Kidney International, Vol. 67 (2005), pp.

1483–1488

Higher prevalence of anemia with diabetes mellitus in moderate

kidney insufficiency: The Kidney Early Evaluation Program
TAREK M. EL-ACHKAR, SUZANNE E. OHMIT, PETER A. MCCULLOUGH, ERROL D. CROOK,
WENDY W. BROWN, RICHARD GRIMM, GEORGE L. BAKRIS, WILLIAM F. KEANE, and JOHN M. FLACK
Department of Internal Medicine, Wayne State University, Detroit, Michigan; Chief of Weight Control Center, William Beaumont
Hospital, Royal Oak, Michigan; Division of Nephrology, Wayne State University, and the John D. Dingell VAMC, Detroit,
Michigan; Division of Nephrology, St. Louis University, St. Louis, Missouri; Department of Medicine, Hennepin County Medical
Center, Minneapolis, Minnesota; and Rush Medical College, Chicago, Illinois

Higher prevalence of anemia with diabetes mellitus in moderate
kidney insufficiency: The Kidney Early Evaluation Program.
Background. The Kidney Early Evaluation Program (KEEP
2.0) cross-sectional, community-based study, targeted individ-
uals at increased risk for kidney disease and measured blood
glucose, creatinine, and hemoglobin.
Methods. KEEP 2.0 screening data were used to determine
the prevalence of anemia by level of kidney function and dia-
betes status. Estimated glomerular filtration rate (EGFR) was
calculated using serum creatinine values, and categorized as
≥90, 60–89, 30–59 and <30 mL/min/1.73m2 . Anemia was de-
fined as hemoglobin <12 g/dL in men and in women aged >50
years, and <11 g/dL in women ≤50 years. Diabetes was defined
as participant-reported diagnosis, fasting glucose >125 mg/dL,
or nonfasting glucose >200 mg/dL.
Results. Data were available on 5380 participants screened
from August 2000 through December 2001. Diabetes was
present in 26.9% of participants, and anemia in 7.7%; 15.9%
of participants had at least moderately reduced kidney func-
tion (EGFR <60 mL/min/1.73m2 ). In participants with diabetes,
anemia prevalence at the 4 levels of descending EGFR were
8.7%, 7.5%, 22.2%, and 52.4%, compared with 6.9%, 5.0%,
7.9%, and 50.0% in persons without diabetes. In a multivari-
able model, participants of non-white race/ethnicity, those with
diabetes and those with EGFR <30 or 30–59 mL/min/1.73m2
had significantly increased odds of anemia. In addition, a sig-
nificant sex-diabetes interaction was identified; odds of anemia
were 4-fold greater in men than women with diabetes relative
to sex-matched participants without diabetes.
Conclusion. Diabetes was independently correlated with
anemia, more so in men than women, and may be linked to
premature expression of anemia in persons with moderate re-
ductions in kidney function.
Anemia is a common feature of end-stage renal disease
(ESRD), but it also accompanies lesser reductions in kid-
ney function [1, 2]. The degree of anemia roughly approx-
imates the severity of kidney dysfunction [3, 4]. Diabetes
mellitus is a risk factor for chronic kidney insufficiency
and is the leading cause of ESRD [5]. Diabetes-related
anemia has been observed in the advanced uremia of di-
abetic nephropathy; however, diabetes affects the hema-
tologic system in several ways [6]. Recent studies have
linked anemia with relatively low serum erythropoietin
in persons with either type 1 or type 2 diabetes even with-
out advanced kidney disease or overt uremia [7–9]. Pa-
tients with diabetic nephropathy may have worse clinical
outcomes related to anemia than other etiologies of kid-
ney failure. Nevertheless, anemia in diabetes likely rep-
resents a complex interplay between diabetes, level of
kidney function, and medication exposures. Anemia is a
treatable condition that has been linked to left ventric-
ular hypertrophy, cardiovascular morbidity, more rapid
loss of kidney function, and poor quality of life [10, 11].
The Kidney Early Evaluation Program (KEEP 2.0) is
an ongoing community-based health-screening program
sponsored by the National Kidney Foundation (NKF)
that identifies individuals at increased risk for kidney
disease, and encourages them to seek follow-up health
care. We used KEEP 2.0 data to examine the relation-
ship of anemia with diabetes and kidney function, and to
determine whether this relationship differed in men and
women.

Key words: anemia, diabetes, estimated glomerular filtration rate,
hemoglobin, K/DOQI, kidney disease, kidney function.
Received for publication July 7, 2004
and in revised form September 2, 2004
Accepted for publication October 14, 2004


C 2005 by the International Society of Nephrology
METHODS
From August 2000 through December 2001, 135 screen-
ing programs in 33 states were carried out by 31 NKF affil-
iates. Eligible participants were at least 18 years old, with
diabetes or hypertension, or with a family history of di-
abetes, hypertension, or kidney disease [12]. Some NKF
affiliates targeted recruitment efforts in areas with large

1483
1484 El-Achkar et al: Anemia with diabetes and reduced kidney function
African American populations because of their known
high prevalence of diabetes and hypertension. All par-
ticipants provided informed consent prior to data col-
lection. Data were collected by questionnaire on demo-
graphic characteristics and medical history via participant
self-report. Medication status was not obtained during
KEEP health assessments. Systolic and diastolic blood
pressures were measured, and blood specimens were col-
lected and processed for determination of blood glu-
cose, creatinine, and hemoglobin. In analyses presented
here we focused on the following variables: age, sex,
race/ethnicity, hemoglobin, blood glucose, and diabetes
status, and serum creatinine to estimate glomerular fil-
tration rate (EGFR).
Diabetes was diagnosed by participant self-report, fast-
ing glucose values >125 mg/dL, or nonfasting glucose
levels >200 mg/dL. Serum creatinine values were con-
sidered abnormal if values were >1.5 mg/dL for men
and >1.3 mg/dL for women [13]. Participant EGFR val-
ues (mL/min/1.73m2 ) were calculated using the simpli-
fied Modification of Diet in Renal Disease (MDRD) for-
mula (186.3 × [serum creatinine, mg/dL −1.154 ] × [age,
years −.203 ]); calculated values were multiplied by .742
for women, and by 1.21 for African Americans [abstract;
Levey AS, Greene T, Kusek JW, Beck GJ, and MDRD
Study Group. J Am Soc Nephrol 11:155A, 2000; A0828,
14] Based on stages of chronic kidney disease (CKD)
from the Kidney Disease Outcomes Quality Initiative
(K/DOQI) guidelines, EGFR values were categorized us-
ing the following cut-points: ≥90 (CKD 1), 60–89 (CKD
2), 30–59 (CKD 3), and <30 (CKD 4, 5) mL/min/1.73m2
[15]. We grouped those with EGFR <15 mL/min/1.73m2
(CKD 5) with those having less severe dysfunction be-
cause of the small number of such individuals (N = 9).
EGFR values <60 mL/min/1.73m2 (CKD 3–5) were con-
sidered abnormal, and indicated at least moderately re-
duced kidney function [15]. Participants were categorized
as anemic based on the K/DOQI guidelines definition
for anemia that warrants investigation in chronic renal
insufficiency [K/DOQI—hemoglobin values <12.0 g/dL
for men and for postmenopausal women (>50 years old),
and <11.0 g/dL for premenopausal women (≤50 years
old)] [10].
Statistical analyses were performed with the SAS sta-
tistical package (release 8.2; SAS Institute Inc., Cary,
NC, USA). Descriptive analyses were used to character-
ize the study population by demographics, including sex,
age, race/ethnicity, and by medical status, including mean
hemoglobin, prevalence of anemia and of diabetes, and
categories of kidney function. The associations of ane-
mia prevalence or mean hemoglobin value with level of
kidney function, stratified by diabetes status, were exam-
ined using chi-square statistics for categorical variables,
and Wilcoxon rank-sum statistics or analysis of variance
(ANOVA) techniques for continuous variables. Odds of
anemia by level of kidney function and diabetes status
were also examined in multivariable logistic regression
models, adjusted for age, sex, and race/ethnicity, with
generation of odds ratios (OR) and consideration of in-
teractions. Analyses presented here are limited to those
individuals with complete data on key variables—
diabetes, anemia, and EGFR.
RESULTS
The population screened was composed of 6071 eligi-
ble participants, including 1956 men and 4115 women.
Among these, complete data on key variables were avail-
able for 5380 (89%) participants; characteristics of this
population by diabetes status are presented in Table 1.
Participants were predominately women (67%), and
African American or white (79%); participant mean age
was 53 years (SD = 15.6). Diabetes was present in 27% of
participants, including 30% of men and 25% of women
(P = .001); 95% of those categorized as such reported
a diabetes diagnosis. Only 5% of participants had ab-
normal serum creatinine values [13]; however, 16% had
EGFR values less than 60 mL/min/1.73m2 , indicating at
least moderately reduced kidney function [15]. Partici-
pants with diabetes were more likely than those without
diabetes to have abnormal serum creatinine and EGFR
values. Men were more likely than women to have ab-
normal serum creatinine values (8% vs. 4%, P < .001);
however, women were slightly more likely than men to
have EGFR values less than 60 mL/min/1.73m2 (17% vs.
15%, P = .06). Mean EGFR values did not significantly
differ by sex [mean 81.5 (SD = 22.2, range 7–207) for
men compared with mean 81.2 (SD = 23.6, range 10–
239) for women, P = .06]. Three percent of men and 10%
of women had anemia according to the K/DOQI defini-
tion (P < .001) [10]. Twelve percent of participants with
diabetes were categorized as anemic compared with 6%
of those without diabetes.
Prevalence of anemia by K/DOQI category of kidney
function was examined for participants with and with-
out diabetes (Fig. 1A). Anemia prevalence was signif-
icantly greater in persons with diabetes compared to
persons without diabetes in the kidney function cate-
gories 30–59 mL/min/1.73m2 (22.2% vs. 7.9%, P < .001)
and 60–89 mL/min/1.73m2 (7.5% vs. 5.0%, P = .015).
Mean hemoglobin values were significantly lower in per-
sons with diabetes compared to persons without diabetes
(13.0 g/dL vs. 13.6 g/dL, P < .001) in the kidney function
category of 30–59 mL/min/1.73m2 , but did not differ for
persons with or without diabetes in the kidney function
category of 60–89 mL/min/1.73m2 (13.7 g/dL for both,
P = 1.0). No significant differences in anemia prevalence
(P = .88) or mean hemoglobin values (11.5 g/dL vs. 11.9 g/
dL, P = .58) were detected at EGFR values below 30 mL/
 El-Achkar et al: Anemia with diabetes and reduced kidney function 1485

Table 1. Demographic characteristics and medical status of 5380 participants in a health screening program designed to identify individuals at
increased risk for kidney disease (KEEP 2.0, August 2000–December 2001)
Diabetes status
Present Absent Total
(N = 1445, 26.9%) (N = 3935, 73.1%) (N = 5380) P value
Mean age years (range) 58.4 (18–101) 50.3 (18–100) 52.5 (18–101) <0.001
Mean EGFR mL/min/1.73m2 (range) 78.7 (10–208) 82.3 (7–239) 81.3 (7–239) <0.001
Mean hemoglobin g/dL (range) 13.5 (7.1–18.1) 13.6 (5.7–18.6) 13.6 (5.7–18.6) 0.031
N (%) N (%) N (%)
Race/ethnicity <0.001
African American 543 (37.6) 1752 (44.5) 2295 (42.7)
White 514 (35.6) 1419 (36.1) 1933 (35.9)
Hispanic 174 (12.0) 400 (10.2) 574 (10.7)
Other/unknown 214 (14.8) 364 (9.3) 578 (10.7)
Gender 0.001
Men 525 (36.3) 1244 (31.6) 1769 (32.9)
Women 920 (63.7) 2691 (68.4) 3611 (67.1)
Abnormal creatinine 120 (8.3) 165 (4.2) 285 (5.3) <0.001
Anemia 168 (11.6) 246 (6.3) 414 (7.7) <0.001
EGFR mL/min/1.73m2 <0.001
<30 [CKD 4, 5]a 21 (1.5) 22 (0.6) 43 (0.8)
30–59 [CKD 3] 307 (21.2) 507 (12.9) 814 (15.1)
60–89 [CKD 2] 671 (46.4) 2072 (52.7) 2743 (51.0)
≥90 [CKD 1] 446 (30.9) 1334 (33.9) 1780 (33.1)
a
Stages of chronic kidney disease (CKD) [14].

60 Table 2. Results from a multivariable logistic regression model

Diabetes evaluating factors associated with anemia among 5380 participants in
52.4
50 Non-diabetes a community health-screening program
50
OR 95% CI P value
40
P < 0.001 Age years 1.03 1.02–1.04 <0.001
30 Male gender .26 .19–.35 <0.001
22.2 Race/ethnicity
20 African American 3.04 2.32–3.99 <0.001
P = 0.015
Hispanic 2.12 1.42–3.17 <0.001
8.7
10 7.9 7.5 6.9 Other 2.07 1.41–3.03 <0.001
Anemia prevalence, %

5
White Reference
0 Diabetes 1.63 1.30–2.03 <0.001
<30 30-59 60-89 >89 EGFR mL/min/1.73m2
<30 12.32 6.16–24.64 <0.001
70 30–59 1.38 1.01–1.89 0.045
60–89 .68 .53–.88 0.003
60 54.2 ≥90 Reference
52.2
50 P = 0.038
In a separate similar model, a statistically significant sex-diabetes interaction
40.4 was observed (OR = 4.3, 95% CI 2.2–8.5, P < .001), indicating that relative to
40 sex-matched nondiabetic participants, men with diabetes had greater odds of
P = 0.009
anemia than women with diabetes.
30 23.5 P < 0.001
20.8
20 16.2
10.1 8.7
10 7.5 5.7 5.9 4.5 6.2
4.8
ered EGFR divided into 10 mL/min/1.73m2 strata and
0
<31 31-40 41-50 51-60 61-70 71-80 >80 re-examined the prevalence of anemia in persons with
Estimated GFR, mL/min/1.73m2 and without diabetes (Fig. 1B) [1]. Statistically significant
(P < .05) differences in prevalence of anemia by diabetes
Fig. 1. Distributions of anemia prevalence, by K/DOQI categories, and status were present at each of the three EGFR categories
by 10 mL/min/1.73m2 increments of estimated GFR, among participants
with or without diabetes.
between 31–60 mL/min/1.73m2 , but did not significantly
differ for any of the other categories.
In a multivariable logistic regression model that con-
min/1.73m2 ; however, this category had a small number sidered factors associated with odds of anemia, di-
of studied subjects (N = 43). abetes status, and both EGFR categories <30 and
To more precisely evaluate the relationship of diabetes 30–59 mL/min/1.73m2 were significantly associated with
and anemia with level of kidney function, we consid- increased likelihood of anemia (Table 2). Results from
1486 El-Achkar et al: Anemia with diabetes and reduced kidney function
Table 3. Mean hemoglobin (HGB) for 10 mL/min/1.73m2 increments of EGFR stratified by diabetes status and sex
Men
EGFR with
mL/min/1.73m2 diabetes
>80 14.6
71–80 14.5
61–70 14.1
51–60 13.9 (P = 0.006)
41–50 13.3 (P < 0.001)
31–40 12.2 (P < 0.001)
<31 12.2 (P < 0.001)
a
P value for the difference in mean HGB at each strata from mean HGB at EGFR >80 mL/min/1.73m2 of corresponding column; only statistically significant (<0.05)
values are presented.
these analyses also indicated increased odds of ane-
mia with older age, among women, and among African
American, Hispanic, and other non-white participants.
In a similar separate model, a significant sex-diabetes in-
teraction was identified (OR 4.3, 95% CI 2.2–8.5, P <
.001), indicating that relative to sex-matched participants
without diabetes, men with diabetes had greater odds of
anemia than women with diabetes. In unadjusted analy-
ses, men with diabetes were more likely than men without
diabetes to have anemia (8.0% vs. 1.1%, OR 7.6, 95% CI
4.1–14.1, P < .001), as were women with diabetes com-
pared to women without diabetes (13.7% vs. 8.6%, OR
1.7, 95% CI 1.3–2.1, P < .001); however, differential odds
of anemia by diabetes status for men and women were
apparent (Breslow-Day test for homogeneity of ORs,
P < .001). Mean hemoglobin values were significantly
lower for men with diabetes compared to men without
diabetes (14.2 g/dL vs. 14.7 g/dL, P < .001), but were not
different for women with and without diabetes (13.1 g/dL
for both, P = .69).
In Table 3, mean hemoglobin values were compared
in 10 mL/min/1.73m2 increments of EGFR for men and
women stratified by diabetes status; category-specific
mean hemoglobin values at multiple EGFR strata were
compared for differences with mean hemoglobin values
at EGFR >80 mL/min/1.73m2 . Among men with dia-
betes, significantly lower mean hemoglobin values were
observed at all EGFR categories less than 60 mL/min/
1.73m2 . In contrast, among women with diabetes and
among men and women without diabetes, significantly
lower mean hemoglobin values were not apparent un-
til more advanced levels of kidney dysfunction (EGFR
<31 mL/min/1.73m2 ).
DISCUSSION
We examined the prevalence of anemia across a broad
range of kidney function according to diabetes status,
among individuals at increased risk of kidney disease
participating in a community-based screening program
Men Women Women
without with without
diabetes diabetes diabetes
Mean HGB g/dL (P value a )
14.7 13.0 12.9
14.9 13.4 13.2
14.8 13.4 13.3
14.7 13.1 13.5
14.3 12.9 13.2
14.2 12.4 13.0
12.8 (P < 0.001) 10.8 (P < 0.001) 11.2 (P < 0.001)
that targeted persons with risk factors for CKD. There
were significant differences in the prevalence of anemia
(and mean hemoglobin values) between persons with
and without diabetes at levels of moderately impaired
kidney function (EGFR 30–59 mL/min/1.73m2 –CKD 3).
This interesting relationship was less evident in the mildly
impaired and normal levels of kidney function—EGFR
≥60 mL/min/1.73m2 (CKD 1, 2). At the level of severely
impaired kidney function (EGFR <30 mL/min/1.73m2 –
CKD 4, 5), anemia was frequently identified (51%), but
with no significant difference in prevalence of anemia by
diabetes status, possibly due to the small number of stud-
ied subjects. In multivariable models adjusted for par-
ticipant demographic characteristics, diabetes status and
both moderate (EGFR 30–59 mL/min/1.73m2 –CKD 3)
and significant (EGFR <30 mL/min/1.73m2 –CKD 4, 5)
reductions in kidney function were associated with in-
creased odds of anemia. Both men and women with
diabetes had increased prevalence of anemia; however,
diabetes conferred greater odds of anemia in men than
in women.
Our findings of increased odds of anemia among older
participants and among African American and Hispanic
participants are consistent with findings from recent anal-
yses of a population-based survey evaluating the health
status of the United States population [2]. That study also
found, as we did, greater overall prevalence of anemia
among women compared with men; prevalence estimates
were higher in our study for both men and women, as our
study targeted an at-risk rather than general population.
We used the K/DOQI definition of anemia that warrants
investigation in chronic renal insufficiency [10]; however,
several other definitions of anemia have been developed
and published [2, 16, 17]. Analyses carried out using these
other definitions (data not shown) generated similar re-
sults and identical conclusions, even though prevalence
varied.
We used the simplified Modification of Diet in
Renal Disease (MDRD) formula to calculate EGFR
values (mL/min/1.73m2 ); this validated formula was
 El-Achkar et al: Anemia with diabetes and reduced kidney function
derived from measured GFR, and uses only demographic
characteristics and serum creatinine values to estimate
GFR [14]. Other formulas, including the widely used
Cockcroft-Gault formula [18], have also been used to
estimate level of kidney function (creatinine clearance).
Analyses carried out using EGFR values calculated based
on the body surface area adjusted-Cockcroft-Gault for-
mula (data not shown) confirmed the findings presented.
The relationship of diabetes to anemia has been clas-
sically related to advanced kidney damage and uremia.
Anemia is a common manifestation of chronic kidney
failure, and contributes to multiple adverse outcomes,
possibly via decreased tissue oxygen delivery and utiliza-
tion [10, 19]. A significant correlation between hemat-
ocrit and creatinine clearance was found to occur below
40 mL/min/1.73m2 in an earlier study, and when GFR fell
below 20 mL/min/1.73m2 in another study [3, 4]. Hsu et
al demonstrated that hematocrit decreased progressively
below EGFR of 50 mL/min/1.73m2 in men and women,
with men having a larger decrease in hematocrit [1]; this
observation is consistent with our findings among KEEP
participants. Similar observations were made recently us-
ing health status survey data from a representative sample
of the United States population [2, 20]. It is important to
note, however, that diabetes status was not considered in
these studies.
Low levels of erythropoietin have been observed in
persons with anemia and either type 1 or type 2 diabetes,
but without advanced renal disease or overt uremia [7–9].
For example, Bosman et al found that 13 of 27 patients
with type 1 diabetes, nephropathy, and serum creatinine
<2 mg/dL had unexplained anemia with relatively low
erythropoietin levels, compared to none of 26 patients
without diabetes but with glomerulonephritis [8]. Simi-
larly, hemoglobin was lower in 19 patients with type 2
diabetes and kidney failure compared with 21 patients
without diabetes who also had kidney failure and compa-
rable serum creatinine levels [9]. These studies typically
used clinic populations, had small sample sizes, and none
documented, as we did, a link between anemia and a spe-
cific level of glomerular filtration rate in persons with
diabetes.
There are several plausible mechanisms that may ex-
plain the link between diabetes and anemia, as well as
the differential diabetes-related anemia risk in men com-
pared with women. Renal denervation attributable to
diabetic autonomic neuropathy can reduce splanchnic
sympathetic stimulation of erythropoietin production [7,
8, 21]. Also, diabetes may adversely affect peritubular and
interstitial structures in the renal cortex, the site of ery-
thropoietin production, even prior to the development of
overt nephropathy. This may attenuate the release of ery-
thropoietin in response to the hypoxic stimuli of anemia
[8, 22]. Furthermore, both diabetes and reduced kidney
function have been linked to depressed androgen levels
1487
[23]. Androgens stimulate erythropoiesis by increasing
erythropoietin production and by direct augmentation of
marrow stem cells [24]. Finally, kidney failure has been
associated with reduced testosterone levels in men [25].
On the other hand, premenopausal and postmenopausal
women with diabetes have higher bioavailable testos-
terone levels then women without diabetes [26].
An important issue to consider is what absolute drop in
hemoglobin would be considered clinically relevant, thus
prompting treatment in CKD. Despite the presence of
guidelines for desirable hemoglobin in CKD [10], the is-
sue remains controversial, as most publications deal with
treatment after development of anemia, and none, as yet,
has examined the effect of avoidance of anemia on clini-
cal outcomes [27]. Recently, the contribution of anemia to
cardiomyopathy in kidney failure has been examined [27,
28]. Previous studies have demonstrated that even minor
drops in hemoglobin (0.5–1.0 g/dL) may augment the risk
of developing left ventricular hypertrophy in persons with
kidney dysfunction [29], may increase risks of mortality
and heart failure [30], and may lead to a more rapid loss
of kidney function in persons with diabetic nephropathy
[11].
Our study has several limitations. First, we cannot rule
out the possibility that other comorbid conditions, such
as iron-deficiency or other etiologies of anemia, may have
been differentially distributed across the KEEP popula-
tion according to gender and diabetes status. Also, we
have no information regarding the types of medications
[e.g., renin angiotensin system (RAS) drugs, erythropoi-
etin] participants were taking. This might be viewed as
a major limitation of our study given that angiotensin-
converting enzyme (ACE) inhibitors are recommended
treatments for persons with diabetes, and are known to
depress erythropoiesis [31]. However, if our findings were
explained by use of RAS agents, we would have to postu-
late differential exposure and/or effect on erythropoiesis
among men and women. Furthermore, we may have mis-
classified diabetes status given our lack of detailed infor-
mation regarding medications, as well as because fasting
blood glucose levels were not routinely available; how-
ever, 95% of those so categorized reported a diabetes
diagnosis. In addition, our study has the limitations com-
mon to cross-sectional epidemiologic investigations that
preclude the determination of causal associations. Finally,
the study was a targeted-screening of individuals at risk
for CKD and, thus, is not a representative sampling of
the United States population. Self-selection bias is inher-
ent when recruitment strategies rely on volunteer partici-
pants [12]; persons who participate in screening programs
tend to be those who are health conscious and better ed-
ucated. The predominance of women attending KEEP
screenings is consistent with previous observations that
women are more likely than men to utilize preventive
health services [32, 33].
1488 El-Achkar et al: Anemia with diabetes and reduced kidney function
CONCLUSION
In a community-based screening, the prevalence of
anemia was increased at less depressed levels of EGFR in
persons with diabetes compared to persons without dia-
betes. Both men and women with diabetes had increased
prevalence of anemia; however, diabetes conferred
greater odds of anemia in men than in women. These
observations are biologically plausible and provocative.
Nevertheless, these findings require confirmation in other
cohorts that have more complete patient-level data, in-
cluding information on medication exposures.
ACKNOWLEDGMENTS
The Kidney Early Evaluation Program (KEEP)TM is a program of
the National Kidney Foundation, Inc., and was supported by Ortho
Biotech and Bayer Diagnostics.
Reprint requests to John M. Flack, M.D., M.P.H., Wayne State Uni-
versity and the Detroit Medical Center, University Health Center, Suite
2E, 4201 St. Antoine, Detroit, MI 48201.
E-mail: jflack@med.wayne.edu
REFERENCES
1. HSU CY, BATES DW, KUPERMAN GJ, CURHAN GC: Relationship be-
tween hematocrit and renal function in men and women. Kidney Int
59:725–731, 2001
2. ASTOR BC, MUNTNER P, LEVIN A, et al: Association of kidney func-
tion with anemia: The Third National Health and Nutrition Exam-
ination Survey (1988–1994). Arch Intern Med 162:1401–1408, 2002
3. RADTKE HW, CLAUSSNER A, ERBES PM, et al: Serum erythropoietin
concentration in chronic renal failure: Relationship to degree of
anemia and excretory renal function. Blood 54:877–884, 1979
4. CHANDRA M, CLEMONS GK, MCVICAR MI: Relation of serum ery-
thropoietin levels to renal excretory function: Evidence for lowered
set point for erythropoietin production in chronic renal failure. J Pe-
diatr 113:1015–1021, 1988
5. UNITED STATES RENAL DATA SYSTEM: USRDS 1998 Annual Report,
Bethesda, MD, NIH, National Institute of Diabetes and Digestive
and Kidney Diseases, 1998
6. JONES RL, PETERSON CM: Hematologic alterations in diabetes mel-
litus. Am J Med 70:339–352, 1981
7. WINKLER AS, MARSDEN J, CHAUDHURI KR, et al: Erythropoietin de-
pletion and anaemia in diabetes mellitus. Diabet Med 16:813–819,
1999
8. BOSMAN DR, WINKLER AS, MARSDEN JT, et al: Anemia with erythro-
poietin deficiency occurs early in diabetic nephropathy. Diabetes
Care 24:495–499, 2001
9. ISHIMURA E, NISHIZAWA Y, OKUNO S, et al: Diabetes mellitus in-
creases the severity of anemia in non-dialyzed patients with renal
failure. J Nephrol 11:83–86, 1998
10. NATIONAL KIDNEY FOUNDATION: NKF-DOQI Clinical Practice
Guidelines for the Treatment of Anemia of Chronic Renal Failure,
New York, National Kidney Foundation, 2000 update, www.kdoqi.
org
11. KEANE WF, BRENNER BM, DE ZEEUW D, et al: The risk of devel-
oping end-stage renal disease in patients with type 2 diabetes and
nephropathy: The RENAAL Study. Kidney Int 63:1499–1507, 2003
12. BROWN WW, PETERS RM, OHMIT SE, et al: Early detection of kid-
ney disease in community settings: The Kidney Early Evaluation
Program (KEEP). Am J Kidney Dis 42:22–35, 2003
13. JONES CA, MCQUILLAN GM, KUSEK JW, et al: Serum creatinine lev-
els in the United States: the third National Health and Nutrition
Examination Survey (NHANES III). Am J Kid Dis 32:1–9, 1998
14. LEVEY AS, BOSCH JP, LEWIS JB, et al: A more accurate method to
estimate glomerular filtration rate from serum creatinine: A new
prediction equation. Ann Intern Med 130:461–470, 1999
15. NATIONAL KIDNEY FOUNDATION: K/DOQI clinical practice guidelines
for chronic Kidney disease: Evaluation, classification and stratifica-
tion. Am J Kidney Dis 39:S1–S266, 2002
16. WORLD HEALTH ORGANIZATION: Nutritional Anemias: Report of a
WHO Scientific Group, Geneva, Switzerland, World Health Orga-
nization, 1968
17. DALLMAN PR, YIP R, JOHNSON C: Prevalence and causes of anemia
in the United States, 1976–1980 (NHANES II). Am J Clin Nutrition
39:437–445, 1984
18. COCKCROFT DW, GAULT MH: Prediction of creatinine clearance
from serum creatinine. Nephron 16:31–41, 1976
19. LEVIN A, SINGER J, THOMPSON CR, et al: Prevalent left ventricular
hypertrophy in the predialysis population: Identifying opportunities
for intervention. Am J Kidney Dis 27:347–354, 1996
20. HSU CY, MCCULLOCH CE, CURHAN GC: Epidemiology of anemia as-
sociated with chronic renal insufficiency among adults in the United
States: Results from the Third National Health and Nutrition Ex-
amination Survey. J Am Soc Nephrol 13:504–510, 2002
21. BIAGGIONI I, ROBERTSON D, KRANTZ S, et al: The anemia of primary
autonomic failure and its reversal with recombinant erythropoietin.
Ann Intern Med 121:181–186, 1994
22. SCHUSTER SJ, KOURY ST, BOHRER M, et al: Cellular sites of extrarenal
and renal erythropoietin production in anaemic rats. Br J Haematol
81:153–159, 1992
23. ANDERSSON B, MARIN P, LISSNER L, et al: Testosterone concentra-
tions in women and men with NIDDM. Diabetes Care 17:405–411,
1994
24. NAETS JP, WITTEK M. The mechanism of action of androgens on
erythropoiesis. Ann NY Acad Sci 149:366, 1968
25. HOLDSWORTH S, ATKINS RC, DE KRETSER DM: The pituitary-
testicular axis in men with chronic renal failure. N Engl J Med
296:1245–1249, 1977
26. PHILLIPS GB, TUCK CH, JING TY, et al: Association of hyperandro-
genemia and hyperestrogenemia with type 2 diabetes in Hispanic
postmenopausal women. Diabetes Care 23:74–79, 2000
27. DONNE RL, FOLEY RN: Anemia management and cardiomyopathy
in renal failure. Nephrol Dial Transplant 17:37–40, 2002
28. FOLEY RN, PARFREY PS, SARNAK MJ: Clinical epidemiology of car-
diovascular disease in chronic renal disease. Am J Kidney Dis
32:112–119, 1998
29. LEVIN A, THOMPSON CR, ETHIER J, et al: Left ventricular mass index
increase in early renal disease: Impact of decline in hemoglobin.
Am J Kidney Dis 34:125–134, 1999
30. HARNETT JD, KENT GM, FOLEY RN, PARFREY PS: Cardiac function
and hematocrit level. Am J Kidney Dis 25:3–7, 1995
31. KAMPER AL, NIELSEN OJ: Effect of enalapril on haemoglobin and
serum erythropoietin in patients with chronic nephropathy. Scand
J Clin Lab Invest 50:611–618, 1990
32. BERTAKIS KD, AZARI R, HELMS LJ, et al: Gender differences in the
utilization of health care services. J Fam Pract 49:147–152, 2000
33. GREEN CA, POPE CR: Gender, psychosocial factors and the use of
medical services: A longitudinal analysis. Soc Science Med 48:1363–
1372, 1999