Kidney International, Vol. 64, Supplement 87 (2003), pp.

S32–S39

Anemia as a risk factor for cardiovascular disease

AREMA A. PEREIRA and MARK J. SARNAK
Division of Nephrology, Department of Medicine, Tufts-New England Medical Center

Anemia as a risk factor for cardiovascular disease. In the pres- quate tissue oxygenation in the anemic state is achieved
ent review we examine the physiologic response to chronic by both non-hemodynamic and hemodynamic adapta-
anemia and describe potential adverse effects of anemia on
myocardial and large arterial remodeling. We present observa- tions. Non-hemodynamic adaptations include increases
tional data demonstrating that anemia is a risk factor for cardio- in erythropoietin production and increases in intra-eryth-
vascular disease (CVD) outcomes in patients with chronic kid- rocytic concentrations of 2,3-diphosphoglycerate [1]. He-
ney disease and patients with heart failure. We also present modynamic changes include systemic arterial dilatation,
data that have evaluated the relationship of level of hematocrit
to CVD outcomes in patients with ischemic heart disease and which leads to a decreased systemic vascular resistance,
in patients in the general population. The results from the and reduced afterload, which in turn may increase stroke
latter studies have been inconclusive and have been limited by volume [2]. Anemia also results in decreased blood vis-
lack of knowledge of the cause of anemia. This is potentially cosity, which increases venous return and thus, augments
important because iron deficiency anemia may, in fact, improve
endothelial function. We conclude that additional randomized preload. Finally, the presence of anemia activates the
controlled trials of treatment of anemia are needed in chronic sympathetic nervous system, which results in an increase
kidney disease and patients with heart failure; however, further in heart rate [3]. Increased preload, heart rate, and stroke
exploration of the cause of anemia and the effect of different volume, as well as reduced afterload all act to raise car-
types of anemia on various CVD outcomes are needed in
patients with ischemic heart disease and patients in the general diac output.
population. Over the long term, adaptations that initially increase
cardiac output may lead to left ventricular enlargement
and eccentric left ventricular hypertrophy (LVH), char-
Chronic anemia may have adverse effects on myocar- acterized by increased LV internal dimensions and a
dial and large arterial remodeling. Decreased iron stores normal ratio of wall thickness to cavity diameter (Fig.
and, therefore, potentially, iron deficiency anemia may, 1). The latter is well appreciated in patients with sickle
however, improve endothelial function and thereby de- cell disease [4], as well as in patients who are anemic
crease cardiovascular disease (CVD) outcomes. In this secondary to their CKD [5]. Patients with CKD also
review we discuss observational and controlled trials that have a high prevalence of comorbid conditions such as
have evaluated the effect and treatment of chronic ane- hyperparathyroidism and a high calcium/phosphorus
mia on CVD outcomes in 3 populations: patients with product, which may contribute to myocardial fibrosis and
chronic kidney disease (CKD); patients with CVD; and calcium deposition. In this setting, the effects of anemia
subjects in the general population. We choose not to on myocardial remodeling may be aggravated, resulting
focus on the relationship of anemia to outcomes in sub-
in more severe LVH and less reversible pathology [6].
jects who are undergoing surgery, patients with malig-
Left ventricular hypertrophy, in turn, may predispose to
nancy, and patients in the intensive care unit as this is
heart failure or aggravate ischemic heart disease (IHD).
beyond the scope of the current article.
It remains unknown whether more subtle changes in
cardiac remodeling may occur in patients with less severe
PATHOPHYSIOLOGY: POTENTIAL ADVERSE degrees of anemia and in patients with other causes of
EFFECTS OF CHRONIC ANEMIA ON THE anemia.
HEART AND BLOOD VESSELS
Effects of anemia on the myocardium Effects of anemia on the blood vessel wall
Although the exact role of anemia in promoting CVD Chronic anemia may also adversely affect large blood
is currently not well understood, maintenance of ade- vessels, resulting in arterial hypertrophy and remodeling,
occurring as a result of sustained increases in cardiac
Key words: anemia, cardiovascular disease, chronic kidney disease, output [7, 8]. In theory, this may reverse the early vasodi-
general population. latation characteristic of anemia and lead to increased
 2003 by the International Society of Nephrology systemic vascular resistance, which further contributes

S-32
 Pereira and Sarnak: Anemia and cardiovascular disease S-33

CVD; and (2) anemia may be a marker of the severity

of underlying heart disease. For example, in patients
with heart failure, anemia may be due to hemodilution
associated with the severity of heart failure [12].

DOES THE CAUSE OF ANEMIA AFFECT CVD

RISK? A POTENTIAL PROTECTIVE EFFECT
OF IRON DEFICIENCY
In 1981, Sullivan [13] proposed that iron depletion
and, by inference, potentially iron deficiency anemia may
protect against IHD. Although controversial, there have
been several lines of support for this hypothesis.

Different rates of IHD incidence in men and women

Fig. 1. Schematic relationship of left ventricular (LV) mass and hema- The crude IHD death rate in United States men and
tocrit (Hct) in the general population. The U-shaped curve indicates women aged 35 to 74 is 219 and 91 per 100,000 popula-
an increase of cardiac mass with both anemia and polycythemia, albeit
due to different mechanisms (upper panel). Reproduced with permis- tion, respectively [14]. The incidence of IHD is consis-
sion from reference [51]. tently higher in men of all age groups compared to
women in the same age groups, with the maximal differ-
ence of 300% in the 45 to 54 age group.
The difference in the incidence of IHD appears to
to the development of LVH. Early on, these changes
parallel the changes in iron stores seen with increasing
may be reversible; however, in conditions such as CKD,
age. In men, iron stores, assessed by serum ferritin con-
they may become permanent [6].
centration, rise after adolescence [15]. In women, iron
Effects of anemia on precipitating ischemia in subjects stores remain low and only begin to rise after the age
with prevalent CVD of 45 years [16]. The maximal sex difference in serum
ferritin level is reached at approximately 45 years of age
It has been shown that the resting human heart can
[17].
withstand acute severe isovolemic anemia as low as 5.0
g/dL, without evidence of inadequate systemic oxygen- Why iron deficiency may be protective
ation. Acute isovolemic hemoglobin reduction to 5.0
Redox-active iron may contribute to lipid peroxida-
g/dL in 33 healthy resting subjects produced no evidence
tion, endothelial cell activation, and generation of reac-
of hypoxia, assessed by change in oxygen consumption
or an increase in plasma lactate concentration [9]. Analy- tive oxygen species [18]. Increased oxidative stress, in
sis of Holter readings suggests that at this level of hemo- turn, may be associated with impaired action of endothe-
globin concentration in the resting healthy population, lium-derived nitric oxide in patients with atherosclerosis,
myocardial ischemia would occur only infrequently [9]. thus contributing to endothelial dysfunction [19]. Iron
However, animal experiments and studies in postopera- chelation has been shown to improve endothelial func-
tive patients with CVD suggest that the presence of IHD tion in patients with coronary artery disease, supporting
may significantly impair the heart’s ability to tolerate the theory that iron may damage the endothelium [20].
low levels of hemoglobin [10, 11]. Observational studies linking iron stores to IHD
Some [21], but not all [22] studies have noted an associ-
OTHER REASONS WHY ANEMIA MAY BE ation between increased iron stores and risk of IHD.
ASSOCIATED WITH CVD OUTCOMES For example, a study of 9920 Canadian men and women
It is important to recognize that, although studies may demonstrated an association between levels of serum
demonstrate an association between anemia and CVD iron and risk of fatal acute myocardial infarction. The
outcomes, this does not necessarily imply that anemia is rate ratio in the highest category of serum iron levels
the cause and, therefore, a treatable cause of CVD. That compared to the lowest category was 2.18 for men and
is, confounding from unmeasured factors, or residual 5.53 for women, respectively [21].
confounding from measured factors may be playing a Studies from the Framingham population have also
role. Two examples include the following: (1) anemia been consistent with the iron hypothesis [23]. That is,
may be associated with an unmeasured risk factor, such the risk of IHD in women increased equally by natural
as inflammatory status, which in turn is the cause of both menopause or by surgical menopause (independent of
the anemia and the causal risk factor associated with oophorectomy). Thus, uterine factors, rather than hor-
S-34 Pereira and Sarnak: Anemia and cardiovascular disease

monal factors, might be responsible for the protection Portoles et al [28] studied 11 patients with reduced GFR
of premenopausal women against IHD. who received recombinant human erythropoietin for a
6-month period. The hemoglobin increased from 9.0
Blood donation may reduce IHD risk g/dL to 11.7 g/dL, and there was a trend toward de-
If iron deficiency was protective, one could hypothe- creased LV thickness and a significant decrease in LVMI
size that frequent donation of blood may reduce IHD. (178 g/m2 to 147 g/m2) following treatment. Similarly,
In fact, in a prospective cohort of 2682 Finnish men Hayashi et al [29] studied 9 patients with reduced glomer-
followed for up to 5.5 years, blood donation was shown ular filtration rate (GFR), with a mean hematocrit (Hct)
to decrease the risk of myocardial infarction by 86% of 23.6% and mean LVMI of 140.6 g/m2 at baseline.
[24]. This finding, however, has not been consistent. For These patients were initially partially treated to a goal
example, although 655 blood donors had half the risk of Hct of 30% and subsequently to a goal Hct of 40%.
CVD events compared to 3200 non-donors in an unad- Left ventricular mass index decreased to 126.9 g/m2 after
justed analysis, this difference was no longer significant partial correction of the anemia (Hct, 32%), and to 111.2
after adjustment for potential confounders [25]. g/m2 after full correction of the anemia (Hct, 39%) (P ⬍
0.0001). Finally, randomized trials have demonstrated
significant beneficial cardiovascular effects in terms of
OVSERVATIONAL STUDIES AND
exercise capacity after treatment of anemia [30, 31].
INTERVENTIONAL TRIALS RELATING
Of interest are 2 ongoing studies. The first is the Cor-
ANEMIA TO CVD OUTCOMES
rection of Hemoglobin and Outcomes in Renal Insuffi-
Patients with CKD ciency (CHOIR) Study, which randomized 2000 patients
Relationship between anemia and LVH in patients with with a baseline hemoglobin of 10.5 to 11 g/dL to a target
reduced glomerular filtration rate. In a cross-sectional hemoglobin of 13 to 13.5 g/dL with the use of recombi-
study of 175 patients attending a kidney disease clinic, nant human erythropoietin versus placebo. The primary
Levin et al [5] evaluated the relationship of hemoglobin outcome is a composite of all-cause mortality, myocar-
level to echocardiographic findings. The population con- dial infarction, stroke, and hospitalization for congestive
sisted of 115 men and 60 women with mean age of 51 heart failure. The second is the Cardiovascular Reduc-
(range, 20-82). Mean creatinine clearance (Ccr) was 25.5 tion by Early Anemia Treatment with Epoetin Beta
mL/min, and LVH was present in 38.9% of the popula- (CREATE) Study, which randomized patients to imme-
tion. The authors demonstrated that anemia and systolic diate treatment with epoetin beta to a target hemoglobin
blood pressure were the most important modifiable risk of 13 to 15 g/dL, or to late treatment with a target hemo-
factors associated with the presence of LVH. In multiple globin of 10.5 to 11.5 g/dL, once their hemoglobin levels
logistic regression analysis each 1g/dL decrease in hemo- have fallen below 10.5 g/dL. The primary outcome is
globin was associated with a 6% increase risk of LVH time to first cardiovascular event.
(P ⫽ 0.0062). Similar results have been noted in other Relationship between anemia and CVD outcomes in
studies [26]. dialysis patients. Decreased hemoglobin levels have
Results of longitudinal studies have also demonstrated been associated with a higher risk for CVD events in
an association between hemoglobin and left ventricular dialysis patients. For example, in a cohort of 432 dialysis
parameters. For example, Levin et al [27] evaluated fac- patients followed for an average of 41 months, each g/dL
tors associated with LV growth (LVG) in a prospective lower hemoglobin was associated with a 1.28 (P ⫽ 0.018),
study of 246 patients with echocardiograms at baseline 1.20 (P ⫽ 0.046), and 1.14 (P ⫽ 0.024) adjusted risk of
and 12 month follow-up. The overall prevalence of LVH de novo and recurrent congestive failure, as well as all-
at baseline was 36%. Left ventricular growth, defined cause mortality, respectively [32]. More recently, Collins
by an increase in left ventricular mass index (LVMI) of et al [33] studied a cohort of ⬃67,000 United States
⬎20% or ⬎20 g/m2 from baseline to 12 months, occurred Renal Data System hemodialysis patients and assessed
in 25% of the population. Multivariable regression analy- the effect of baseline Hct on the risks of cardiac mortality
sis demonstrated a 1.32 increased odds of LVG for each at 1 year [33]. In multivariable analysis, they demon-
0.5 g/dL decrease in hemoglobin (P ⫽ 0.004). Combined, strated that for patients with Hct values of ⱖ36%, cardiac
the results of both of these studies suggest that the pres- mortality rates were not different when compared with
ence of anemia may be a risk factor for LVG and devel- patients with hematocrit values of 33% to 36%. Patients
opment of LVH early in the course of CKD. with Hct ⬍30% or 30% to 33% were found to have a
Treatment of anemia in patients with reduced glomeru- higher risk of cardiac death than patients with Hct 33%
lar filtration rate. There have been a number of small to 36% [relative risk (RR) of 1.57 and 1.25, respectively;
non-randomized studies that have demonstrated im- (P ⬍ 0.001 for both)].
provement of cardiac parameters with correction of ane- Treatment of anemia in dialysis patients. Uncontrolled
mia in patients in the earlier stages of CKD [28, 29]. studies have suggested that treatment of anemia is associ-
 Pereira and Sarnak: Anemia and cardiovascular disease S-35

ated with reduction in left ventricular mass and favorable mia is an independent risk factor for outcomes in patients
outcomes, such as decreased all-cause and CVD mortal- with heart failure. For example, in the Studies of Left
ity in dialysis patients [34]. Similarly, treatment of ane- Ventricular Dysfunction (SOLVD), each 1% lower Hct
mia from a Hct of 25% to 34% has been shown to at baseline was associated with a 3% increase risk in all-
decrease exercise-induced myocardial ischemia, diag- cause mortality in multivariable analysis [39].
nosed by EKG monitoring [35]. Similar findings were noted in a study by Horwich et
The results from randomized controlled trials have, al [41], who demonstrated an inverse association be-
however, been less conclusive. The Normal Hematocrit tween Hb level and mortality in a study of 1061 patients
Trial enrolled ⬃1200 patients with IHD or congestive with advanced heart failure (NYHA functional class III
heart failure who were randomized to a Hct goal of or IV and LVEF ⬍40%). In multivariable analysis, each
either 30% or 42%. The primary end points were time 1 g/dL lower Hb was associated with a relative risk of
to first myocardial infarction or death. Patients random- 1.13 (CI, 1.045–1.22) for all-cause mortality [41].
ized to the higher Hct group showed a trend toward Treatment of anemia in heart failure. In an uncon-
increased mortality (though not significant) compared trolled study, Silverberg et al [40] used a combination
to patients in the lower Hct group. The exact explanation of subcutaneous erythropoietin and intravenous iron to
for these results remains unknown although potential correct anemia in 26 patients with persistent, severe heart
reasons included increased access thrombosis in the failure. Treatment resulted in improvements in mean Hb
higher Hct target, as well as increased iron use in the from 10.2 to 12.1 g/dL (P ⫽ 0.001), and a decrease in
higher Hct group [36]. The Canadian Normalization of mean NYHA functional class from 3.7 to 2.7. In addition,
Hemoglobin Trial randomized a total of 146 patients the dose of diuretics and the number of hospitalizations
with either asymptomatic concentric LVH or LV dilata- decreased.
tion to receive doses of recombinant human erythropoie- In a controlled trial, Silverberg et al [42] randomized
tin designed to achieve a target hemoglobin (Hb) level 32 patients with moderate to severe heart failure and
of either 10 or 13 g/dL. Results showed that in the con- Hb levels of 10.0 to 11.5 g/dL, to a target Hb of 12.5
centric LVH group and in the LV dilatation group, the g/dL versus placebo. Intravenous iron and erythropoietin
changes in LVMI and LV volume index, respectively,
were used in the treatment group to achieve the Hb goal.
were similar at both target Hb concentrations. However,
The treatment group showed an improvement in NYHA
patients with concentric LVH were less likely to develop
functional class by 42%, in contrast to patients with un-
progressive LV dilatation if they were randomized to
treated anemia, who showed a decline in NYHA class.
the high Hb target [37].
There was also a reduction in the need for oral and
Results are now being awaited from a larger ongoing
intravenous diuretics, as well as hospitalizations in the
controlled trial in 544 asymptomatic patients with con-
treatment compared with the placebo group.
centric LVH and normal LV cavity volume who are
Finally, Mancini et al [43] have studied the effect of
being randomized to a target Hb of 14 g/dL versus 10
treatment of anemia on exercise performance in patients
to 11 g/dL. The primary study outcome is percent change
with heart failure. Twenty-six patients with anemia and
in LV cavity volume and secondary outcomes, is percent
change in LV mass. heart failure were randomized to receive erythropoietin
These trials suggest that, although there may be some or placebo for 3 months. Results revealed that patients
benefit to treatment of anemia in hemodialysis patients in the treatment group had increased peak oxygen con-
prior to development of clinical CVD, additional trials sumption and exercise capacity in comparison with the
should be conducted in the earlier stages of CKD to placebo group.
evaluate whether many of the cardiac changes that occur
Patients with coronary artery disease
as GFR declines can be prevented.
The frequency and impact of reduced Hct was investi-
Patients with heart failure gated in a retrospective analysis of 78,974 Medicare ben-
Prevalence of anemia in heart failure. The prevalence eficiaries 65 years or older who were hospitalized with
of anemia is high in patients with heart failure, increasing acute myocardial infarction [44]. The authors noted that
as the severity of the heart failure worsens [38–40]. For patients with lower Hct values on admission had higher
example, Silverberg [40] demonstrated that the preva- 30-day mortality rates (Fig. 2). Furthermore, blood trans-
lence of anemia (Hb ⬍12 g/dL) was 9.1% and 19.2% fusion was associated with a reduction in 30-day mortal-
in patients with New York Heart Association (NYHA) ity among patients whose Hct on admission ranged from
functional class I or II, respectively, while it was as high 5.0% to 24.0% (adjusted odds ratio, 0.22; 95% CI, 0.11–
as 79.1% in patients with NYHA functional class IV. 0.45) to 30.1% to 33.0% (adjusted odds ratio, 0.69; 95%
Relationship between anemia and outcomes. Several CI, 0.53–0.89). The authors concluded that transfusion
studies have now demonstrated that the presence of ane- might improve outcomes in elderly patients who are ane-
S-36 Pereira and Sarnak: Anemia and cardiovascular disease

Fig. 2. Kaplan-Meier 30-day survival curves

according to hematocrit (Hct) category. Re-
produced with permission from reference [44].

mic and who have suffered an acute myocardial in- However, in many studies it appears that an elevated Hct
farction [44]. remains an independent risk factor for CVD outcomes,
However, not all studies are consistent with this find- despite adjustment for potential confounding risk factors
ing. A recent study examined the association of anemia [50]. Second, a higher Hct may lead to increased blood
with 1-year mortality among 30,341 patients hospitalized viscosity and, thereby, increased peripheral resistance
with acute myocardial infarction in 1986 (prethrombo- and potentially LVH (Fig. 1) [51]. Third, increases in
lytic era, N ⫽ 15,584) and 1996 (thrombolytic era, N ⫽ Hct may lead to red blood cell aggregation [52], platelet
14,757). In unadjusted analyses, a higher mortality was activation, adhesion and aggregation, and, therefore, in-
observed among patients with acute myocardial in- crease the risk of thrombus formation [53].
farction and anemia. However, after adjustment for de- In the following paragraphs we discuss a few of the
mographics, severity of CVD, and other comorbid condi- large epidemiologic studies in which the relationship of
tions there was no difference in the anemic group Hct to CVD outcomes has been evaluated (Table 1).
compared with the non-anemic group [45]. The Framingham Study. The relationship between
To our knowledge there are no randomized controlled baseline Hct and the risk of CVD morbidity and mortal-
trials that have evaluated the effect of treatment of ane- ity over a period of 34 years of follow-up was assessed
mia in patients with IHD. in 5209 men and women in the Framingham Study. In
women aged 35 to 64, a non-linear relationship was noted
Patients in the general population between level of Hct and IHD mortality. Women with
Shape of relationship between level of hematocrit and a Hct ⬎45% or a Hct ⬍40% were both at significantly
CVD outcomes. Until this point, we have focused pri- increased risk compared with those with an Hct of 42%
marily on anemia as a risk factor for CVD outcomes. to 43%, after adjustment for CVD risk factors. In women
However, there are potential reasons to suspect that a over the age of 65, only a Hct less than 39% was signifi-
high Hct may also be harmful to the heart and blood cantly associated with an increased risk of IHD mortality.
vessels; therefore, the demonstration of non-linear rela- In men, no association was found between level of Hct
tionships between level of Hct and CVD outcomes may and IHD mortality. However, men aged 35 to 64 ap-
not be unexpected. Although a few [abstract; Anker SD peared to be at increased risk for angina pectoris with
et al: ESC Congress, September 2, 2002], but not all a high Hct [48].
[46, 47] studies in heart failure have also suggested a Puerto Rico Health Program. The Puerto Rico Heart
J/U-shaped relation between level of Hct and CVD, this Health Program was a prospective study of IHD in 2555
relationship has been better appreciated in the general rural and 6151 urban men aged 45 to 64 years [54].
population [48]. Subjects were followed for 8 1/4 years. A higher Hct was
Potential adverse effects of high hematocrit on CVD. associated with an increased risk of myocardial in-
The exact mechanism why a higher Hct may be a risk farction, coronary insufficiency, and IHD mortality, for
factor for CVD is not clear. First, the association between those living in urban areas and for those aged 55 to
elevated Hct and CVD may reflect confounding from 64 years in rural areas. In multivariable analyses the
lung disease. That is, chronic obstructive pulmonary dis- relationship remained significant only for urban men
ease, a risk factor for CVD outcomes, may induce hypo- (RR, 1.5 for every 10% increase in Hct).
xemia and, thereby, stimulation of erythropoiesis [49]. The Honolulu Heart Program. The relationship be-
 Pereira and Sarnak: Anemia and cardiovascular disease S-37

Table 1. Effect of high or low Hct on CVD outcomes in the general population

Effect of low Effect of high

Name of study Author, year Definition of CVD Inclusion N Hct on CVDa Hct on CVDb
Puerto Rico Heart Health Program Morbidity and mortality secondary
Sorlie, 1980 to IHD Urban men 6151 ⫺ ⫹
Rural men 2555 ⫺ ⫺
The Honolulu Heart Program Morbidity and mortality secondary
Carter, 1982 to IHD Men 8006 ⫺ ⫺
Oslo Ischemic Study Erikssen, 1993 CVD mortality Men 2014 ⫺ ⫹
The Framingham Study Gagnon, 1994 Morbidity and mortality secondary Men and women 5209 ND ND
to CVDg Men 35–64 ND ⫺ ⫹c
Men ⬎64 ND ⫺ ⫹e
Women 35–64 ND ⫹d ⫹f
Women ⬎64 ND ⫹f ⫺
NHANES II Study Brown, 2001 Mortality secondary to IHD Men 4213 ⫺ ⫺
Women 4683 ⫺ ⫺
ARIC Study Sarnak, 2002 Morbidity and mortality secondary Men and women 14410 ⫹ ND
to IHD Men 6267 ⫺ ND
Women 8143 ⫹ ND
Abbreviations are: IHD, ischemic heart disease; CVD, cardiovascular disease; NHANES, National Health and Nutrition Examination Survey; Hct, hematocrit.
⫹, statistically significant association present; ⫺, no association detected; ND, no data available.
a,b
Effect of Hct on CVD outcomes in multivariable analyses
c
Significant association for angina pectoris only
d
Significant association for IHD mortality only
e
Significant association for CVD mortality only
f
Significant association for both CVD morbidity and mortality
g
CVD morbidity in Framingham Study defined as either IHD (including angina pectoris), intermittent claudication, congestive heart failure, stroke, or transient
ischemic attack. CVD mortality in Framingham Study defined as death from diseases of the heart or blood vessels, including the above causes.

tween baseline Hct and IHD was assessed in 8000 Japa-

nese men who were followed for a mean of 10 years in
the Honolulu Heart Program. A higher Hct at baseline
was noted in those who died of IHD. However, after
adjustment for other CVD risk factors, this association
did not persist [55].
Oslo Ischemic Study. The relationship of Hct to CVD
mortality was assessed in 488 healthy Norwegian men
who were followed for a period of 16 years. After adjust-
ment for other CVD risk factors, each 2 standard devia-
tions higher Hct at baseline was associated with a RR
for CVD mortality of 2.9 at 10 years and 1.9 at 16 years
(P ⬍ 0.005) [56].
NHANES II Mortality Study. All-cause and cause-
specific mortality were obtained from death certificates
16.8 years after 8896 adults aged 30 to 75 were enrolled in
the Second National Health and Nutrition Examination
survey (NHANES II) Mortality Study [57]. The crude
IHD mortality rate per 10,000 population among men
was found to be 42.6, 31.9, and 46.3 among those with
Hct in the lower (19% to 43%), middle (43.2% to
45.2%), and upper (45.5% to 61.75%) tertiles, respec-
tively (Fig. 3). The corresponding crude IHD mortality
rates among women were 12.6, 18.6, and 22.7 among
those with Hct in the lower (14.5% to 39.0%), middle
(39.2% to 41.5%), and upper (41.7% to 58.5%) tertiles
(Fig. 3). After adjustment for other CVD risk factors,
there was no association between Hct and IHD mortality
among men. In women, subjects in the upper tertile
showed a trend toward increased IHD mortality com-
Fig. 3. Unadjusted mortality rate per 10,000 population for coronary
pared to those in the lower tertile (RR, 1.3, 95% CI 0.9– heart disease and diseases of the heart among (A ) men and (B ) women
1.9). aged 30 to 75 years. Reproduced with permission from reference [57].
S-38 Pereira and Sarnak: Anemia and cardiovascular disease

CONCLUSION
Anemia appears to be an independent risk factor for
surrogates of CVD, such as LVH, and with clinical CVD
outcomes in patients with CKD or heart failure. In these
2 populations, randomized controlled trials are needed
to assess if this is due to a cause-and-effect relationship
and whether treatment of anemia can reduce CVD out-
comes.
In patients with IHD and in subjects in the general
population the relationship between anemia with CVD is
inconclusive. Future observational studies that evaluate
Fig. 4. Kaplan-Meier survival analysis for cardiovascular disease
this relationship should attempt to identify both the
(CVD) in men and women stratified by presence versus absence of cause of the anemia as well as surrogates of CVD. The
anemia. Log-rank statistics were significant (P ⫽ 0.03 for men, P ⫽ 0.04 reason for the latter is that anemia itself may result in
for women) for differences between anemic and non-anemic subjects.
Dotted line represents anemia; solid line represents nonanemia. Repro-
ventricular remodeling and LVH, while iron deficiency
duced with permission from reference [58]. anemia could theoretically improve endothelial function.
Therefore, the possibility of competing risks may be at
play.

Atherosclerosis Risk in Communities Study (ARIC). Reprint requests to Mark J. Sarnak, M.D., Box 391, Tufts-New En-
gland Medical Center, 750 Washington Street, Boston, MA 02111.
The ARIC study is a prospective cohort study of athero- E-mail: msarnak@tufts-nemc.org
sclerosis and its risk factors in 4 United States communi-
ties. Anemia was defined using World Health Organiza- REFERENCES
tion criteria, namely a Hb level ⬍12 g/dL in women and 1. Horina JH, Schwaberger G, Brussee H, et al: Increased red
⬍13 g/dL in men. A total of 14,410 subjects without cell 2,3-diphosphoglycerate levels in haemodialysis patients treated
CVD at baseline had Hb levels measured. Three hundred with erythropoietin. Nephrol Dial Transplant 8:1219–1222, 1993
2. Duke M, Abelmann WH: The hemodynamic response to chronic
men (4.8%) and 1058 women (13.0%) were anemic at anemia. Circulation 39:503–515, 1969
baseline. The presence of anemia was associated with 3. Anand IS, Chandrashekhar Y, Ferrari R, et al: Pathogenesis of
CVD outcomes in men and women (Fig. 4). Anemia was oedema in chronic severe anaemia: Studies of body water and
sodium, renal function, haemodynamic variables, and plasma hor-
also independently associated with an increased risk of mones. Br Heart J 70:357–362, 1993
CVD in the entire cohort [hazards ratio (HR), 1.41, 95% 4. Werdehoff SG, Moore RB, Hoff CJ, et al: Elevated plasma endo-
CI 1.01-1.95]. In subgroup analyses, the hazard ratios thelin-1 levels in sickle cell anemia: Relationships to oxygen satura-
tion and left ventricular hypertrophy. Am J Hematol 58:195–199,
were in the same direction, although not statistically 1998
significant in all cases [58]. For example, the HR (95% 5. Levin A, Singer J, Thompson CR, et al: Prevalent left ventricular
hypertrophy in the predialysis population: Identifying opportuni-
CI) was 1.33 (0.85–2.08), 1.71 (1.01–2.89), 1.18 (0.74–1.88), ties for intervention. Am J Kidney Dis 27:347–354, 1996
and 1.63 (1.03–2.59) in men, women, African Americans, 6. Metivier F, Marchais SJ, Guerin AP, et al: Pathophysiology of
and Whites, respectively. anaemia: Focus on the heart and blood vessels. Nephrol Dial Trans-
plant 15(Suppl 3):S14–S18, 2000
Interpretation of the results. These studies unfortu- 7. Gibbons GH, Dzau VJ: The emerging concept of vascular remod-
nately do not provide us with any definitive answers eling. N Engl J Med 330:1431–1438, 1994
regarding the relationship of anemia and level of Hct to 8. London GM, Parfrey PS: Cardiac disease in chronic uremia:
pathogenesis. Adv Ren Replace Ther 4:194–211, 1997
CVD. It appears that the relationship between Hct and 9. Weiskopf RB, Viele MK, Feiner J, et al: Human cardiovascular
CVD may vary depending on the gender, the age group and metabolic response to acute, severe isovolemic anemia. JAMA
279:217–221, 1998
being studied, the population being studied, and the de-
10. Levy PS, Kim SJ, Eckel PK, et al: Limit to cardiac compensation
gree of anemia being evaluated. Of note, none of the during acute isovolemic hemodilution: Influence of coronary steno-
studies provide us detailed data of the causes of low sis. Am J Physiol 265:H340–349, 1993
11. Carson JL, Duff A, Poses RM, et al: Effect of anaemia and
Hct. This is particularly important in relation to the iron cardiovascular disease on surgical mortality and morbidity. Lancet
deficiency hypothesis presented above. 348:1055–1060, 1996
The analysis of the ARIC cohort was unique in that 12. Androne AS, Katz SD, Lund L, et al: Hemodilution is common
in patients with advanced heart failure. Circulation 107:226–229,
it evaluated anemia rather than tertiles or quintiles of 2003
Hct. In this respect it is important to note that only 13. Sullivan JL: Iron and the sex difference in heart disease risk.
approximately 10% of subjects had anemia. Therefore, Lancet 1:1293–1294, 1981
14. American Heart Association: Heart Disease and Stroke: Statis-
it is possible that only the lowest decile of Hct is associ- tics-2003 update. www.americanheart.org, 2003
ated with CVD outcomes. 15. Cook JD, Lipschitz DA, Miles LE, et al: Serum ferritin as a
 Pereira and Sarnak: Anemia and cardiovascular disease S-39

measure of iron stores in normal subjects. Am J Clin Nutr 27:681– disease who are receiving hemodialysis and epoetin. N Engl J Med
687, 1974 339:584–590, 1998
16. Burt MJ, Halliday JW, Powell LW: Iron and coronary heart 37. Foley RN, Parfrey PS, Morgan J, et al: Effect of hemoglobin
disease. BMJ 307:575–576, 1993 levels in hemodialysis patients with asymptomatic cardiomyopathy.
17. Sullivan JL: Iron, hematocrit, and the sex difference in heart Kidney Int 58:1325–1335, 2000
disease. Arch Pathol Lab Med 117:966–967, 1993 38. Ezekowitz JA, McAlister FA, Armstrong PW: Anemia is com-
18. Pratico D, Pasin M, Barry OP, et al: Iron-dependent human mon in heart failure and is associated with poor outcomes: Insights
platelet activation and hydroxyl radical formation: Involvement of from a cohort of 12 065 patients with new-onset heart failure.
protein kinase C. Circulation 99:3118–3124, 1999 Circulation 107:223–225, 2003
19. Cooper CE: Nitric oxide and iron proteins. Biochim Biophys Acta 39. Al-Ahmad A, Rand WM, Manjunath G, et al: Reduced kidney
1411:290–309, 1999 function and anemia as risk factors for mortality in patients with
left ventricular dysfunction. J Am Coll Cardiol 38:955–962, 2001
20. Duffy S, Biegelson E, Holbrook M, et al: Iron chelation improves
40. Silverberg DS, Wexler D, Blum M, et al: The use of subcutaneous
endothelial function in patients with coronary disease. Circulation
erythropoietin and intravenous iron for the treatment of the ane-
103:2799–2804, 2001 mia of severe, resistant congestive heart failure improves cardiac
21. Morrison HI, Semenciw RM, Mao Y, et al: Serum iron and risk and renal function and functional cardiac class, and markedly re-
of fatal acute myocardial infarction. Epidemiology 5:243–246, 1994 duces hospitalizations. J Am Coll Cardiol 35:1737–1744, 2000
22. Manttari M, Manninen V, Huttunen JK, et al: Serum ferritin 41. Horwich TB, Fonarow GC, Hamilton MA, et al: Anemia is associ-
and ceruloplasmin as coronary risk factors. Eur Heart J 15:1599– ated with worse symptoms, greater impairment in functional capac-
1603, 1994 ity and a significant increase in mortality in patients with advanced
23. Kannel WB, Hjortland MC, McNamara PM, et al: Menopause heart failure. J Am Coll Cardiol 39:1780–1786, 2002
and risk of cardiovascular disease: The Framingham study. Ann 42. Silverberg DS, Wexler D, Sheps D, et al: The effect of correction
Intern Med 85:447–452, 1976 of mild anemia in severe, resistant congestive heart failure using
24. Tuomainen TP, Salonen R, Nyyssonen K, et al: Cohort study of subcutaneous erythropoietin and intravenous iron: A randomized
relation between donating blood and risk of myocardial infarction controlled study. J Am Coll Cardiol 37:1775–1780, 2001
in 2682 men in eastern Finland. BMJ 314:793–794, 1997 43. Mancini DM, Katz SD, Lang CC, et al: Effect of erythropoietin
25. Meyers DG, Strickland D, Maloley PA, et al: Possible associa- on exercise capacity in patients with moderate to severe chronic
tion of a reduction in cardiovascular events with blood donation. heart failure. Circulation 107:294–299, 2003
Heart 78:188–193, 1997 44. Wu W, Rathore SS, Wang Y, et al: Blood transfusion in elderly
26. Greaves SC, Gamble GD, Collins JF, et al: Determinants of left patients with acute myocardial infarction. N Eng J Med 345:1230–
ventricular hypertrophy and systolic dysfunction in chronic renal 1236, 2001
failure. Am J Kidney Dis 24:768–776, 1994 45. Al Falluji N, Lawrence-Nelson J, Kostis JB, et al: Effect of
27. Levin A, Thompson CR, Ethier J, et al: Left ventricular mass anemia on 1-year mortality in patients with acute myocardial in-
index increase in early renal disease: Impact of decline in hemoglo- farction. Am Heart J 144:636–641, 2002
46. Sarnak MJ, Al-Ahmad A, Rand WM, et al: Is a high hematocrit
bin. Am J Kidney Dis 34:125–134, 1999
level good for patients with heart failure? J Am Coll Cardiol
28. Portoles J, Torralbo A, Martin P, et al: Cardiovascular effects
39:1704–1705, 2001
of recombinant human erythropoietin in predialyis patients. Am 47. Fonarow GC, Horwich TB, Hamilton MA, et al: Anemia and
J Kidney Dis 29:541–548, 1997 poor prognosis in advanced heart failure. J Am Coll Cardiol
29. Hayashi T, Suzuki A, Shoji T, et al: Cardiovascular effect of 40:2204, 2002
normalizing the hematocrit level during erythropoietin therapy in 48. Gagnon D, Zhang T, Brand F, et al: Hematocrit and the risk of
predialysis patients with chronic renal failure. Am J Kidney Dis cardiovascular disease-The Framingham Study: A 34-year follow-
35:250–256, 2000 up. Am Heart J 127:674–682, 1994
30. Teehan BP, Krantz S, Stone WA, et al: Double-blind, placebo- 49. Kannel WB, D’Agostino RB, Belanger AJ: Fibrinogen, cigarette
controlled study of the therapeutic use of recombinant human smoking, and risk of cardiovascular disease: Insights from the Fra-
erythropoietin for anemia associated with chronic renal failure in mingham Study. Am Heart J 113:1006–1010, 1987
pre-dialysis patients. Am J Kidney Dis 18:50–59, 1991 50. Lowe GD, Drummond MM, Lorimer AR, et al: Relation between
31. Revicki DA, Brown RE, Feeny DH, et al: Health-related quality the extent of coronary artery disease and blood viscosity. BMJ
of life associated with recombinant human erythropoietin therapy 85:673–674, 1980
for predialysis chronic renal disease patients. Am J Kidney Dis 51. Schunkert H, Hense HW: A heart price to pay for anaemia.
25:548–554, 1995 Nephrol Dial Transplant 16:445–448, 2001
32. Foley RN, Parfrey PS, Harnett JD, et al: The impact of anemia 52. Lowe GD, Forbes CD: Platelet aggregation, haematocrit, and fi-
on cardiomyopathy, morbidity, and mortality in end-stage renal brinogen. Lancet 1:395–396, 1985
disease. Am J Kidney Dis 28:53–61, 1996 53. Harrison MJ, Pollock SS, Weisblatt E: Haematocrit and platelet
33. Collins AJ, Li S, St Peter W, et al: Death, hospitalization, and aggregation. Lancet 2:991–992, 1984
economic associations among incident hemodialysis patients with 54. Sorlie P, Garcia-Palmieri M, Costas R, et al: Hematocrit and
hematocrit values of 36 to 39%. J Am Soc Nephrol 12:2465–2473, risk of coronary heart disease: The Puerto Rico Heart Health
Program. Am Heart J 101:456–461, 1981
2001
55. Carter C, McGee D, Reed D, et al: Hematocrit and the risk of
34. London GM, Pannier B, Guerin AP, et al: Alterations of left
coronary heart disease: The Honolulu Heart Program. Am Heart
ventricular hypertrophy in and survival of patients receiving hemo- J 105:674–679, 1983
dialysis: Follow-up of an interventional study. J Am Soc Nephrol 56. Erikssen G, Thaulow E, Sandvik L, et al: Hematocrit: A predictor
12:2759–2767, 2001 of cardiovascular mortality? J Intern Med 235:493–499, 1993
35. Wizemann V, Kaufmann J, Kramer W: Effect of erythropoietin on 57. Brown D, Giles W, Croft J: Hematocrit and the risk of coronary
ischemia tolerance in anemic hemodialysis patients with confirmed heart disease mortality. Am Heart J 142:657–663, 2001
coronary artery disease. Nephron 62:161–165, 1992 58. Sarnak MJ, Tighiouart H, Manjunath G, et al: Anemia as a risk
36. Besarab A, Bolton WK, Browne JK, et al: The effects of normal factor for cardiovascular disease in The Atherosclerosis Risk in
as compared with low hematocrit values in patients with cardiac Communities (ARIC) study. J Am Coll Cardiol 40:27–33, 2002