Limited information is available about gender and ethnic differences in red cell distribution width (RCDW) with regard to its relation to mortality in a population free of cardiovascular (CV) disease and diabetes. To assess gender and ethnic differences in RCDW and their effect on the association between RCDW and mortality, the Third National Health and Nutritional Examination Survey (n = 15,460, 1988 to 1994) data were examined. Multivariate adjusted Cox proportional hazard analysis was performed to assess effect of gender and ethnicity on the association between RCDW and mortality (total, CV disease, and coronary heart disease [CHD]). RCDW (mean ± SE) was greater in black women (13.1 ± 0.03) and men (13.4 ± 0.02) compared to women of white (12.9 ± 0.02) and other (13.0 ± 0.07) ethnicities and men of white (13.3 ± 0.02) and other (13.3 ± 0.07) ethnicities, respectively (p <0.001). The interaction between RCDW and gender was statistically significant for all study outcomes (p <0.001) but nonsignificant for RCDW and ethnicity. After adjusting for key variables, RCDW in women was associated with adjusted hazard ratios of 1.22 (95% confidence interval [CI] 1.14 to 1.31) for all-cause mortality, 1.17 (95% CI 1.07 to 1.28) for CV deaths, and 1.18 (95% CI 1.03 to 1.35) for CHD deaths; in men, adjusted hazard ratios were 1.29 (95% CI 1.20 to 1.38) for all-cause mortality, 1.27 (95% CI 1.17 to 1.37) for CV deaths, and 1.25 (95% CI 1.13 to 1.39) for CHD deaths (p <0.05 for all). In conclusion, blacks and men have significantly greater RCDWs compared to whites and women. Greater RCDW is associated with a greater risk of mortality in men compared to women, whereas no effect modification is observed by ethnicity.
Red cell distribution width (RCDW) is a mathematical index calculated from the mean corpuscular value (MCV) and is representative of heterogeneity in the size of red blood cells. It is traditionally used for the differential diagnosis of anemia of different causes, with more recent data emerging to support its independent role in predicting mortality and cardiovascular (CV) events. Because there is a paucity of data exploring gender and ethnic differences in RCDW and their effect on outcomes, we sought to determine whether RCDW values differ by gender and ethnicities. We further aimed at exploring the effect of gender and ethnicity on the association between RCDW and mortality, including CV and coronary heart disease (CHD) mortality.
Methods
Data from the Third National Health and Nutrition Examination Survey (NHANES III; 1988 to 1994), a high-quality survey designed to provide health and nutritional information of a nationally representative noninstitutionalized civilian cohort, were analyzed. Informed consent was obtained from each participant and institutional review board approval was obtained by the National Center for Health Statistics. Detailed information on the study design, data collection, and quality control protocols has been previously published.
Analysis of data was performed for those ≥20 years old after excluding diabetics (those who had physician diagnosis of diabetes or those with hemoglobin A1c ≥6.5 according to the current American Diabetic Association definition of diabetes or those using oral hypoglycemic medications or insulin for diabetes), those with previous CV disease, and those with missing values of the study variables (n = 15,460). Gender, race/ethnicities (white, African-American, and other), smoking status (current smoker vs not), and family history of CHD were self-reported. Body mass index (kilogram per meter squared) was divided into 3 categories: normal (<25), overweight (25 to 29.9), and obese (≥30). Iron deficiency was defined as ≥2 of the following 3 criteria: transferrin saturation <15%, serum ferritin level <12 ng/ml, and erythrocyte protoporphyrin level >69.8 μm (1.24 μmol/L). Vitamin B12 deficiency was defined as serum vitamin B12 level <200 pg/dl (147.56 pmol/L), and folate deficiency was defined as a red blood cell folate level <102.6 ng/ml (232.49 nmol/L) or a serum folate level <2.6 ng/ml (5.89 nmol/L). Deficiency of nutritional factors was defined as the presence of iron, folate, or vitamin B12 deficiency.
Laboratory procedures in the NHANES were performed using standard procedures established by protocols published previously. In brief, RCDW, MCV, and hemoglobin were measured using the Beckman Automated Coulter Counter (Beckman Coulter, Inc., Indianapolis, Indiana) method of counting and sizing, combined with an automatic diluting and mixing device for sample processing, and a single-beam photometer for hemoglobinometry, respectively. Serum creatinine levels were measured using a multichannel analyzer (Roche, Hitachi 737; Boehringer Mannheim Diagnostics, Indianapolis, Indiana). Standardized creatinine was calculated according to the NHANES guidelines using the following formula: standard creatinine = 1.013 × serum creatinine (milligrams per deciliter) + 0.147. Estimated glomerular filtration rate was calculated using the Modification of Diet in Renal Disease method. Serum folate and vitamin B12 levels were measured using a radioassay kit (Quantaphase Folate, Bio-Rad Laboratories, Hercules, California). Serum total and high-density lipoprotein cholesterol levels were measured using the Hitachi 704 Analyzer (Boehringer Mannheim Diagnostics). C-reactive protein levels were quantified with latex-enhanced nephelometry.
Main outcome measurements of the study were all-cause death, mortality from a CV cause (CV death), and mortality from ischemic heart disease (CHD death). The International Classification of Diseases, Tenth Revision was used by the National Center for Health Statistics, which used the National Death Index to ascertain mortality status (median follow-up duration 14.5 years). All-cause death was defined as death from any underlying cause. International Classification of Diseases, Tenth Revision codes 53 to 75 were used to define CV death and codes 58 to 63 were used to define CHD death.
Statistical analyses were performed using STATA 10 (STATA Corp., College Station, Texas). Baseline characteristics were compared using standard statistical tests such as chi-square test for categorical variables and analysis of variance or Student’s t test for continuous variables, whichever was appropriate. Adjusted RCDW means were derived for gender and ethnicities from linear regression estimates adjusting for age, body mass index, hypertension, hyperlipidemia, smoking, hemoglobin, MCV, estimated glomerular filtration rate (log-transformed), C-reactive protein (log-transformed), and deficiency of nutritional factors. Spearman correlation coefficients (rho) were generated to assess the correlation between RCDW and other continuous variables. We performed Cox proportional hazard analysis to evaluate for potential gender and ethnic differences in risk association between RCDW and other outcome variables. Statistical significance for the interaction term was tested using the likelihood-ratio test for nested models with and without interaction terms for gender and RCDW and ethnicity and RCDW. We then performed separate gender-based stratified analyses where RCDW was evaluated as a continuous variable because the interaction term between RCDW and gender was found to be significant. After adjusted Cox proportional hazard analysis, models were generated where variables were added to the hazard model based on their statistical and clinical significance: model 1—adjusted for age (and gender for ethnicity-based analysis and ethnicities for gender-based analysis); model 2—adjusted for model 1 plus body mass index, hypertension, hyperlipidemia, smoking, hemoglobin, MCV, and deficiency of nutritional factors; and model 3—model 2 plus estimated glomerular filtration rate and C-reactive protein. For each model, assumption of proportionality was tested using Schoenfeld residuals. Variables that violated this assumption were stratified. Kaplan–Meier estimated survival curves were generated for study outcomes by RCDW quartiles stratified according to gender. For the entire analysis, a p value <0.05 was considered statistically significant.
Results
Table 1 lists baseline characteristics distribution across gender and race/ethnic strata. African-American men and women had higher RCDW values than their white counterparts. Women had higher serum levels of high-density lipoprotein cholesterol, total cholesterol, and C-reactive protein and lower hemoglobin, MCV, and estimated glomerular filtration rate values compared to men (p <0.001 for all).
| Women | Men | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| White | Black | Other | p Value | Overall | White | Black | Other | p Value | Overall | |
| (n = 5,644) | (n = 2,368) | (n = 247) | (n = 4,965) | (n = 1,957) | (n = 279) | |||||
| Age (years), mean ± SD | 48.9 ± 20.1 | 41.2 ± 16.5 | 40.8 ± 16.4 | <0.001 | 46.5 ± 19.4 | 47.6 ± 19.7 | 41.8 ± 16.3 | 40.9 ± 17.6 | <0.001 | 45.8 ± 19.0 |
| Hypertension | 35.9% | 35.7% | 21.1% | <0.001 | 35.5% | 34.8% | 38.2% | 26.5% | <0.001 | 35.4% |
| Hyperlipidemia | 28.5% | 21.6% | 16.6% | <0.001 | 26.1% | 22.3% | 17.1% | 19.4% | <0.001 | 20.7% |
| High-density lipoprotein cholesterol (mg/dl), mean ± SD | 55 ± 15 | 57 ± 16 | 54 ± 15 | <0.001 | 56 ± 16 | 46 ± 13 | 53 ± 18 | 45 ± 12 | <0.001 | 48 ± 14 |
| Total cholesterol (mg/dl), mean ± SD | 208 ± 45 | 200 ± 45 | 196 ± 43 | <0.001 | 206 ± 45 | 203 ± 41 | 198 ± 42 | 200 ± 42 | <0.001 | 201 ± 41 |
| Nonhigh-density lipoprotein cholesterol (mg/dl), mean ± SD | 153 ± 46 | 142 ± 45 | 142 ± 43 | <0.001 | 150 ± 46 | 157 ± 43 | 145 ± 44 | 154 ± 44 | <0.001 | 154 ± 43 |
| Current smoker | 20.5% | 28.1% | 11.7% | <0.001 | 22.4% | 29.1% | 40.9% | 32.6% | <0.001 | 32.5% |
| Family history of coronary artery disease ⁎ | 15.5% | 11.8% | 9.5% | <0.001 | 13.9% | 11.6% | 9.2% | 5.2% | <0.001 | 10.5% |
| Body mass index (kg/m 2 ) | ||||||||||
| <25 | 46.9% | 35.9% | 53.4% | <0.001 | 43.9% | 40.4% | 46.4% | 54.1% | <0.001 | 42.5% |
| 25–29.9 | 30.0% | 29.6% | 21.9% | 29.6% | 41.4% | 34.4% | 31.5% | 39.1% | ||
| ≥30 | 23.1% | 34.5% | 24.7% | 26.4% | 18.2% | 19.2% | 14.3% | 18.3% | ||
| Hemoglobin (g/dl), mean ± SD | 13.0 ± 1.1 | 12.5 ± 1.2 | 13.2 ± 1.2 | <0.001 | 13.1 ± 1.2 | 15.1 ± 1.1 | 14.5 ± 1.2 | 15.2 ± 1.1 | <0.001 | 14.9 ± 2.0 |
| Mean corpuscular volume, mean ± SD | 89.9 ± 5.1 | 86.9 ± 6.5 | 88.2 ± 6.3 | <0.001 | 89.0 ± 5.8 | 90.6 ± 4.6 | 88.4 ± 6.0 | 90.4 ± 5.2 | <0.001 | 90.0 ± 5.2 |
| C-reactive protein (mg/dl), mean ± SD † | 0.47 ± 0.8 | 0.54 ± 0.8 | 0.40 ± 0.4 | <0.001 | 0.49 ± 0.8 | 0.39 ± 0.6 | 0.41 ± 0.6 | 0.32 ± 0.4 | 0.001 | 0.39 ± 0.6 |
| Estimated glomerular filtration rate (ml/min/1.73 m 2 ), mean ± SD † | 55 ± 12 | 66 ± 12 | 62 ± 11 | <0.001 | 59 ± 13 | 61 ± 11 | 70 ± 13 | 64 ± 11 | <0.001 | 64 ± 12 |
| Nutritional factors deficiency | 21.3% | 34.3% | 20.6% | <0.001 | 25.0% | 15.3% | 26.5% | 16.5% | <0.001 | 18.4% |
| All-cause death | 21.9% | 14.2% | 8.5% | <0.001 | 19.3% | 24.7% | 21.2% | 13.3% | <0.001 | 23.3% |
| Cardiovascular disease death ‡ | 9.9% | 5.8% | 2.4% | <0.001 | 8.5% | 10.0% | 6.8% | 5.4% | <0.001 | 8.9% |
| Coronary heart disease death § | 5.2% | 2.6% | 0.8% | <0.001 | 4.3% | 5.9% | 3.2% | 2.5% | <0.001 | 5.0% |
⁎ Defined as any relative who had a heart attack at <50 years of age.
‡ Death resulting from cardiovascular disease, cerebrovascular disease, and atherosclerotic heart disease.
Table 2 lists baseline characteristics distribution across quartiles of RCDW. Higher quartiles were associated with a graded increase in mean values for age, serum high-density lipoprotein cholesterol, total cholesterol, and C-reactive protein levels and a decrease in mean values for hemoglobin, MCV, and estimated glomerular filtration rate. Proportions of African-Americans increased and those of whites decreased with higher RCDW quartiles. A graded increase in mortality was associated with increasing RCDW quartiles. Prevalence of hypertension, hyperlipidemia, current smoking, obesity, and nutritional factor deficiency also increased with increasing RCDW quartiles.
| RCDW Quartiles | p Value | ||||
|---|---|---|---|---|---|
| 1 (≤12.4) | 2 (>12.4–≤12.9) | 3 (12.9–≤13.5) | 4 (>13.5) | ||
| (n = 3,426) | (n = 4,212) | (n = 3,815) | (n = 4,007) | ||
| Age (years), mean ± SD | 39 ± 16 | 43 ± 18 | 48 ± 19 | 53 ± 20 | <0.001 |
| Men | 46.9% | 49.2% | 47.8% | 42.2% | <0.001 |
| Race/ethnicity | |||||
| Whites | 77.0% | 75.9% | 68.4% | 54.1% | <0.001 |
| Blacks | 18.8% | 20.1% | 28.5% | 43.5% | |
| Others | 4.2% | 3.9% | 3.1% | 2.4% | |
| Hypertension | 26.0% | 30.7% | 37.7% | 46.3% | <0.001 |
| Hyperlipidemia | 21.8% | 23.6% | 24.4% | 24.4% | 0.032 |
| High-density lipoprotein cholesterol (mg/dl), mean ± SD | 51 ± 15 | 51 ± 14 | 52 ± 16 | 54 ± 17 | <0.001 |
| Total cholesterol (mg/dl), mean ± SD | 199 ± 43 | 203 ± 42 | 206 ± 43 | 206 ± 45 | <0.001 |
| Nonhigh-density lipoprotein cholesterol (mg/dl), mean ± SD | 147 ± 43 | 152 ± 44 | 155 ± 44 | 152 ± 47 | <0.001 |
| Current smoker | 20.6% | 24.8% | 29.4% | 32.9% | <0.001 |
| Family history of coronary artery disease | 13.3% | 12.9% | 13.2% | 11.1% | 0.011 |
| Body mass index (kg/m 2 ) | |||||
| <25 | 50.2% | 44.2% | 39.8% | 39.8% | <0.001 |
| 25–29.9 | 33.6% | 35.1% | 35.9% | 31.7% | |
| ≥30 | 16.2% | 20.8% | 24.3% | 28.5% | |
| Hemoglobin (g/dl), mean ± SD | 14.2 ± 1.4 | 14.2 ± 1.4 | 14.0 ± 1.4 | 13.3 ± 1.7 | <0.001 |
| Mean corpuscular volume, mean ± SD | 90.4 ± 4.1 | 90.2 ± 4.3 | 89.8 ± 4.9 | 87.7 ± 7.4 | <0.001 |
| C-reactive protein (mg/dl), mean ± SD ⁎ | 0.35 ± 0.5 | 0.38 ± 0.6 | 0.42 ± 0.6 | 0.62 ± 1.1 | <0.001 |
| Estimated glomerular filtration rate (ml/min/1.73 m 2 ), mean ± SD ⁎ | 63 ± 12 | 61 ± 12 | 61 ± 13 | 59 ± 14 | <0.001 |
| Nutritional factors deficiency | 14.6% | 17.2% | 21.9% | 33.2% | <0.001 |
| All-cause death | 9.3% | 15.3% | 22.4% | 36.3% | <0.001 |
| Cardiovascular disease death † | 3.3% | 6.3% | 9.2% | 15.5% | <0.001 |
| Coronary heart disease death ‡ | 1.8% | 3.4% | 4.8% | 8.3% | <0.001 |
⁎ Sample size for model 3 excluded subjects with missing values on estimated glomerular filtration rate and C-reactive protein levels (n = 10,959).
† Death resulting from cardiovascular disease, cerebrovascular disease, and atherosclerotic heart disease.
Figure 1 shows that mean values of RCDW were higher for men and women of African-American ethnicities compared to their counterparts after adjusting for variables described earlier. Table 3 presents Spearman correlations between RCDW and other continuous variables across gender and racial/ethnic strata. As presented, RCDW had a moderate but significant degree of correlation with age but a weak but significant correlation with C-reactive protein and estimated glomerular filtration rate in men. In women, age, body mass index, C-reactive protein, and estimated glomerular filtration rate correlated weakly but significantly with RCDW. Table 4 presents the effect of gender on the association between RCDW and mortality. The interaction term between RCDW and gender was statistically significant for all outcome measurements (p for interaction <0.05). Each unit (0.1%) increase in RCDW in women and men was significantly associated with higher adjusted hazard ratios for all-cause mortality, CV mortality, and CHD mortality. Figure 2 shows Kaplan–Meier survival curve estimates for study outcomes across RCDW quartiles for women and men. In quartile 4 (RCDW >13.5) men had significantly worse survival compared to women.
| Women | Men | |||||
|---|---|---|---|---|---|---|
| White | Black | Other | White | Black | Other | |
| (n = 5,644) | (n = 2,368) | (n = 247) | (n = 4,965) | (n = 1,957) | (n = 279) | |
| Age (years) | 0.26 ⁎ | 0.23 ⁎ | 0.12 § | 0.39 ⁎ | 0.34 ⁎ | 0.31 ⁎ |
| Body mass index (kg/m 2 ) | 0.14 ⁎ | 0.16 ⁎ | 0.13 ‡ | 0.02 § | 0.01 § | 0.09 § |
| Total cholesterol (mg/dl) | 0.09 ⁎ | 0.07 ⁎ | 0.06 § | 0.07 ⁎ | 0.07 † | 0.04 § |
| High-density lipoprotein cholesterol (mg/dl) | −0.03 ‡ | −0.04 § | −0.01 § | 0.07 ⁎ | 0.05 ‡ | −0.01 § |
| C-reactive protein (mg/dl) | 0.14 ⁎ | 0.15 ⁎ | 0.10 § | 0.21 ⁎ | 0.25 ⁎ | 0.16 ‡ |
| Estimated glomerular filtration rate (ml/min/1.73 m 2 ) | −0.14 ⁎ | −0.12 † | 0.02 § | −0.24 ⁎ | −0.11 ⁎ | −0.19 † |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
