Izaks GJ, Westendorp RGJ, Knook DL. The Definition of Anemia in Older Persons. JAMA. 1999;281(18):1714–1717. doi:10.1001/jama.281.18.1714
Author Affiliations: Section of Gerontology and Geriatrics, Department of General Internal Medicine (Drs Izaks, Westendorp, and Knook) and Department of Clinical Epidemiology (Dr Westendorp), Leiden University Medical Center, Leiden, the Netherlands.
Context Whether hemoglobin concentrations defined as
anemia by the World Health Organization (WHO) are associated with
increased mortality in older persons is not known.
Objective To investigate the association between hemoglobin
concentration and cause-specific mortality in older persons.
Design Community-based study conducted from 1986 to 1996
(follow-up period, 10 years).
Setting Leiden, the Netherlands.
Participants A total of 1016 community residents aged 85 years and
older were eligible and 872 agreed to have a blood sample taken.
Hemoglobin concentration was measured in 755 persons (74%).
Main Outcome Measures Hemoglobin concentration, 10-year survival,
and primary cause of death. According to the WHO criteria, anemia was
defined as a hemoglobin concentration below 7.5 mmol/L (120 g/L) in
women and below 8.1 mmol/L (130 g/L) in men.
Results Compared with persons with a normal hemoglobin
concentration, the mortality risk was 1.60 (95% confidence interval
[CI], 1.24-2.06; P<.001) in women with anemia, and 2.29
(95% CI, 1.60-3.26; P<.001) in men with anemia. In both
sexes, the mortality risk increased with lower hemoglobin
concentrations. In persons without self-reported clinical disease at
baseline, the mortality risk of anemia was 2.21 (95% CI, 1.37-3.57;
P=.002). Mortality from malignant and
infectious diseases was higher in persons with anemia.
Conclusions Anemia defined by the WHO criteria was associated with
an increased mortality risk in persons aged 85 years and older. The
criteria are thus appropriate for older persons. A low hemoglobin
concentration at old age signifies disease.
Anemia is defined
by the World Health Organization (WHO),1 but for older persons the criteria for anemia may not be as widely accepted as for
younger people. The hemoglobin concentration is lower on average in
people of older age and could therefore necessitate an adjustment of
the criteria.1 On the other hand, the decline in hemoglobin concentration during aging is small and may not require unique
The aim of this study was to investigate the association between
hemoglobin concentration and mortality in older persons. We used
mortality data to determine whether the WHO criteria for hemoglobin
concentration and the WHO definition of anemia are appropriate for
persons aged 85 years and older. We postulated that a low hemoglobin
concentration represents underlying disease and supports evaluating the
patient for possible causes of anemia.
The subjects were participants of the community-based Leiden
85-plus Study. The primary goal of the study was to investigate the
association between human leukocyte antigen (HLA) phenotypes and
aging.3 Persons were included if they were inhabitants of
Leiden, the Netherlands, and aged 85 years and older at the start of
the study (December 1, 1986). There were no exclusion criteria. The
study population comprised a total of 1258 persons of whom 73% were
women. The Committee on Medical Ethics of the Leiden University Medical
Center approved the study.
All persons were visited at their place of residence. The request for
participation was made by telephone, a short home visit, or through
relatives if the person was residing in a nursing home. After verbal
informed consent, a physician interviewed the persons on their medical
history and activities of daily living and tested their cognitive
function with the Mini-Mental State Examination.4 At the end of the interview, permission was asked to collect a venous blood
sample. The blood samples were drawn during a separate home call.
The blood samples were collected in sterile, EDTA tubes (Becton
Dickinson Vacutainer Systems, Meylan Cedex, France) and processed
within 3 hours in the central laboratories of Leiden University Medical
Center. The measurements were done with an automated system (Coulter
counter, Coulter Electronics, Hialeah, Fla).
Anemia was defined by the criteria of the WHO.1 The reference interval for the hemoglobin concentration was set at 7.5 to
10.0 mmol/L (120-160 g/L) in women and at 8.1 to 11.2 mmol/L (130-180
g/L) in men. The reference interval for the red blood cell volume was
set at 80 to 100 fL in both sexes.
All participants were followed for mortality up to October 1,
1996, for a total follow-up period of 10 years. The places and dates of
death and the numbers of the death certificates were obtained from the
civic registries.5 Two persons who moved abroad during the study period were not included in the analysis.
The primary causes of death were assessed by linking the death
certificate numbers to the causes of death coded by a physician of the
Dutch Central Bureau of Statistics.5 The causes of death were classified according to the International Classification of
Diseases, Ninth Revision.6 From 1996 onward, the death certificates were coded according to the International
Classification of Diseases, 10th Revision.7 For
purposes of the study, they were reclassified according to the ninth
revision. The codes 390 to 459 were categorized as "cardiovascular
disorders," the codes 140 to 239 as "malignant neoplasms," and
the codes 460 to 519 as "respiratory diseases." Selected
respiratory tract infections (codes 460-466.1, 475, 480-487.8,
510.0-510.9, and 513.0-513.1) were included in the category "all
infections." The category "all infections" also included
tuberculosis (codes 010-018.9 and 137-137.4), septicemia (codes
038-038.9), and infections of the kidney and urinary tract (codes
590-590.9 and 599.0). Other infectious diseases were not recorded in
Continuous data are presented as medians and interquartile ranges. The
mortality risk of anemia and the 95% confidence interval (CI) was
estimated by a Cox proportional hazards regression model. (A linear
spline model produced very similar results; only the Cox model is
presented here.) The mortality risk of persons with anemia was
estimated in comparison to persons with a normal hemoglobin
concentration. We used 4 regression models. In model 1, adjustment was
made for age and sex. In model 2, adjustment was made for age, sex, and
self-reported diseases at baseline associated with anemia: malignant
neoplasm, infectious disease, thyroid disease, peptic ulcer, renal
failure, and rheumatoid disease. In model 3, adjustment was made for
age, sex, and functional status defined as any dependency in
activities of daily living and cognitive impairment as measured by a
Mini-Mental State Examination score below 24 points.4 In
model 4, the mortality risk of anemia was estimated only for persons
without self-reported clinical disease, after adjustment for age and
In an additional analysis, the hemoglobin concentration was divided in
categories from 6.5 to 10.0 mmol/L with an increment of 0.5 mmol/L. For
each hemoglobin category, a Cox proportional hazard model was used to
estimate the mortality risk and the 95% CI. The mortality risk for
persons with a particular hemoglobin category was compared with all
Mortality risks were estimated for 2 different periods: 0 to 5 years
after the date of blood sampling and 5 to 10 years after blood
sampling. Survival time was calculated from the date of blood sampling
Differences in age were tested by the Mann-Whitney U test.
Differences in proportions were tested by the χ2 test or, if appropriate, the Fisher exact test. All analyses were performed with
the statistical package SPSS for Windows, version 6.1 (SPSS Inc,
During follow-up, 21 persons who were initially included were found to
not fulfill the age criteria, and 221 persons died before the
interview. Thus, 1016 persons were eligible for the study. Of these, 57
persons (6%) refused the interview, 2 persons were not traceable, and
1 person was erroneously not interviewed. A total of 956 persons
(94%), of whom 872 persons (86% of the eligible number) approved of a
blood sample, gave consent to the interview. If only a small amount of
blood could be drawn, HLA typing was performed first because this was
the primary goal of the study. The hemoglobin concentration was
assessed in 755 persons (74% of the eligible number).
Anemia was found in 17% of the women and in 28%
of the men (Table 1). For both sexes, anemia was
mostly normocytic (80% of all anemic women and 92% of all anemic
men). The median age (interquartile range) was 90 years (88-93 years)
in persons with anemia and 89 years (88-91 years) in persons with a
normal hemoglobin concentration (P=.003).
At baseline, anemia was associated with diseases known to cause a
decrease in hemoglobin concentration. Malignant neoplasms were found in
19 (13%) of the persons with anemia and in 28 (5%) of the persons
with a normal hemoglobin concentration
P<.001). A history of peptic ulcer was present in
3 (2%) of the persons with anemia
compared with none of those with a normal hemoglobin concentration
(P=.03). Any infection was found in 7 (5%)
of the persons with anemia, and in 11 (2%) of the persons with a
normal hemoglobin concentration (P=.07). There
were no differences between persons with anemia and persons with a
normal hemoglobin concentration for other diseases.
At baseline, 29 (21%) of the persons with anemia had no reported
clinical disease compared with 148 persons (26%) with a normal
In the first 5 years after blood sampling, the mortality risk was
increased in persons with anemia (Table 2). For women with anemia, the mortality
risk in this period was 1.60 (95% CI, 1.24-2.06; P<.001)
compared with women with a normal hemoglobin concentration. For men
with anemia, the mortality risk was 2.29 (95% CI, 1.60-3.26;
P<.001) compared with men with a normal hemoglobin
concentration. There were no differences in mortality risk between
persons with microcytic, normocytic, or macrocytic anemia (Table 2).
The mortality risk associated with anemia was similar after
adjustment was made for diseases at baseline associated with anemia,
and also if adjustment was made for functional impairments (Table 2). If the analysis was restricted to persons free from self-reported
clinical disease at baseline, the mortality risk of
anemia was higher (Table 2).
The mortality risk increased with lower hemoglobin
concentrations for both sexes. In women, the mortality risk was
increased in persons with a hemoglobin concentration below 8.0 mmol/L
and was highest in persons with a hemoglobin concentration of 6.5
mmol/L: 2.20 (95% CI, 1.35-3.58; P=.002)
compared with all other concentrations (Figure 1). In men, the mortality risk was
increased in persons with a hemoglobin concentration below 8.5 mmol/L
and was also highest in persons with a hemoglobin concentration of
6.5 mmol/L: 2.54 (95% CI, 1.33-4.87; P=.005)
compared with all other concentrations (Figure 2).
In the period of 5 to 10 years after blood sampling, there were no
differences in mortality risk (Table 2).
During follow-up, 133 (86%) of the 151 persons with anemia had
died compared with 390 (65%) of the 599 persons with a normal
hemoglobin concentration. Malignant neoplasms and infections were more
often noted as the primary cause of death in persons with anemia,
whereas respiratory diseases were more often noted as the primary cause
of death in persons with a normal hemoglobin concentration. The
difference in the distribution of the primary causes of death was
P=.04) (Table 3).
The hemoglobin concentration was not measured in 201 persons who gave
consent to participate in the study. The mortality risk of these
persons was 1.24 (95% CI, 1.03-1.48; P=.02)
compared with the study sample. There were no differences in sex
distribution, age, or primary causes of death.
The mortality risk in persons with anemia, defined
according to the WHO criteria, was increased 2-fold compared with
persons with a normal hemoglobin concentration. The mortality risk
increased with lower hemoglobin concentrations. The association between
a low hemoglobin concentration and increased mortality could not be
explained by diseases at baseline or by functional impairment.
Moreover, a low hemoglobin concentration was also associated with an
increased mortality risk in older persons without clinical disease.
Although clinical disease was determined by patient self-report,
previous research has found self-report to be reasonably
Although chosen arbitrarily, the WHO criteria for anemia
were confirmed by our data. Anemia is defined by the WHO as a
hemoglobin concentration below 7.5 mmol/L (120 g/L) in women and below
8.1 mmol/L (130 g/L) in men.1 These hemoglobin
concentrations are based on data from young persons, but also were
associated with an increased mortality risk in persons aged 85 years
and older. Thus, the WHO criteria for anemia are also appropriate for
persons aged 85 years and older, and age-dependent criteria are not
A cause for anemia is found in most older persons with a low hemoglobin
concentration.10,11 In our study, anemia was also
associated with a poor health status. As in other
studies,10,11 malignant neoplasms and infections were more
often found in older persons with anemia. These diseases were also more
often marked as the primary cause of death. Thus, anemia at old age was
probably due to disease. Because of this increased mortality risk,
hemoglobin levels below normal are a reason for further investigation
of older persons in clinical practice.
The higher prevalence of anemia in men was also described
in other studies. The frequency of anemia varies between 27% to 40%
in men aged 85 years and older and between 16% to 21% in women aged
85 years and older.12,13 The mortality risk of men in our
cohort was slightly higher than in women.3 This suggests
that the higher frequency of anemia in men can be explained by a higher
prevalence of underlying diseases.12
A poor health status might also play a role if the hemoglobin
concentration is high. Compared with the lowest mortality risk, a small
increase in the mortality risk was seen in older persons with a high
hemoglobin concentration. A higher hemoglobin concentration is caused,
for example, by dehydration or obstructive pulmonary disease. In our
study, respiratory diseases were more often noted as the primary cause
of death in persons with a normal hemoglobin concentration than in
persons with anemia. These disorders will lead both to a higher
hemoglobin concentration and an increase in mortality risk.
In conclusion, the mortality risk was increased in older
persons with anemia if anemia was defined by the WHO
criteria.1 Thus, the WHO criteria are appropriate for older persons. Anemia in older persons is due to disease and not to aging.
Therefore, further clinical investigation is warranted if an older
person's hemoglobin concentration is below the WHO normal values—even
if the person is without apparent clinical disease.