death (13.2 vs 13.7, P < .001).
TABLE 1.
Baseline characteristics of study participants
| All subjects | No CHD Total = 14 738 N (%) | Definite or Probable MI Total = 694 N (%) | P valueb | No Acute CHD Death Total = 15 211 N (%) | Acute CHD Death Total = 221 N (%) | P valueb | |
|---|---|---|---|---|---|---|---|
| Age, years | |||||||
| ≤63 | 8222 (53.3) | 7984 (54.2) | 238 (34.3) | <.001 | 8155 (53.6) | 67 (30.3) | <.001 |
| >63 | 7210 (46.7) | 6754 (45.8) | 456 (65.7) | 7056 (46.4) | 154 (69.7) | ||
| Sex | |||||||
| Female | 10 088 (65.4) | 9734 (66.1) | 354 (51.0) | <.001 | 9971 (65.6) | 117 (52.9) | <.001 |
| Male | 5344 (34.6) | 5004 (34.0) | 340 (49.0) | 5240 (34.5) | 104 (47.1) | ||
| Body mass index (kg/m2) | 29.4 (6.32) | 29.4 (6.3) | 29.9 (6.6) | .021 | 29.4 (6.3) | 29.8 (6.8) | .34 |
| Race | |||||||
| White | 9128 (59.2) | 8718 (59.2) | 410 (59.1) | .969 | 9025 (59.3) | 103 (46.6) | <.001 |
| Black | 6304 (40.9) | 6020 (40.9) | 284 (40.9) | 6186 (40.7) | 118 (53.4) | ||
| Geographic Location | |||||||
| Other | 6271 (40.6) | 6014 (40.8) | 257 (37.0) | .082 | 6190 (40.7) | 81 (36.7) | .007 |
| Stoke Belt | 5406 (35.0) | 5138 (34.9) | 268 (38.6) | 5307 (34.9) | 99 (44.8) | ||
| Buckle | 3755 (24.3) | 3586 (24.3) | 169 (24.4) | 3714 (24.4) | 41 (18.6) | ||
| Annual Income | |||||||
| <$20 K | 2419 (15.7) | 2276 (15.4) | 143 (20.6) | <.001 | 2365 (15.6) | 54 (24.4) | <.001 |
| $20 K– $34 K | 3494 (22.6) | 3316 (22.5) | 178 (25.7) | 3430 (22.6) | 64 (29.0) | ||
| $35 K – $74 K | 4741 (30.7) | 4545 (30.8) | 196 (28.2) | 4687 (30.8) | 54 (24.4) | ||
| >$75 K | 2796 (18.1) | 2707 (18.4) | 89 (12.8) | 2772 (18.2) | 24 (10.9) | ||
| Refused | 1982 (12.9) | 1894 (12.9) | 88 (12.7) | 1957 (12.9) | 25 (11.3) | ||
| Diabetes mellitus | 2777 (18.0) | 2557 (17.4) | 220 (31.7) | <.001 | 2704 (17.8) | 73 (33.0) | <.001 |
| Current smoking | 2156 (14.0) | 2029 (13.8) | 127 (18.3) | <.001 | 2115 (13.9) | 41 (18.6) | .048 |
| Hypertension | 7680 (49.8) | 7239 (49.1) | 441 (63.5) | <.001 | 7532 (49.5) | 148 (67.0) | <.001 |
| Cholesterol medication | 4071 (26.4) | 3844 (26.1) | 227 (32.7) | <.001 | 4002 (26.3) | 69 (31.2) | .1 |
| Alcoholic drinks/ week | 2.0 (6.0) | 2.0 (5.9) | 2.1 (8.8) | .57 | 2.0 (6.0) | 1.9 (6.2) | .86 |
| CVD (baseline stroke) | 877 (5.7) | 790 (5.4) | 87 (12.5) | <.001 | 841 (5.5) | 36 (16.3) | <.001 |
| Hemoglobin quintile | |||||||
| 1 | 3290 (21.3) | 3083 (20.9) | 207 (29.8) | <.001 | 3212 (21.1) | 78 (35.3) | .004 |
| 2 | 3217 (20.9) | 3093 (21.0) | 124 (17.9) | 3182 (20.9) | 35 (15.8) | ||
| 3 | 2993 (19.4) | 2868 (19.5) | 125 (18.0) | 2953 (19.4) | 40 (18.1) | ||
| 4 | 3009 (19.5) | 2912 (19.8) | 97 (14.0) | 2975 (19.6) | 34 (15.4) | ||
| 5 | 2923 (18.9) | 2782 (18.9) | 141 (20.3) | 2889 (19.0) | 34 (15.4) | ||
| (Mean, SD) | (Mean, SD) | P value c | (Mean, SD) | (Mean, SD) | P value c | ||
| Age, years | 63 (9.6) | 63 (9.6) | 67 (9.2) | <.001 | 63 (9.6) | 69 (9.4) | <.001 |
| Hemoglobin (g/dL) | 13.6 (1.4) | 13.6 (1.4) | 13.6 (1.7) | .501 | 13.7 (1.4) | 13.2 (1.6) | <.001 |
| Hematocrit (%) | 40.4 (4.1) | 40.4 (4.1) | 40.4 (4.9) | .762 | 40.5 (4.1) | 39.4 (4.8) | .002 |
| MCV (<80 fl) | 673 (4.4) | 635 (4.3) | 38 (5.5) | .169 | 657 (4.3) | 16 (7.2) | .052 |
| Total Cholesterol (mg/dL) | 194 (39.6) | 194 (39.3) | 192 (44.0) | .167 | 194 (39.5) | 191 (44.0) | .317 |
| Estimated GFR27 (mL/min/BSA) | 88 (20) | 88 (19) | 80 (23) | <.001 | 88 (19) | 78 (27) | <.001 |
| HDL (mg/dL) | 54 (16.5) | 54 (16.6) | 50 (15.3) | <.001 | 54 (16.5) | 52 (15.2) | .085 |
| LDL (mg/dL) | 115 (34.6) | 115 (34.5) | 113 (37.8) | .241 | 115 (34.6) | 114 (38.3) | .836 |
| SBP (mmHg) | 126 (16) | 125 (16) | 132 (19) | <.001 | 126 (16) | 134 (21) | <.001 |
| CRP (median mg/L, 25% - 75%)a | 2.2 (0.9–5.0) | 2.2 (0.9–4.9) | 3.0 (1.3–6.2) | <.001 | 2.2 (0.9–5.0) | 3.3 (1.4–7.7) | <.001 |
Abbreviations: CHD, coronary heart disease; CRP, C-reactive protein; GFR, glomerular filtration rate; HDL, high-density lipoprotein; HTN, hypertension; LDL, low-density lipoprotein; SBP, systolic blood pressure.
CRP is log-transformed to conduct t test.
P value calculated using Chi-squared.
P value calculated using t-test.
Initial analysis using hemoglobin as a continuous variable in a Cox proportional hazard model found a HR of CHD of 0.84 (95% CI 0.77–0.92) per SD higher hemoglobin adjusting for traditional risk factors, and a HR of 0.91 (95% CI 0.83–0.99) in the extended risk factor model. Addition of a hemoglobin squared term in the model identified a significant non-linear association between hemoglobin and CHD with both models (P < .001). To examine this non-linear association, hemoglobin was divided into age, race, and sex-specific quintiles (Table 2). Women, blacks, and older individuals had lower hemoglobin cutoffs for the quintiles. Adjusting for traditional risk factors, increased hazard ratios were seen for the lowest and highest quintiles vs the fourth quintile (Q1 HR 1.77, 95% CI 1.39–2.26; Q5 HR 1.33, 95% CI 1.03–1.73) with no statistically significant differences in risk noted for the second and third quintiles (second vs fourth, HR 1.15, 95% CI 0.88–1.49; third vs fourth, 1.26, 95% CI 0.97–1.64).
TABLE 2.
Age, sex, and race specific hemoglobin percentiles used to define quintiles
| Age | Sex | Race | 20th Hgb (g/dL) | 40th Hgb (g/dL) | 60th Hgb (g/dL) | 80th Hgb (g/dL) |
|---|---|---|---|---|---|---|
| Age > 63 | Female | Black | 11.7 | 12.4 | 12.9 | 13.6 |
| White | 12.6 | 13.2 | 13.8 | 14.3 | ||
| Male | Black | 12.7 | 13.5 | 14.1 | 14.8 | |
| White | 13.7 | 14.5 | 15 | 15.7 | ||
| Age ≤ 63 | Female | Black | 11.9 | 12.6 | 13.1 | 13.7 |
| White | 12.7 | 13.3 | 13.8 | 14.4 | ||
| Male | Black | 13.3 | 14.1 | 14.7 | 15.4 | |
| White | 14.2 | 14.8 | 15.4 | 16 |
Note: Quintile 1: <20th percentile, Quintile 2: 20th-40th percentile, Quintile 3: 40th-60th percentile, Quintile 4: 60th-80th percentile, Quintile 5: >80th percentile.
Subsequent analyses for interaction between hemoglobin and age, race, and sex, found significant interaction between hemoglobin quintiles and race in the traditional (P = .025) and extended (P = .053) risk factor models. There were no significant interactions between hemoglobin quintiles and age (P = .12) or sex (P = .59). Therefore, results for the quintile analyses were stratified by race (white vs black). For whites, the hazard for incident CHD was lowest for the fourth quintile compared to all other quintiles, so this was used as the reference group. In the traditional model, white participants in all quintiles had increased risk of CHD compared to the fourth quintile (Table 3). The risk was highest in the lowest (HR 2.28, 95% CI 1.61, 3.33) and highest (HR 1.94, 95% CI 1.35, 2.79) hemoglobin quintiles. For blacks, the hemoglobin quintile with the lowest event rate was the second quintile and this was used as the reference group for analysis. Only the lowest hemoglobin quintile was associated with an increased risk for incident CHD events (first vs second: HR 1.70, 95% CI 1.20, 2.41). Higher hemoglobin was not associated with incident CHD risk in black participants (fifth vs second: HR 1.05, 95% CI 0.71–4.57; fourth vs second: HR 1.24, 95% CI 0.85–1.82).
TABLE 3.
Association of hemoglobin quintiles and coronary heart disease events among blacks and whites
| Incident Coronary Heart Disease | ||||||
|---|---|---|---|---|---|---|
|
| ||||||
| Traditional Risk Factor Model |
Extended Risk Factor Model |
|||||
| Hemoglobin | Black | White | Black | White | Black | White |
| Quintile | Observed IR (cases per 1000 person-years) | HR (95% CI) | HR (95% CI) | HR (95% CI) | HR (95% CI) | |
| 1 | 9.30 | 8.61 | 1.70 (1.20, 2.41) | 2.28 (1.61, 3.33) | 1.50 (1.04, 2.15) | 2.18 (1.52, 3.13) |
| 2 | 5.15 | 5.21 | Ref | 1.58 (1.09, 2.29) | Ref | 1.56 (1.06, 2.29) |
| 3 | 5.56 | 5.51 | 1.10 (0.74, 1.64) | 1.73 (1.20, 2.50) | 1.16 (0.77, 1.74) | 1.79 (1.23, 2.62) |
| 4 | 5.99 | 3.09 | 1.24 (0.85, 1.82) | Ref | 1.23 (0.84, 1.84) | Ref |
| 5 | 5.50 | 7.09 | 1.05 (0.71, 4.57) | 1.94 (1.35, 2.79) | 1.09 (0.72, 1.65) | 1.94 (1.33, 2.82) |
| Acute coronary heart disease death | ||||||
| 1 | 4.19 | 2.87 | 2.20 (1.21, 4.01) | 4.67 (2.08, 10.45) | 1.82 (0.97, 3.40) | 4.06 (1.81, 9.13) |
| 2 | 1.87 | 1.21 | 1.03 (0.52, 2.04) | 2.28 (0.95, 5.50) | 1.05 (0.52, 2.11) | 1.96 (0.80, 4.81) |
| 3 | 1.79 | 1.78 | Ref | 3.70 (1.16, 8.49) | Ref | 3.62 (1.57, 8.35) |
| 4 | 2.84 | 0.60 | 1.71 (0.9, 3.21) | Ref | 1.61 (0.83, 3.11) | Ref |
| 5 | 2.47 | 1.00 | 1.27 (0.65, 2.49) | 1.68 (0.66, 4.26) | 1.34 (0.67, 2.67) | 1.71 (0.67, 4.35) |
Abbreviations: CI, confidence interval, IR, incidence rate, HR, hazard ratio, Ref, reference.
Incidence of acute CHD death was numerically higher for blacks than for whites in the upper and lower hemoglobin quintiles. Considering acute CHD death events in whites, participants in the highest hemoglobin quintile had a non-significantly increased risk (Q5 vs Q4: HR 1.68, 95% CI 0.66, 4.26), with the lowest quintile having a very elevated risk (Q1 vs Q4: HR 4.67 95% CI 2.08, 10.45). Similarly, for blacks, there was a more pronounced risk for acute CHD deaths than what was seen for incident CHD in the first hemoglobin quintile (Q1 vs Q3, HR 2.20, 95% CI 1.21, 4.01). In blacks, no significantly increased risk of acute CHD death was seen in the higher quintiles (fifth vs third: HR 1.27, 95% CI 0.65–2.49; fourth vs third: HR 1.71, 95% CI 0.90–3.21).
A secondary analysis was performed with hematocrit that also showed a non-linear association with incident CHD. When assessing hematocrit quintiles in relation to incident CHD there was also a significant interaction by race, but not sex or age. Hazard ratios for incident CHD and acute CHD death by hematocrit quintiles differed from what was observed with hemoglobin (Table S1). Reference hematocrit quintile differed for hemoglobin and hematocrit in blacks and whites (Figure 1). In blacks, the lowest hemoglobin and hematocrit quintile were associated with incident CHD compared to the reference quintile (Hgb, HR 1.7, 95% CI 1.2–2.4; Hct, 2.1, 95% CI 1.6–3.3). In whites, no hematocrit quintile was associated with total CHD compared to the reference.
FIGURE 1.
Adjusted hazard ratios in blacks (A) and whites (B) for incident coronary heart disease events by hemoglobin and hematocrit quintile
4 |. DISCUSSION
Our results provide data from a large, geographically diverse biracial cohort study that hemoglobin concentration is an independent risk factor for incident CHD and acute CHD death in blacks and whites, but with different concentrations associated with adverse events. After multivariable adjustment, an increased risk for incident CHD was seen in whites among all hemoglobin quintiles compared to the reference quintile, and the risk was most pronounced in the lowest quintile. In contrast, among blacks only those in the lowest hemoglobin quintile had an increased risk of incident CHD. Acute CHD death risk was observed for blacks and whites in the lowest hemoglobin quintile, but not in the higher hemoglobin quintiles. No difference in risk was found between men and women or by age when using age, sex, and race-specific hemoglobin quintiles.
The largest studies examining CHD outcomes related to hemoglobin concentrations are from Korea. The Korean National Health Insurance Service National Screening Cohort (NHIS-HEALS) included 292 194 men and women age > 40 years old from 2002 to 2013.3 MI was identified by ICD-10 codes and after a median follow up of 7.8 years there were 559 events identified. Hemoglobin was divided into sex-specific quintiles and an increased hazard ratio for MI was found in women with a hemoglobin <11 g/Dl, compared to 12–12.9 g/dL (HR 2.21, 95% CI 1.11–4.42). No statistically significant increases were seen for men or women in the higher or lower quintiles. Interestingly, previous data from the Korean Heart Study from 1993–2004 examining 407 858 men and women did demonstrate an increased risk for ischemic heart disease in men (but not women) in the highest and lowest hemoglobin quintiles.3 In contrast to these studies, the current study evaluated white and black participants in the United States. The outcome definition also differed as our primary outcome was based on prospective follow up and adjudication of outcomes. Despite the significantly larger sample size of the NHIS-HEALS study, the number of events observed over a similar follow-up interval was higher in the current study (incidence rate 69.1 vs 2.5 per 10 000 person years), likely a result of a population with a higher baseline risk for CHD and robust follow up and prospective identification of events.
The Framingham Heart Study,28 which enrolled a predominately white population, demonstrated an increased risk for CHD and CHD mortality in younger women (age 35–64 years) in the lowest and highest quintiles of hemoglobin. Data from the Second National Health and Nutrition Examination Survey (NHANES II) study (8896 adults aged 30–75) demonstrated no increased risk for CHD mortality for men and women overall when comparing the highest tertile of hemoglobin concentration to the lowest after multivariable adjustment.29 However, similar to the Framingham study, women <65 years old in the highest tertile of hemoglobin did have higher CHD mortality (RR 2.2, 95% CI 1.0–4.6). Two cohort studies examined the risk of elevated hematocrit on CHD specifically in men. The first study (The British Regional Heart Study, 7735 men aged 40–59) demonstrated that men with a hematocrit ≥46% had an increased risk of ischemic heart disease events (RR 1.27, 95% CI 1.06–1.51) after multivariate adjustment.30 The second and smaller study from The Tampere Adult Population Cardiovascular Risk study (TAMRISK) in Finland (670 men, all 55 years old), also demonstrated an increased risk of CHD mortality, with a hematocrit ≥50% compared to those with a hematocrit <50 after multivariate adjustment (HR 1.8, 95% CI 1.1–2.7).31
Most prospective cohort studies have enrolled predominately white participants and so the association between hemoglobin and CHD has not been fully evaluated in blacks. The Atherosclerosis Risk in Communities (ARIC) study (14 410 men and women without baseline CVD) included 3833 blacks and found that anemia (Hgb <13 g/dL in men, and < 12 g/dL in women) was an independent risk factor for CVD mortality (HR 1.41 95% CI 1.01–1.95).11 Note, CHD was not specifically examined. No significant interaction was found between hemoglobin and age, race, or sex. Discrepant findings in previous cohort studies could potentially be attributed to varying methodologies, covariates included, patient populations, whether hemoglobin or hematocrit were used, and the way the data were divided into groups for comparisons. The World Health Organization’s definition of anemia (men <13.0 g/dL, women <12.0 g/dL), as a method to stratify hemoglobin, is in some studies imprecise, when evaluating a diverse patient population.20,32 It is unclear if some studies excluded assessment of non-linear associations between hemoglobin and CHD, but studies that evaluated certain cut-off levels for hemoglobin or compared higher levels to lower levels could have potentially missed or underestimated the risk of CHD. Our analysis highlights the different results that can be obtained within the same population and the same grouping of data into quintiles, with some results differing between our hemoglobin and hematocrit analysis. While hemoglobin and hematocrit are strongly correlated and are often thought of interchangeably, the tests are not exactly the same. In fact, genome wide association studies have found different genetic loci associated with hemoglobin and hematocrit.14
It is possible that findings of CHD risk at higher and lower hemoglobin concentrations could indicate occult diseases, or inflammatory (autoimmune) diseases, that could confound the association between hemoglobin and CHD. Although not adjusted for specifically, these conditions would be rare, and adjustment was performed with the CRP. As expected, we did see higher CRP values in those with incident CHD and CHD mortality, but accounting for this and other variables (eGFR) known to influence hemoglobin levels18 in our extended risk factor model did not significantly change our findings compared to the traditional risk factor model. Iron studies were not performed on study participants and could cause lower hemoglobin concentrations, possibly increasing CHD events. However, a sensitivity analysis was performed excluding participants with a reduced mean corpuscular volume (MCV), which could be indicative of iron deficiency (Table S2), and the results remained similar. Information on lung diseases or pulmonary function was not available, and might confound the relationship between CHD and hemoglobin. Annual income and geographic location were adjusted for in the extended risk factor model, but this did not account for the differences in outcomes seen among white and black participants, although additional unmeasured socioeconomic factors may contribute.
The mechanism for why higher hemoglobin may contribute to CHD is unclear. Smoking is known to cause an increase in hemoglobin concentrations and would contribute to higher CHD events and current smoking status was adjusted for in the multivariate models. After enrollment, if white active smokers were significantly heavier smokers than blacks, this might account for the differences observed in the risk for higher hemoglobin seen in white participants but not blacks. However, it would also be expected that this would confer a higher risk for CHD mortality which was not observed. Red blood cell phenotypic profiles are known to vary based on race, and there may be other single nucleotide polymorphisms that affect RBC lifespan and hence hemoglobin concentrations. It is possible that these RBC structural or surface proteins may interact with the vessel wall and influence CHD. In our analysis, we were unable to more fully explore subtypes of incident CHD that could provide additional insight into a mechanism of action. It is possible that the effects of hemoglobin may have different outcomes depending on the type of MI (type 1 or type 2 MI33).
A strength of our analysis was the stratification into age, sex, and race-specific quintiles, which has not been performed in previous analyses. This helps control for the known differences in hemoglobin by these factors and better allows simultaneous examination of risk at higher and lower hemoglobin values in a diverse patient population. Other studies have found differences in cardiovascular outcomes by hemoglobin concentrations based on sex, which we did not find. This is likely due to the creation of age, race, and sex-specific quintiles rather than more arbitrary hemoglobin cut-offs. Additional strengths of this study include the broad area of geographic recruitment, large number of black participants’ recruited and consistent definition of CHD with ongoing follow up and rigorous adjudication processes. The laboratory data quality and consistency is robust as the baseline hemoglobin measure was analyzed in a central laboratory preventing differences in measurements that could be attributed to different laboratory techniques in other observational studies. Baseline hemoglobin and other laboratory testing occurred at an ideal time with in-home samples obtained in healthy individuals and are unlikely to be influenced by acute illnesses. A limitation of the data is that baseline complete blood counts were not performed in the first 8608 participants and was missing or not analyzable for some subsequent participants. Every effort to ensure optimal transportation of samples was made but it is possible that pre-analytical variables effected sample quality. Our findings are also limited by the observational study design that precludes causal inference and may not be fully generalizable to non-study participants. However, this issue is assuaged by the prospective study design which eliminates the possibility of reverse causality in explaining findings.
5 |. CONCLUSION
Hemoglobin is an independent risk factor for CHD and CHD mortality in whites and blacks, but with different hemoglobin concentrations conferring different risks by race. Whether hemoglobin concentration is causative or reflects underlying unmeasured comorbid conditions that cannot be fully accounted for in our analysis requires further work. In contrast to other large cohort studies, we did not find a difference in the relationship between hemoglobin and risk for CHD by sex or age, likely because of the use of age, sex, and race-specific quintiles. Further research is needed to explore possible mechanisms mediating increased CHD risk associated with hemoglobin levels, and to understand why the risk profiles differ among whites and blacks. Our data point to “optimal” hemoglobin reference ranges based on age, sex, and race-specific hemoglobin values but not to interventions based on changing hemoglobin concentrations. For example, detection of higher hemoglobin might prove useful in motivating smoking cessation efforts or could be used to prompt an evaluation for other known modifiable risk factors. Whether hemoglobin is an “innocent” bystander or causal factor in CHD is not known, but lower and higher hemoglobin in whites and lower hemoglobin in blacks should be viewed as an independent risk factor for CHD.
Supplementary Material
SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of this article.
ACKNOWLEDGMENTS
This research project is supported by cooperative agreement U01 NS041588 co-funded by the National Institute of Neurological Disorders and Stroke (NINDS) and the National Institute on Aging (NIA), National Institutes of Health, Department of Health and Human Service. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NINDS or the NIA. Representatives of the NINDS were involved in the review of the manuscript but not directly involved in the collection, management, analysis or interpretation of the data. The CHD outcomes were obtained from grant R01 HL80477 from the National Heart, Lung, Blood Institute. Funding for measurement of complete blood count was provided by an investigator-initiated grant from Amgen. Representatives from Amgen had no role in the analysis plan, reviewing the manuscript, or decision to publish. We thank the other investigators, the staff, and the participants of the REGARDS study for their valuable contributions. A full list of participating REGARDS investigators and institutions can be found at http://www.regardsstudy.org
Funding information
Amgen; Department of Health; National Institutes of Health; National Institute on Aging; National Institute of Neurological Disorders and Stroke
Footnotes
DISCLOSURE OF INTEREST
The authors report no conflicts of interest. This paper is not under consideration elsewhere and none of the paper’s contents have been previously published.
AUTHOR CONTRIBUTIONS
All authors were involved in the conception and design or analysis and interpretation of the data, drafting of the manuscript or revising it critically, and read and approved the final manuscript.
REFERENCES
- 1.Erikssen G, Thaulow E, Sandvik L, Stormorken H, Erikssen J. Haematocrit: a predictor of cardiovascular mortality? J Intern Med. 1993;234(5):493–499. [DOI] [PubMed] [Google Scholar]
- 2.Lee G, Choi S, Kim K, et al. Association of hemoglobin concentration and its change with cardiovascular and all-cause mortality. J Am Heart Assoc. 2018;7(3):e007723. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Kim MY, Jee SH, Yun JE, Baek SJ, Lee DC. Hemoglobin concentration and risk of cardiovascular disease in Korean men and women - the Korean heart study. J Korean Med Sci. 2013;28(9):1316–1322. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Zakai NA, Katz R, Hirsch C, et al. A prospective study of anemia status, hemoglobin concentration, and mortality in an elderly cohort: the cardiovascular health study. Arch Intern Med. 2005;165(19):2214–2220. [DOI] [PubMed] [Google Scholar]
- 5.Writing Group, Lloyd-Jones DM, Adams RJ, et al. Heart disease and stroke statistics—2010 update: a report from the American Heart Association. Circulation. 2010;121(7):e46–e215. [DOI] [PubMed] [Google Scholar]
- 6.Nordenberg D, Yip R, Binkin NJ. The effect of cigarette smoking on hemoglobin levels and anemia screening. JAMA. 1990;264(12):1556–1559. [PubMed] [Google Scholar]
- 7.Malenica M, Prnjavorac B, Bego T, et al. Effect of cigarette smoking on haematological parameters in healthy population. Med Arch. 2017; 71(2):132–136. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Weiss G, Goodnough LT. Anemia of chronic disease. N Engl J Med. 2005;352(10):1011–1023. [DOI] [PubMed] [Google Scholar]
- 9.Walton BL, Lehmann M, Skorczewski T, et al. Elevated hematocrit enhances platelet accumulation following vascular injury. Blood. 2017; 129:2537–2546. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Price JF, Mowbray PI, Lee AJ, Rumley A, Lowe GD, Fowkes FG. Relationship between smoking and cardiovascular risk factors in the development of peripheral arterial disease and coronary artery disease: Edinburgh Artery Study. Eur Heart J. 1999;20(5):344–353. [DOI] [PubMed] [Google Scholar]
- 11.Sarnak MJ, Tighiouart H, Manjunath G, et al. Anemia as a risk factor for cardiovascular disease in the atherosclerosis risk in communities (aric) study. J Am Coll Cardiol. 2002;40(1):27–33. [DOI] [PubMed] [Google Scholar]
- 12.Metivier F, Marchais SJ, Guerin AP, Pannier B, London GM. Pathophysiology of anaemia: focus on the heart and blood vessels. Nephrol Dial Transplant. 2000;15(suppl 3):14–18. [DOI] [PubMed] [Google Scholar]
- 13.Anemia Levin A. and left ventricular hypertrophy in chronic kidney disease populations: a review of the current state of knowledge. Kidney Int Suppl. 2002;61(80):35–38. [DOI] [PubMed] [Google Scholar]
- 14.Ganesh SK, Zakai NA, van Rooij FJA, et al. Multiple loci influence erythrocyte phenotypes in the CHARGE consortium. Nat Genet. 2009;41(11):1191–1198. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Cheng C, Chan J, Cembrowski GS, van Assendelft OW. Complete blood count reference interval diagrams derived from NHANES III: stratification by age, sex, and race. Lab Hematol. 2004;10(1):42–53. [PubMed] [Google Scholar]
- 16.Salive ME, Cornoni-Huntley J, Guralnik JM, et al. Anemia and hemoglobin levels in older persons: relationship with age, gender, and health status. J Am Geriatr Soc. 1992;40(5):489–496. [DOI] [PubMed] [Google Scholar]
- 17.Nilsson-Ehle H, Jagenburg R, Landahl S, Svanborg A. Blood haemoglobin declines in the elderly: implications for reference intervals fromage 70 to 88. Eur J Haematol. 2000;65(5):297–305. [DOI] [PubMed] [Google Scholar]
- 18.Zakai NA, McClure LA, Prineas R, et al. Correlates of anemia in American blacks and whites the REGARDS renal ancillary study. Am J Epidemiol. 2009;169(3):355–364. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Sorlie PD, Garcia-Palmieri MR, Costas R, Havlik RJ. Hematocrit and risk of coronary heart disease: the Puerto Rico health program. Am Heart J. 1981;101(4):456–461. [DOI] [PubMed] [Google Scholar]
- 20.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] [PMC free article] [PubMed] [Google Scholar]
- 21.Dorsheimer L, Assmus B, Rasper T, et al. Association of mutations contributing to clonal hematopoiesis with prognosis in chronic ischemic heart failure. JAMA Cardiol. 2019;4(1):25–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Jaiswal S, Natarajan P, Silver AJ, et al. Clonal hematopoiesis and risk of atherosclerotic cardiovascular disease. N Engl J Med. 2017;377(2):111–121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Jaiswal S, Fontanillas P, Flannick J, et al. Age-related clonal hematopoiesis associated with adverse outcomes. N Engl J Med. 2014;371 (26):2488–2498-2498. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Howard VJ, Cushman M, Pulley L, et al. The reasons for geographic and racial differences in stroke study: objectives and design. Neuroepidemiology. 2005;25(3):135–143. [DOI] [PubMed] [Google Scholar]
- 25.Redmond N, Richman J, Gamboa CM, et al. Perceived stress is associated with incident coronary heart disease and all-cause mortality in low- but not high-income participants in the reasons for geographic and racial differences in Stroke study. J Am Heart Assoc. 2013;2(6):1–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Safford MM, Brown TM, Muntner PM, et al. Association of race and sex with risk of incident acute coronary heart disease events. JAMA. 2012;308(17):1768–1774. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Levey AS, Stevens LA, Schmid CH, et al. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150(9):604–612. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Gagnon DR, Zhang T-J, Brand FN, Kannel WB. Hematocrit and the risk of cardiovascular disease—the Framingham study: a 34-year follow-up. Am Heart J. 1994;127(3):674–682. [DOI] [PubMed] [Google Scholar]
- 29.Brown DW, Giles WH, Croft JB. Hematocrit and the risk of coronary heart disease mortality. Am Heart J. 2001;142(4):657–663. [DOI] [PubMed] [Google Scholar]
- 30.Wannamethee G, Shaper AG, Whincup PH. Ischaemic heart disease: association with haematocrit in the British Regional Heart Study. J Epidemiol Commun Health. 1994;48(2):112–118. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Kunnas T, Solakivi T, Huuskonen K, Kalela A, Renko J, Nikkari ST. Hematocrit and the risk of coronary heart disease mortality in the TAMRISK study, a 28-year follow-up. Prev Med. 2009;49(1):45–47. [DOI] [PubMed] [Google Scholar]
- 32.Cappellini MD, Motta I. Anemia in clinical practice-definition and classification: does hemoglobin change with aging? Semin Hematol. 2015; 52(4):261–269. [DOI] [PubMed] [Google Scholar]
- 33.Thygesen K, Alpert JS, Jaffe AS, et al. Joint ESC/ACCF/AHA/WHF Task Force for Universal Definition of Myocardial InfarctionThird universal definition of myocardial infarction. J Am Coll Cardiol. 2012;60 (16):1581–1598. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.