Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2013 Sep 16.
Published in final edited form as: J Androl. 2012 May 17;33(6):1332–1341. doi: 10.2164/jandrol.111.015651

Association Between Sex Steroid Hormones and Hematocrit in a Nationally Representative Sample of Men

Channing J Paller *, Meredith S Shiels †, Sabine Rohrmann ‡, Andy Menke †, Nader Rifai §, William G Nelson *,║, Elizabeth A Platz *,†,║, Adrian S Dobs *,║,
PMCID: PMC3774012  NIHMSID: NIHMS448338  PMID: 22604627

Abstract

Low or high hematocrit levels are associated with increased morbidity and mortality, mediated via anemia or thromboembolic events, respectively. It is therefore important to identify factors that influence hematocrit. Although androgens are known to stimulate hematopoietic cells, it is unknown whether circulating sex steroid hormones affect hematocrit. The association between serum sex steroid hormone concentrations and hematocrit in men aged ≥20 years was evaluated in a cross-sectional study of 1273 men in the Third National Health and Nutrition Examination Survey (1988–1991). Outcomes were low (<10th percentile), high (>90th percentile), and mean hematocrit. Men with low free testosterone levels had a lower hematocrit than men with normal free testosterone levels (P = .03), although no relationship was found between total testosterone level and hematocrit. The relationship between sex hormone–binding globulin (SHBG) and hematocrit was complex, with both low (P < .001) and high (P = .01) SHBG levels associated with lower hematocrit in men aged ≥20 years and only high (P = .01) SHBG levels in men aged ≥50 years. The odds ratio (OR) of high vs normal hematocrit increased as total estradiol (OR, 2.84; P trend = .04) and free estradiol (OR, 2.23; P trend = .09) levels increased. In this nationally representative study of men, sex steroid hormone levels, particularly low free testosterone and high SHBG levels, were associated with lower hematocrit, and high total and free estradiol levels were associated with high hematocrit. Thus, changes in sex hormone levels with aging may contribute to the increased prevalence of anemia and thromboembolic stroke in men as they age.

Keywords: Testosterone, estradiol, sex hormone–binding globulin, NHANES III

Both low and high levels of hematocrit have been observed to be associated with adverse health outcomes. Specifically, anemia, defined as low hemoglobin (<13 g/dL), is associated with an increased risk of falls and fractures, cognitive impairment, mortality, and decreased physical ability (Penninx, 2007). Anemic men older than 85 years had more than twice the risk of death than men without anemia (Izaks et al, 1999). In the Third National Health and Nutrition Examination Study (NHANES III), a nationally representative sample of Americans, anemia was present in 11% of men older than age 65 years (Guralnik et al, 2004). Among hospitalized older patients, the prevalence of anemia is even higher (24% to 40%) (Joosten et al, 1992; Smieja et al, 1996). High hematocrit was associated with an increased risk of death from cardiovascular disease in both the Honolulu Heart Program (Carter et al, 1983) and Framingham Heart Study (Gagnon et al, 1994). In addition, in disease states with elevated red cell mass, as in polycythemia vera, the prevalence of thrombosis ranges from 12% to 39% (Elliott and Tefferi, 2005).

It is unknown why anemia is so common in elderly men. Changes in sex steroid hormone levels with age may alter hematocrit and could contribute to this finding. It is well documented that with increasing age, serum testosterone concentrations decline in men and the prevalence of clinically low testosterone levels increases (Harman et al, 2001). There is some evidence that suggests that androgens raise hematocrit. For example, hemoglobin levels rise in boys during puberty (Garn et al, 1981) and with pharmacological doses of androgens in patients with hypogonadism (Tenover, 1992; Sih et al, 1997; Saad et al, 2007), whereas hemoglobin levels decline with chemical or surgical castration (Weber et al, 1991; Fonseca et al, 1998). Additionally, previous research has shown that men older than 65 years with low testosterone levels have a higher risk of anemia (Ferrucci et al, 2006). However, these studies have focused on select populations (eg, pubescent, hypogonadal, castrated, and elderly men), effectively assessing the association among men with very high or low levels of testosterone. Estrogen levels also decline with age in men, although not to the same extent as testosterone levels (Orwoll et al, 2006). There is no known mechanism by which estradiol directly increases hematocrit; however, estradiol may blunt the effects of hypoxia-induced erythropoietin expression (Mukundan et al, 2002). In addition, men have higher age-matched hemoglobin concentrations than women at particular altitudes; this effect is seen after puberty but is not evident at postmenopausal ages (Leon-Velarde et al, 2000).

We hypothesized that sex steroid hormone levels in men, in both diseased and healthy populations, are associated with hematocrit levels, leading to alterations in risk of clinical and subclinical disease, respectively (Figure). In this study, we evaluated the association of serum levels of testosterone, estradiol, and their main binding globulin, sex hormone–binding globulin (SHBG) with hematocrit across the full male adult age range using data from a national population-based study.

Figure.

Figure

Open in a new tab

Possible explanations for the observation that changes in hormone levels may alter hematocrit, leading to a change in health outcomes.

Materials and Methods

Study Population

NHANES III, a cross-sectional study of the US population aged ≥2 months, was conducted between 1988 and 1994 by the National Center for Health Statistics. NHANES III used a multistage, stratified, clustered probability sample with over-sampling of Mexican Americans, non-Hispanic African Americans, and older Americans to ensure an adequate sample size for generating prevalence estimates for these groups. In total, 33 944 people participated, of which 14 781 were male. The protocols for the conduct of NHANES III were approved by the institutional review board of the National Center for Health Statistics, US Centers for Disease Control and Prevention. Informed consent was obtained from all participants. The assay of these stored serum specimens for sex steroid hormones was approved by institutional review boards at the Johns Hopkins Bloomberg School of Public Health and the National Center for Health Statistics, US Centers for Disease Control and Prevention.

NHANES III was conducted in 2 phases (1988–1991 and 1991–1994). Participants were randomly assigned to morning or afternoon examination sessions. Unbiased national estimates of health and nutrition characteristics can be generated when using either phase and the morning or afternoon sessions. For our analysis, 2205 men aged ≥20 years who attended the morning examination session of the first phase were eligible for inclusion. We selected the morning session to decrease the variance due to diurnal sex hormone production. The NHANES III repository held serum from 1470 of these men. We excluded men with a self-reported history of cancer (n = 40); missing testosterone, estradiol, or SHBG concentrations (n = 21); missing data on hematocrit (n = 26); and missing data on percent body fat, physical activity, smoking status, or number of cigarettes smoked per day (n=110), leaving a final study population of 1273 men.

Measurement of Hematocrit

Detailed documentation of the laboratory methods used to measure hematocrit in NHANES III has been published (Gunter, 1996). Whole blood was collected in tubes containing EDTA, and a complete blood count was performed (Gunter, 1996). Hematocrit was calculated using a Coulter S-Plus Jr electronic counter (Coulter Electronics, Hialeah, Florida) (Gunter, 1996).

Sex Steroid Hormones

Sex steroid hormone concentrations were measured in blood samples that were centrifuged, aliquoted, and subsequently stored at −70°C in Dr Nader Rifai's laboratory (Children's Hospital, Boston, Massachusetts). Competitive electrochemiluminescence immunoassays were performed on a 2010 Elecsys autoanalyzer (Roche Diagnostics, Indianapolis, Indiana) to measure serum concentrations of testosterone, estradiol, and SHBG. Laboratory technicians were blinded to sample identifiers and participant characteristics. The lowest detection limits of the assay were testosterone, 20 ng/dL; estradiol, 5 pg/mL; and SHBG, 3 nmol/L. Coefficients of variation for quality control specimens were testosterone, 5.9% and 5.8% at 2.5 and 5.5 μg/L, respectively; estradiol, 6.5% and 6.7% at 102.7 and 474.1 μg/L, respectively; and SHBG, 5.3% and 5.9% at 5.3 and 16.6 nmol/L, respectively. In a separate quality control sample run for this study, the interassay coefficient of variation was 2.5% for an estradiol concentration of 39.4 pg/mL, which is in the range of typical male estradiol concentrations. Serum hormone concentrations detected in the men in NHANES III were generally within the range of reference values used for men (testosterone, 194–833 ng/dL; estradiol, ≤50 pg/mL [Beers et al, 2006]; and SHBG, 13–71 nmol/L [Fauci et al, 2008]). Serum total testosterone, total estradiol, SHBG, and albumin concentrations were used to estimate free testosterone and free estradiol levels with mass action equations (Vermeulen et al, 1999; Rinaldi et al, 2002).

Assessment of Covariates

Covariates considered in the analysis were those known to influence either hormone concentrations or hematocrit. Participants in NHANES III provided information on age, race/ethnicity, smoking status, number of cigarettes smoked per day, and physical activity. A standardized method was used to code and classify each type of physical activity by rate of energy expenditure (Ainsworth et al, 1993; Rohrmann et al, 2005). Weekly frequency of moderate and vigorous walking, jogging, or running; biking; swimming; aerobics; dancing; calisthenics; gardening; and lifting weights was measured. Other physical activities were considered to be at the moderate level if they met age-specific cutoffs of the metabolic equivalent (MET) of the activity when compared with rest: ≥3 METs for ages 20 to 39 years; ≥2.5 METs for ages 40 to 64 years; ≥2.0 METs for ages 65 to 79 years, and ≥ 1.26 METs for ages ≥79 years. Levels of moderate physical activities were categorized as follows: 0, 0 to 4.7, and ≥4.8 times/week. Additionally, those who were able to walk 1 mile without stopping, irrespective of whether the intent was exercise, were categorized as being physically active (>0 times/week). Percent body fat was calculated from bioelectric impedance, height, and weight measurements using previously published formulas (Plan and operation of the Third National Health and Nutrition Examination Survey, 1994; Chumlea et al, 2002). A trained examiner measured height and weight. Bioelectrical impedance was estimated using a 1990B Bio-Resistance body composition analyzer (Valhalla Scientific, San Diego, California).

Statistical Analysis

Statistical analyses were performed using SUDAAN (Research Triangle Park, North Carolina) as implemented in SAS version 9.1 (Cary, North Carolina) software. In each analysis, we applied sampling weights. Those in the highest 10th percentile of hematocrit (≥47.7%) were categorized as having high hematocrit, those in the lowest 10th percentile of hematocrit (<41.0%) were categorized as having low hematocrit, and those with hematocrit between 41.0% and 47.6% were categorized as having normal hematocrit. Select participant characteristics were compared among those with low, normal, and high hematocrits. Both crude and age-adjusted means and proportions were calculated.

Total testosterone, total estradiol, SHBG, free testosterone, and free estradiol levels were categorized into tertiles based on the distributions among all men included in the analysis. Multivariable logistic regression was used to estimate the odds ratios (OR) and 95% confidence intervals of low hematocrit compared with normal hematocrit and high hematocrit compared with normal hematocrit by tertile of hormone concentrations, with the lowest hormone tertile as the reference group. The models were adjusted for age, race/ ethnicity, smoking status (current, former, never) and number of cigarettes smoked per day by current smokers, physical activity, percent body fat, and other hormone levels (tertiles of total estradiol, total testosterone, and SHBG levels were simultaneously included in a model, free testosterone was adjusted for tertiles of total estradiol, and free estradiol was adjusted for tertiles of total testosterone).

The mean hematocrit levels of men with low and high concentrations of testosterone, estradiol, SHBG, free testosterone, and free estradiol were compared with the mean hematocrit levels of men with normal concentrations of these hormones. Cut points for low hormone concentrations were based on either clinical cut points or the 10th percentile of the distribution: total testosterone, <250 ng/dL (Greenspan, 1994); total estradiol, <20 pg/mL (Greenspan, 1994); SHBG, <13 nmol/L (Bukowski et al, 2000); free testosterone, <5 ng/dL (Greenspan, 1994); and free estradiol, <0.59 pg/mL (,10th percentile). The cut points for high hormone concentrations were defined as ≥90th percentile as follows: total testosterone, ≥800 ng/dL; total estradiol, ≥50.0 pg/mL; SHBG, ≥66 nmol/L; free testosterone, ≥160 ng/dL; and free estradiol, ≥1.37 pg/mL. Multivariable linear regression was used to estimate mean hematocrit concentrations, adjusted for age, race/ethnicity, percent body fat, current smoking status and number of cigarettes smoked per day by current smokers, and frequency of physical activity and mutually adjusted for other hormones. A subanalysis was carried out, restricting the analysis to men aged ≥50 years because these men were more likely to have lower testosterone, estradiol, and hematocrit levels.

We repeated the analysis excluding men with chronic kidney disease (CKD) because this condition is a common cause of anemia and then further stratified by the presence or absence of major chronic diseases to assess whether prevalent disease explained the association between hormones and hematocrit (ie, reverse causation) based on our hypothesis in the Figure. We considered prevalent chronic disease to be a history of diabetes, emphysema, chronic bronchitis, stroke, myocardial infarction, or congestive heart failure.

Results

Selected characteristics of this study population are shown in Table 1. After age adjustment, the mean hematocrits in those with low, normal, and high levels were 39.3%, 44.3%, and 48.9%, respectively. The mean age was 41.5 years. Men with low hematocrits were significantly older (51.3 years) than those with normal hematocrits (40.5 years), and men with high hematocrits were significantly younger (39.4 years) than those with normal hematocrits (P < .05). After adjustment for age, men with low hematocrits were more likely to be non-Hispanic African American and less likely to be non-Hispanic white or Mexican American than those with normal hematocrits (P < .05). There was no difference in percent body fat by hematocrit category. Men with high hematocrits were more likely to be current smokers, in agreement with prior analyses (Whitehead et al, 1995). After age adjustment, those with high hematocrits were less likely to be physically active than those with normal hematocrits (P < .05), consistent with previous findings (Wannamethee et al, 2002).

Table 1. Selected characteristics of 1273 men in the morning session of Phase I of NHANES III overall and for those with low (<41%) and high (≥47.7%) hematocrit as compared with normal hematocrit (41% to 47.6%).

Characteristica Total (n = 1273) Unadjusted Values Age-Adjusted Values
Low Hematocrit (n = 187) Normal Hematocrit (n = 964) High Hematocrit (n = 122) Low Hematocrit (n = 187) Normal Hematocrit (n = 964) High Hematocrit (n = 122)
Age, yb 41.5 (0.7) 51.3 (1.6) 40.5 (0.8) 39.4 (1.6) … … …
Age categories, %b
 20–29 y 26.4 (2.4) 12.2 (3.6) 27.9 (3.1) 28.4 (6.0) … … …
 30–39 y 26.2 (2.4) 14.6 (3.9) 27.8 (2.7) 25.2 (5.3) … … …
 40–49 y 19.9 (1.5) 22.5 (4.2) 19.5 (1.8) 20.5 (6.4) … … …
 50–59 y 12.0 (1.1) 18.8 (2.5) 10.4 (1.3) 17.5 (4.6) … … …
 60–69 y 9.4 (1.3) 15.6 (4.0) 9.1 (1.2) 5.9 (1.7) … … …
 70–79 y 4.7 (0.6) 11.9 (2.9) 4.1 (0.6) 2.1 (1.0) … … …
 ≥80 y 1.4 (0.3) 4.5 (1.5) 1.1 (0.2) 0.5 (0.3) … … …
Race/ethnicity, %b,d
 Non-Hispanic white 77.8 (3.2) 73.4 (4.4) 78.5 (3.2) 76.2 (6.7) 70.5 (5.0) 79.1 (0.03) 76.8 (6.7)
 Non-Hispanic 9.4 (1.4) 21.0 (3.3) 8.4 (1.3) 5.3 (1.5) 23.5 (4.0) 8.2 (1.2) 5.1 (1.4)
 African American
 Mexican American 5.0 (0.8) 3.0 (0.8) 5.2 (0.9) 5.7 (1.2) 3.5 (1.0) 4.9 (0.9) 5.3 (1.1)
 Other 7.9 (2.2) 2.6 (1.5) 7.9 (2.2) 12.7 (6.3) 2.5 (1.6) 7.9 (2.2) 12.8 (6.3)
Hematocrit, %b,c,d,e 44.3 (0.2) 39.2 (0.1) 44.4 (0.1) 48.9 (0.1) 39.3 (0.1) 44.3 (0.1) 48.9 (0.1)
Percent body fat, % 24.9 (0.3) 24.5 (0.7) 24.8 (0.3) 25.9 (0.8) 23.7 (0.8) 24.9 (0.3) 26.0 (0.8)
Smoking, %b,c,e
 Never 35.0 (2.4) 28.6 (5.3) 36.8 (2.6) 27.6 (6.9) 34.6 (5.8) 36.4 (5.8) 27.0 (6.8)
 Former 31.0 (2.8) 48.1 (3.6) 30.7 (3.0) 19.5 (6.2) 38.6 (4.0) 30.3 (2.9) 17.3 (6.2)
 Current 34.0 (2.3) 23.3 (4.6) 32.6 (2.5) 55.0 (6.8) 26.8 (5.5) 33.3 (2.6) 55.7 (6.7)
Cigarettes smoked per day by current smokersb,d 19.3 (1.0) 15.2 (1.5) 19.2 (1.3) 21.0 (1.1) 14.6 (1.5) 19.3 (1.2) 20.9 (0.9)
Physical activity, %e
 0 times/wk 8.4 (1.4) 10.4 (2.0) 7.9 (1.6) 10.2 (4.2) 8.8 (1.8) 7.7 (1.3) 12.8 (3.5)
 <4.8 times/wk 40.6 (2.3) 40.4 (3.7) 39.0 (2.3) 52.3 (7.9) 41.9 (3.2) 39.4 (2.4) 48.2 (3.5)
 ≥4.8 times/wk 51.1 (3.0) 49.2 (4.1) 53.1 (3.0) 37.5 (7.7) 49.3 (4.5) 52.9 (3.1) 39.0 (6.4)
Open in a new tab

Abbreviation: NHANES III, Third National Health and Nutrition Examination Survey.

a

Mean or percentage (standard error).

Indicates a significant unadjusted difference between

b

low and normal hematocrit and

c

high and normal hematocrit (P < .05).

Indicates a significant age-adjusted difference between

d

low and normal hematocrit and

e

high and normal hematocrit (P < .05)

Table 2 presents ORs of low (<41%) and high hematocrits (≥47.7%) across tertiles of total and free estradiol, total and free testosterone, and SHBG levels after multivariable adjustment. Men with concentrations of total estradiol in the highest tertile had almost 3 times the odds of high hematocrit (OR, 2.84; 95% confidence intervals, 1.05, 7.66; P trend = .04) than men in the lowest tertile. The odds of high hematocrit also increased with increasing tertiles of free estradiol level (P trend = .09). Although not statistically significant, there was a suggestion that men in the highest tertiles of total and free testosterone and total estradiol levels had decreased odds of low hematocrit (Table 2).

Table 2. Associationa of tertiles of hormones with clinically low (<41%) and high (≥47.7%) hematocrit in men in the morning session of Phase I of NHANES III.

Tertile Range Hematocrit
Low vs Normal High vs Normal
OR 95% CI OR 95% CI
Total testosterone, ng/dL 1 <433 1.00 Reference 1.00 Reference
2 433–600 1.08 0.48, 2.39 1.69 0.82, 3.48
3 ≥601 0.63 0.30, 1.32 1.16 0.45, 3.03
P trend = .2 P trend = 1.0
Total estradiol, pg/mL 1 <31.5 1.00 Reference 1.00 Reference
2 31.5–40.7 0.78 0.35, 1.72 1.50 0.57, 3.92
3 ≥40.8 0.61 0.33, 1.14 2.84 1.05, 7.66b
P trend = .1 P trend = .04
SHBG, nmol/L 1 <29.2 1.00 Reference 1.00 Reference
2 29.2–44.8 1.69 0.80, 3.57 0.81 0.23, 2.87
3 ≥44.9 1.63 0.85, 3.13 1.31 0.29, 5.90
P trend = .2 P trend = .7
Free testosterone, ng/dL 1 <8.4 1.00 Reference 1.00 Reference
2 8.4–12.0 0.52 0.24, 1.14 0.72 0.23, 2.21
3 ≥12.1 0.63 0.28, 1.45 0.64 0.18, 2.26
P trend = .2 P trend = .5
Free estradiol, pg/mL 1 <0.794 1.00 Reference 1.00 Reference
2 0.794–1.04 0.86 0.42, 1.80 1.11 0.46, 2.68
3 ≥1.05 0.81 0.39, 1.68 2.23 0.79, 6.24
P trend = .5 P trend = .09
Open in a new tab

Abbreviations: CI, confidence intervals; NHANES III, Third National Health and Nutrition Examination Survey; OR, odds ratio; SHBG, sex hormone–binding globulin.

a

Adjusted for age, race/ethnicity, smoking status, cigarettes smoked per day by current smokers, physical activity, percent body fat, and mutual adjustment for other hormones.

b

P < .05.

We repeated these analyses excluding men with CKD, and the results were similar (see Supplemental Table, available online at www.andrologyjournal.org). When we excluded men with CKD and then stratified by prevalent chronic disease, the associations of total (P trend = .04) and free estradiol (P trend = .07) levels with high hematocrit in men without prevalent chronic disease were similar to the overall findings, whereas these associations were not present in men with prevalent chronic diseases (Table 3).

Table 3. Associationa of tertiles of hormones with clinically low (<41%) and high (>47.7%) hematocrit in men with (n = 182) and without (n = 996) prevalent chronic disease and who do not have chronic kidney disease in the morning session of Phase I of NHANES III.

Tertile Range Hernatocrit
LOW Vs Normal High vs Normal
None Prevalent Chronic Disease None Prevalent Chronic Disease
OR 95% CI OR 95% CI OR 95% CI OR 95% CI
Total 1 <433 1.00 Reference 1.00 Reference 1.00 Reference 1.00 Reference
 testosterone, 2 433–600 1.37 0.52, 3.58 0.53 0.14, 1.95 1.42 0.61, 3.26 2.50 0.26, 23.62
 ng/dL 3 ≥601 0.64 0.29, 1.43 0.45 0.04, 4.82 1.01 0.35, 2.92 3.84 0.08, 183.60
P trend = .1 P trend = .4 P trend = .9 P trend = .4
Total estradiol, 1 <31.5 1.00 Reference 1.00 Reference 1.00 Reference 1.00 Reference
 pg/mL 2 31.5–40.7 0.95 0.36, 2.50 0.42 0.16, 1.10 1.47 0.53, 4.03 2.87 0.52, 15.65
3 ≥40.8 0.78 0.37, 1.64 0.40 0.06, 2.66 3.03b 1.08, 8.52 0.95 0.06, 14.53
P trend = .5 P trend = .3 P trend = .04 P trend = .6
SHBG, nmol/L 1 <29.2 1.00 Reference 1.00 Reference 1.00 Reference 1.00 Reference
2 29.2–44.8 1.94 0.77, 4.87 3.68 0.62, 21.97 0.68 0.18, 2.60 6.56 0.33, 130.47
3 ≥44.9 1.65 0.70, 3.90 4.59 0.84, 25.14 1.18 0.22, 6.45 6.15 0.52, 70.75
P trend = .3 P trend = .09 P trend = .9 P trend = .4
Free testosterone 1 <8.4 1.00 Reference 1.00 Reference 1.00 Reference 1.00 Reference
 ng/dL 2 8.4–12.0 0.48 0.20, 1.15 0.68 0.15, 3.09 0.53 0.16, 1.77 1.17 0.23, 5.99
3 ≥12.1 0.53 0.22, 1.28 2.81 0.36, 21.71 0.46 0.12, 1.74 2.21 0.17, 28.83
P trend = .2 P trend = .4 P trend = .3 P trend = .5
Free estradiol. 1 <0.794 1.00 Reference 1.00 Reference 1.00 Reference 1.00 Reference
 pg/mL 2 0.794–1.04 0.79 0.28, 2.22 1.08 0.28, 4.19 1.03 0.37, 2.83 2.37 0.26, 21.27
3 ≥1.05 0.94 0.37, 2.34 0.50 0.11, 2.21 2.41 0.84, 6.93 1.11 0.10, 12.88
P trend = .7 P trend = .3 P trend = .07 P trend = .9
Open in a new tab

Abbreviations: CI, confidence intervals; NHANES III, Third National Health and Nutrition Examination Survey; OR, odds ratio; SHBG, sex hormone–binding globulin.

a

Adjusted for age, race/ethnicity, smoking status, cigarettes smoked per day, physical activity, percent body fat, and mutual adjustment for other hormones. Prevalent chronic disease includes history of diabetes, emphysema, chronic bronchitis, stroke, myocardial infarction, or congestive heart failure.

b

P < .05.

Mean hematocrit did not differ in men who had low, normal, or high total testosterone concentrations (Table 4). However, men with low free testosterone concentrations had a lower hematocrit (43.6%) compared with men in the normal range of free testosterone (44.3%; P = .03); the finding was similar when the analysis was restricted to men aged ≥50 years (42.4% vs 43.7%, respectively; P = .008). Hematocrit levels were significantly lower in men with both low SHBG (hematocrit, 41.4%; P < .001) and high SHBG (hemat-ocrit, 43.8%; P = .01) levels as compared with normal SHBG levels (hematocrit, 44.4%). For men aged ≥50 years, those with high SHBG concentrations had lower hematocrits than men with normal SHBG levels (42.7% vs 43.8%; P = .01) (Table 4).

Table 4. Mean hematocrit among men with low, normal, and high concentrations of hormones in the morning session of Phase I of NHANES III.

≥20 Years ≥50 Years
n Hematocrit, % (SE)a P n Hematocrit, % (SE)a P
Total testosterone, ng/dL Low: <250 77 44.1 (0.46) .5 58 43.4 (0.39) .6
Normal: 250–800 1065 44.3 (0.14) Reference 440 43.6 (0.15) Reference
High: ≥801 131 44.3 (0.46) .9 30 43.3 (0.67) .3
Total estradiol, pg/mL Low: <20.0 37 45.1 (0.40) .06 17 44.0 (0.67) .6
Normal: 20.0–50.0 1080 44.3 (0.14) Reference 454 43.6 (0.16) Reference
High: ≥50.0 156 44.7 (0.26) .07 57 43.7 (0.48) .9
SHBG, nmol/L Low: <13 19 41.4 (0.78) ,.001 3 42.4 (2.04) .5
Normal: 13–66 1126 44.4 (0.14) Reference 423 43.8 (0.17) Reference
High: ≥66 128 43.8 (0.23) .01 102 42.7 (0.32) .01
Free testosterone, ng/dL Low: <50 100 43.6 (0.34) .03 88 42.4 (0.43) .008
Normal: 50–160 1053 44.3 (0.15) Reference 434 43.7 (0.16) Reference
High: ≥161 120 44.6 (0.48) .7 6 45.9 (1.38) .1
Free estradiol, pg/mL Low: <0.59 123 43.8 (0.28) .07 73 43.7 (0.57) .9
Normal: 0.59–1.37 1041 44.3 (0.15) Reference 419 43.6 (0.20) Reference
High: ≥1.37 132 44.7 (0.29) .2 36 43.0 (0.55) .3
Open in a new tab

Abbreviations: NHANES III, Third National Health and Nutrition Examination Survey; SE, standard error; SHBG, sex hormone–binding globulin.

a

Adjusted for age, race, percent body fat, smoking status, cigarettes per day, physical activity, and mutually adjustment for other hormones.

Table 5 summarizes the most important findings in this manuscript.

Table 5. Summary of key findings.

Hormone Level Association With Hematocrit Significance
Low free testosterone Lower hematocrit P = .03
High free estradiol Higher hematocrit P trend = .09 (not significant)
High total estradiol Higher hematocrit P trend = .04
Low and high SHBG in men ≥20 years Lower hematocrit P ≤ .001 and P = .01, respectively
High SHBG in men ≥50 years Lower hematocrit P = .01
Open in a new tab

Abbreviation: SHBG, sex hormone–binding globulin.

Discussion

Both low and high hematocrit levels have detrimental health effects. Increased mortality has been documented in patients with anemia, as well as those with erythrocytosis, which may increase predilection for thrombosis through multiple mechanisms (Gagnon et al, 1994; Izaks et al, 1999; Elliott and Tefferi, 2005; Penninx, 2007). Thus, it is important to identify predictors of both high and low levels of hematocrit among men. In this nationally representative sample of men, we found that men with clinically low free testosterone levels had lower hematocrits than men with the normal range of free testosterone. We also observed that high total and free estradiol concentrations were associated with high hematocrit. In addition, both low and high SHBG levels were associated with low hematocrit; the association between high SHBG level and low hematocrit was similar when the analysis was restricted to men aged ≥50 years, whereas the association between low SHBG level and hematocrit was nonsignificant. Although the observed differences in hematocrit were often small (1% to 3%), they were statistically significant, and changes of as little as 2% in hematocrit have been shown to be associated with significant increases (≥20%) in the frequency of major ischemic heart disease events (Carter et al, 1983; Wannamethee et al, 1994).

Our finding that men with clinically low concentrations of free testosterone (<5 ng/dL) had significantly lower hematocrits than those with free testosterone in the normal range is compatible with findings in clinical studies that consistently show an association between low total testosterone level, via surgical or chemical depletion, and low hematocrit (Weber et al, 1991; Fonseca et al, 1998). Our findings are also consistent with those of Ferrucci et al (2006), who in the Tuscany InCHIANTI study of men aged ≥65 years, found that those with low total and bioavailable testosterone concentrations had a higher odd of anemia as assessed by hemoglobin and that men with low baseline bioavailable testosterone had an increased risk of developing anemia. In that study, men with lower total testosterone levels had 5 times the odds of anemia than those with higher total testosterone levels and those with lower free testosterone levels had 13 times the odds of anemia than those with higher free testosterone levels. The smaller OR for total testosterone levels seen by Ferrucci et al (2006) and the lack of significant effect of lower total testosterone levels in our study could be explained by the high variability of older men in SHBG and albumin levels, both of which have strong effects on the bioactive portion of total testosterone. With respect to men aged ≥50 years in our analysis, low free testosterone levels were also associated with a lower hematocrit. The stronger associations observed in the InCHIANTI study than in our study may be due to the older age and thus lower testosterone levels of the men. Unlike clinical studies in which pharmacological doses of exogenous testosterone were administered, resulting in an elevated hematocrit (Tenover, 1992; Sih et al, 1997; Snyder et al, 2000; Saad et al, 2007), we did not observe an increased risk of high hematocrit levels in the highest free or total testosterone tertiles. Similar to our analysis, Yeap et al (2009) explored a cross-sectional analysis of 492 men from Australia aged 30 to 94 years and found that overall, hemoglobin and SHBG levels were inversely correlated. In addition, free testosterone levels were associated with hemoglobin levels, consistent with our findings (Yeap et al, 2009). We hypothesized that androgens would be associated with hematocrit because they are known to activate hematopoietic stem cells and stimulate erythropoietin production, thereby regulating hematopoiesis (Besa, 1994). We did not have a hypothesis about the association between estradiol and hematocrit because to our knowledge, there is no direct mechanistic link between estradiol and erythropoiesis. Yet, we found that men with higher concentrations of total and free estradiol were less likely to have low hematocrits and more likely to have high hematocrits, even after adjustments for total testosterone level and percent body fat. We adjusted for testosterone, which is converted to estradiol via catalysis by aromatase in fat, to be able to determine the independent association for estradiol. Our results for estradiol are compatible with a clinical study of testosterone supplementation in hypogonadal men that found that 12 of 14 men with an elevated hematocrit also had elevated estradiol levels (Dobs et al, 1999). As previously mentioned, estradiol blunts the effects of hypoxia on erythropoietin production in women but has unclear significance in men and would not explain our findings; thus, another possibility is that these findings occurred by chance.

The relationship between SHBG and hematocrit appears to be complex. Consistent with a prior Japanese study of men aged 52 to 74 years (Kato et al, 1992), in our study, high SHBG levels were associated with lower hematocrit levels overall and among men aged ≥50 years. The finding that low SHBG concentrations were associated with lower hematocrit in our overall study population is a novel finding to our knowledge. There is no known mechanism for this finding; however, the different trends seen with age could relate to the fact that SHBG starts to rise in the fourth and fifth decades in men (Feldman et al, 2002) and steroid negative feedback is known to change as men age (Winters and Wang, 2009). In addition, the association between SHBG and hematocrit could be affected by confounding disease states such as liver disease.

Based on previous analysis of the NHANES III data, anemia appears to be related to a deficiency in iron, vitamin B12, and/or folate in one-third of cases, renal insufficiency in one-third, and unexplained anemia in the remaining third of patients (Guralnik et al, 2004). Although renal insufficiency is a known cause of low erythropoietin production, in multivariable-adjusted models, the odds of a glomerular filtration rate <60 mL/min/1.73 m2 did not change with tertiles of hormones with the exception of high free estradiol levels having 3 times the odds of a decreased glomerular filtration rate (Yi et al, 2009). In sensitivity analyses, we excluded men with chronic kidney disease and the result was not remarkably changed (see Supplemental Table).

In healthy men, altered hematocrit may mediate the association between hormones and health outcomes; however, in men with chronic disease, the disease may alter both hematocrit and hormone levels. Thus, we stratified by the presence of major chronic disease. In men without prevalent chronic disease, the association between high estradiol levels and high hematocrit was similar to that of the main analysis; however, in the men with prevalent chronic disease, no associations were observed between any of the hormone levels and hematocrit. Thus, there was no evidence supporting our hypothesis that disease status will alter both hematocrit and hormone levels; however, we could not take into account the duration or severity of these diseases or the nature or success of the treatments for these diseases.

To our knowledge, this is the first study to examine the association between sex steroid hormones and hematocrit in a large, nationally representative sample of the US adult male population. The men in this study were aged ≥20 years, allowing the association between hormones and hematocrit to be evaluated in an age range wider than that in previous studies of older men. Similarly, this analysis benefited from the standardized measurement of demographic characteristics, anthropometric measures, and laboratory analytes. The large sample size and resulting power of the NHANES III analysis provided us with a unique opportunity to detect small differences even after adjustment for relevant covariates, including measures of smoking. Analyses of hemoglobin showed results similar to those for hematocrit (data not shown). However, the study was limited by insufficient power involving men with prevalent disease (Table 3) and men aged >50 years (Table 4). An important limitation of our study is that the cross-sectional design did not allow for discerning the temporality of the observed associations of sex steroid hormones and SHBG with hematocrit.

In this nationally representative study of men, sex steroid hormones, in particular total and free estradiol, free testosterone, and SHBG, appeared to influence hematocrit levels. Although causal relationships cannot be determined from our analysis, future studies should be conducted to address whether it might be beneficial to treat older men who have low circulating testosterone, estradiol, or SHBG levels to increase hematocrit levels and general well-being.

Acknowledgments

This is the 12th study from the Hormone Demonstration Program, which is supported by the Maryland Cigarette Restitution Fund at Johns Hopkins.

References

  1. Ainsworth BE, Haskell WL, Leon AS, Jacobs DR, Jr, Montoye HJ, Sallis JF, Paffenbarger RS., Jr Compendium of physical activities: classification of energy costs of human physical activities. Med Sci Sports Exerc. 1993;25:71–80. doi: 10.1249/00005768-199301000-00011. [DOI] [PubMed] [Google Scholar]
  2. Beers MR, Porter RS, Jones TV, Kaplan JL, Berkwits M. The Merk Manual of Diagnosis and Therapy. 18th. Whitehouse Station, NJ: Merk Research Laboratories; 2006. [Google Scholar]
  3. Besa EC. Hematologic effects of androgens revisited: an alternative therapy in various hematologic conditions. Semin Hematol. 1994;31:134–145. [PubMed] [Google Scholar]
  4. Bukowski C, Grigg MA, Longcope C. Sex hormone-binding globulin concentration: differences among commercially available methods. Clin Chem. 2000;46:1415–1416. [PubMed] [Google Scholar]
  5. Carter C, McGee D, Reed D, Yano K, Stemmermann G. Hematocrit and the risk of coronary heart disease: the Honolulu Heart Program. Am Heart J. 1983;105:674–679. doi: 10.1016/0002-8703(83)90493-3. [DOI] [PubMed] [Google Scholar]
  6. Chumlea WC, Guo SS, Kuczmarski RJ, Flegal KM, Johnson CL, Heymsfield SB, Lukaski HC, Friedl K, Hubbard VS. Body composition estimates from NHANES III bioelectrical impedance data. Int J Obes Relat Metab Disord. 2002;26:1596–1609. doi: 10.1038/sj.ijo.0802167. [DOI] [PubMed] [Google Scholar]
  7. Dobs AS, Meikle AW, Arver S, Sanders SW, Caramelli KE, Mazer NA. Pharmacokinetics, efficacy, and safety of a permeation-enhanced testosterone transdermal system in comparison with bi-weekly injections of testosterone enanthate for the treatment of hypogonadal men. J Clin Endocrinol Metab. 1999;84:3469–3478. doi: 10.1210/jcem.84.10.6078. [DOI] [PubMed] [Google Scholar]
  8. Elliott MA, Tefferi A. Thrombosis and haemorrhage in polycythaemia vera and essential thrombocythaemia. Br J Haematol. 2005;128:275–290. doi: 10.1111/j.1365-2141.2004.05277.x. [DOI] [PubMed] [Google Scholar]
  9. Fauci AS, Braunwald E, Kasper DL, Hauser SL, Longo DL, Jameson JL, Loscalzo J, editors. Harrison's Principles of Internal Medicine. 17th. New York, NY: McGraw Hill Medical; 2008. [Google Scholar]
  10. Feldman HA, Longcope C, Derby CA, Johannes CB, Araujo AB, Coviello AD, Bremner WJ, McKinlay JB. Age trends in the level of serum testosterone and other hormones in middle-aged men: longitudinal results from the Massachusetts male aging study. J Clin Endocrinol Metab. 2002;87:589–598. doi: 10.1210/jcem.87.2.8201. [DOI] [PubMed] [Google Scholar]
  11. Ferrucci L, Maggio M, Bandinelli S, Basaria S, Lauretani F, Ble A, Valenti G, Ershler WB, Guralnik JM, Longo DL. Low testosterone levels and the risk of anemia in older men and women. Arch Intern Med. 2006;166:1380–1388. doi: 10.1001/archinte.166.13.1380. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Fonseca R, Rajkumar SV, White WL, Tefferi A, Hoagland HC. Anemia after orchiectomy. Am J Hematol. 1998;59:230–233. doi: 10.1002/(sici)1096-8652(199811)59:3<230::aid-ajh8>3.0.co;2-2. [DOI] [PubMed] [Google Scholar]
  13. Gagnon DR, Zhang TJ, Brand FN, Kannel WB. Hematocrit and the risk of cardiovascular disease—the Framingham study: a 34-year follow-up. Am Heart J. 1994;127:674–682. doi: 10.1016/0002-8703(94)90679-3. [DOI] [PubMed] [Google Scholar]
  14. Garn SM, Ryan AS, Abraham S, Owen G. Suggested sex and age appropriate values for “low” and “deficient” hemoglobin levels. Am J Clin Nutr. 1981;34:1648–1651. doi: 10.1093/ajcn/34.9.1648. [DOI] [PubMed] [Google Scholar]
  15. Greenspan FBJ. Basic and Clinical Endocrinology. 4th. East Norwalk, CT: Appleton & Lange; 1994. [Google Scholar]
  16. Gunter E, Lewis BG, Koncikowski SM. Laboratory Procedures Used for the Third National Health and Nutrition Examination Survey (NHANES III) 1988–1994. Hyattsville, MD: Centers for Disease Control and Prevention; 1996. [Google Scholar]
  17. Guralnik JM, Eisenstaedt RS, Ferrucci L, Klein HG, Woodman RC. Prevalence of anemia in persons 65 years and older in the United States: evidence for a high rate of unexplained anemia. Blood. 2004;104:2263–2268. doi: 10.1182/blood-2004-05-1812. [DOI] [PubMed] [Google Scholar]
  18. Harman SM, Metter EJ, Tobin JD, Pearson J, Blackman MR. Baltimore Longitudinal Study of Aging. Longitudinal effects of aging on serum total and free testosterone levels in healthy men. Baltimore Longitudinal Study of Aging. J Clin Endocrinol Metab. 2001;86:724–731. doi: 10.1210/jcem.86.2.7219. [DOI] [PubMed] [Google Scholar]
  19. Izaks GJ, Westendorp RG, Knook DL. The definition of anemia in older persons. JAMA. 1999;281:1714–1717. doi: 10.1001/jama.281.18.1714. [DOI] [PubMed] [Google Scholar]
  20. Joosten E, Pelemans W, Hiele M, Noyen J, Verhaeghe R, Boogaerts MA. Prevalence and causes of anaemia in a geriatric hospitalized population. Gerontology. 1992;38:111–117. doi: 10.1159/000213315. [DOI] [PubMed] [Google Scholar]
  21. Kato I, Nomura A, Stemmermann GN, Chyou PH. Determinants of sex hormone levels in men as useful indices in hormone-related disorders. J Clin Epidemiol. 1992;45:1417–1421. doi: 10.1016/0895-4356(92)90203-y. [DOI] [PubMed] [Google Scholar]
  22. Leon-Velarde F, Gamboa A, Chuquiza JA, Esteba WA, Rivera-Chira M, Monge CC. Hematological parameters in high altitude residents living at 4,355, 4,660, and 5,500 meters above sea level. High Alt Med Biol. 2000;1:97–104. doi: 10.1089/15270290050074233. [DOI] [PubMed] [Google Scholar]
  23. Mukundan H, Resta TC, Kanagy NL. 17β-estradiol decreases hypoxic induction of erythropoietin gene expression. Am J Physiol Regul Integr Comp Physiol. 2002;283:R496–R504. doi: 10.1152/ajpregu.00573.2001. [DOI] [PubMed] [Google Scholar]
  24. Orwoll E, Lambert LC, Marshall LM, Phipps K, Blank J, Barrett-Connor E, Cauley J, Ensrud K, Cummings S. Testosterone and estradiol among older men. J Clin Endocrinol Metab. 2006;91:1336–1344. doi: 10.1210/jc.2005-1830. [DOI] [PubMed] [Google Scholar]
  25. Penninx BW. Anemia and Physical Health Decline in Old Age. Amsterdam, Netherlands: Springer; 2007. [Google Scholar]
  26. Plan and operation of the Third National Health and Nutrition Examination Survey, 1988–94. Series 1: programs and collection procedures. Vital Health Stat 1. 1994;(32):1–407. [PubMed] [Google Scholar]
  27. Rinaldi S, Geay A, Dechaud H, Biessy C, Zeleniuch-Jacquotte A, Akhmedkhanov A, Shore RE, Riboli E, Toniolo P, Kaaks R. Validity of free testosterone and free estradiol determinations in serum samples from postmenopausal women by theoretical calculations. Cancer Epidemiol Biomarkers Prev. 2002;11:1065–1071. [PubMed] [Google Scholar]
  28. Rohrmann S, Crespo CJ, Weber JR, Smit E, Giovannucci E, Platz EA. Association of cigarette smoking, alcohol consumption and physical activity with lower urinary tract symptoms in older American men: findings from the third National Health and Nutrition Examination Survey. BJU Int. 2005;96:77–82. doi: 10.1111/j.1464-410X.2005.05571.x. [DOI] [PubMed] [Google Scholar]
  29. Saad F, Gooren L, Haider A, Yassin A. An exploratory study of the effects of 12 month administration of the novel long-acting testosterone undecanoate on measures of sexual function and the metabolic syndrome. Arch Androl. 2007;53:353–357. doi: 10.1080/01485010701730880. [DOI] [PubMed] [Google Scholar]
  30. Sih R, Morley JE, Kaiser FE, Perry HM, 3rd, Patrick P, Ross C. Testosterone replacement in older hypogonadal men: a 12-month randomized controlled trial. J Clin Endocrinol Metab. 1997;82:1661–1667. doi: 10.1210/jcem.82.6.3988. [DOI] [PubMed] [Google Scholar]
  31. Smieja MJ, Cook DJ, Hunt DL, Ali MA, Guyatt GH. Recognizing and investigating iron-deficiency anemia in hospitalized elderly people. CMAJ. 1996;155:691–696. [PMC free article] [PubMed] [Google Scholar]
  32. Snyder PJ, Peachey H, Berlin JA, Hannoush P, Haddad G, Dlewati A, Santanna J, Loh L, Lenrow DA, Holmes JH, Kapoor SC, Atkinson LE, Strom BL. Effects of testosterone replacement in hypogonadal men. J Clin Endocrinol Metab. 2000;85:2670–2677. doi: 10.1210/jcem.85.8.6731. [DOI] [PubMed] [Google Scholar]
  33. Tenover JS. Effects of testosterone supplementation in the aging male. J Clin Endocrinol Metab. 1992;75:1092–1098. doi: 10.1210/jcem.75.4.1400877. [DOI] [PubMed] [Google Scholar]
  34. Vermeulen A, Verdonck L, Kaufman JM. A critical evaluation of simple methods for the estimation of free testosterone in serum. J Clin Endocrinol Metab. 1999;84:3666–3672. doi: 10.1210/jcem.84.10.6079. [DOI] [PubMed] [Google Scholar]
  35. Wannamethee SG, Lowe GD, Whincup PH, Rumley A, Walker M, Lennon L. Physical activity and hemostatic and inflammatory variables in elderly men. Circulation. 2002;105:1785–1790. doi: 10.1161/hc1502.107117. [DOI] [PubMed] [Google Scholar]
  36. Wannmethee SG, Shaper AG, Whincup PH. Ischaemic heart disease: association with haematocrit in British Regional Heart Study. J Epidemiol Community Health. 1994;48:112–118. doi: 10.1136/jech.48.2.112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Weber JP, Walsh PC, Peters CA, Spivak JL. Effect of reversible androgen deprivation on hemoglobin and serum immunoreactive erythropoietin in men. Am J Hematol. 1991;36:190–194. doi: 10.1002/ajh.2830360306. [DOI] [PubMed] [Google Scholar]
  38. Whitehead TP, Robinson D, Allaway SL, Hale AC. The effects of cigarette smoking and alcohol consumption on blood haemoglobin, erythrocytes and leucocytes: a dose related study on male subjects. Clin Lab Haematol. 1995;17:131–138. [PubMed] [Google Scholar]
  39. Winters SJ, Wang C Fortigel Study Group. LH and non-SHBG testosterone and estradiol levels during testosterone replacement of hypogonadal men: further evidence that steroid negative feedback increases as men grow older. J Androl. 2009;31:281–287. doi: 10.2164/jandrol.109.009035. [DOI] [PubMed] [Google Scholar]
  40. Yeap BB, Beilin J, Shi Z, Knuiman MW, Olynyk JK, Bruce DG, Milward EA. Serum testosterone levels correlate with haemoglobin in middle-aged and older men. Intern Med J. 2009;39:532–538. doi: 10.1111/j.1445-5994.2008.01789.x. [DOI] [PubMed] [Google Scholar]
  41. Yi S, Selvin E, Rohrmann S, Basaria S, Menke A, Rifai N, Guallar E, Platz EA, Astor B. Endogenous sex steroid hormones and measures of chronic kidney disease (CKD) in a nationally representative sample of men. Clin Endocrinol (Oxf) 2009;71:246–252. doi: 10.1111/j.1365-2265.2008.03455.x. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES