Not so benign haematology: anaemia of the elderly


Cindy N. Roy, PhD, JHAAC 1A.44, 5501 Hopkins Bayview Circle, Baltimore, MD 21224, USA.


Developed countries, such as the United Kingdom, are experiencing a change in demographics resulting in the largest proportion of adults over 65 years of age that our health systems have ever experienced. As such, haematologists must be prepared to evaluate and treat anaemia in a more complicated patient population, but sufficient evidence-based guidelines are lacking. Critical next steps that must be taken to ensure the best care of this population include the determination of appropriate haemoglobin concentrations for older adults in light of age, gender, race, and comorbidities; the development of interventional trials that address physical performance outcomes in addition to haemoglobin targets; and translational studies which address the molecular pathogenesis of anaemia in older adults with the most advanced scientific approaches.

The number of people over the age of 65 years in the United Kingdom (UK) has been projected to increase 65% by the year 2033 (UK Office for National Statistics, 2009). In the United States, the number of individuals over 65 years is projected to reach 20% of the entire population by the year 2030, while the number of those over the age of 85 years is projected to increase five-fold by the year 2050 (Day, 1996). Japan is considered the first ‘super-ageing’ nation, as 20% of its population was over 65 years of age in 2005, and this demographic group continues to expand (Nishioka et al, 2011). Clearly, our nations are ageing more rapidly than ever, creating new challenges that our health care systems, standards, and providers must meet.

The health care of robust, community-dwelling older adults remains complicated by the presence of multiple chronic conditions that must be managed simultaneously. Such conditions commonly include obesity, diabetes, osteoarthritis, hypertension, and heart disease. While genetic and environmental factors contribute to the development of these diseases, they are not unexpected and generally considered to be the result of ageing.

Anaemia often appears in the background of these conditions, but may be ignored. If it is mild, it may be considered of secondary importance in light of other chronic conditions with well-known mortality risks. Further, evidence-based guidelines for the diagnosis and treatment of anaemia in older adults are lacking, which makes it difficult for the physician to choose an appropriate course of action. This review will focus on the problem of anaemia in community-dwelling older adults, and the current state of knowledge regarding its pathogenesis and treatment. Epidemiological studies clearly link anaemia with morbidity and mortality in older adults, but these studies must be followed with clinical trials designed to test the hypothesis that anaemia impacts significantly on functional ability and survival. Additionally, we will consider the gaps in our knowledge concerning the impact of ageing on haematopoiesis in general, and erythropoiesis in particular. While the scientific advances of the last decade have given us new insight on the problem of anaemia in older adults, the haematology community must continue working to improve diagnosis, to establish standards of practice for proven interventions, and to develop new, more effective therapies.

Prevalence and clinical impact

Much of our knowledge concerning the impact of anaemia on older adults has been developed by epidemiological studies in Europe and North America, which have used the World Health Organization (WHO) criteria to define anaemia as a haemoglobin concentration of <120 g/l for women and <130 g/l for men (Blanc et al, 1968). These studies indicate that anaemia is common in older adults (Table I), affecting around 10% of adults over 65 years of age (Cesari et al, 2004; Guralnik et al, 2004; Zakai et al, 2005; Culleton et al, 2006; Penninx et al, 2006; Patel et al, 2007, 2009; Dong et al, 2008; Milman et al, 2008). The percentage climbs with age (Nilsson-Ehle et al, 2000; Culleton et al, 2006; Denny et al, 2006; Patel et al, 2007; Milman et al, 2008), increasing to 20–30% for adults over 85 years of age (Izaks et al, 1999; Guralnik et al, 2004). This rapid increase in the incidence of anaemia in the later years of life suggests an underlying pathophysiological mechanism that puts older adults at risk for anaemia. We will discuss the most widely accepted hypotheses for this process in later sections of this review.

Table I.   The incidence of anaemia increases with age.
Study populationReferencesAge (years)Incidence* (%)Mortality risk?
  1. *WHO-defined anaemia (Vardiman et al, 2009).

  2. †Hb < 110 g/l.

Cardiovascular Health Study, USA)Zakai et al (2005)65+8·5Yes
Chicago Health Ageing Project, USADong et al (2008)65+28Yes
Invecchiare in CHIANTI, ItalyCesari et al (2004)65+10·3Not determined
Third National Health and Nutrition Examination Survey, USAGuralnik et al (2004)65+10·6Varies by race (Patel et al, 2009)
Calgary Health Region, Alberta, CanadaCulleton et al (2006)66+13Yes†
Health, Ageing and Body Composition Study, USAPatel et al (2007)70+7–26Varies by race
70-year-old people in Göteborg, SwedenNilsson-Ehle et al (2000)70+4–28Not determined
Established Populations for Epidemiological Studies of the Elderly, USADenny et al (2006); Penninx et al (2006)71+12·5Yes
Milman Study, DenmarkMilman et al (2008)80Men: 18
Women: 17
10-year surviving men had higher Hb
Calgary Health Region, Alberta, CanadaCulleton et al (2006)80+20Not determined
Third National Health and Nutrition Examination Survey, USAGuralnik et al (2004)85+Men: 26
Women: 21
Not determined
Leiden, the NetherlandsIzaks et al (1999)85+Men: 28
Women: 17

In addition to demonstrating the prevalence of anaemia in older adults, these epidemiological studies clearly link anaemia to outcomes that impact importantly on older adults. Anaemia has been significantly correlated with mortality in older adults (Izaks et al, 1999; Guralnik et al, 2004; Zakai et al, 2005; Culleton et al, 2006; Denny et al, 2006; Penninx et al, 2006; Patel et al, 2007; Dong et al, 2008; Milman et al, 2008). Anaemia is also correlated with the development of physical, functional, or cognitive decline [reviewed in (Chaves, 2008)]. In some of these studies, even haemoglobin concentrations at the low end of normal were significantly associated with functional disability (Chaves et al, 2002) or declining quality of life (Thein et al, 2009).

The racial diversity of the United States has allowed investigators to study anaemia in African Americans and Caucasians and has revealed that anaemia occurs with a higher frequency in African Americans than Caucasians (Penninx et al, 2004, 2006; Zakai et al, 2005; Denny et al, 2006; Patel et al, 2007, 2009; Dong et al, 2008). However, data from two of these studies, the Third National Health and Nutrition Examination Survey (NHANES III) and Chicago Health Ageing Project (CHAP), indicated that the haemoglobin threshold for a significant mortality risk in older African Americans is 10 g/l below the WHO criteria for anaemia (Dong et al, 2008; Patel et al, 2009). This finding has significant implications for the design of clinical trials that include African Americans, as they may not benefit from interventions designed to increase haemoglobin unless their starting haemoglobin is sufficiently below the WHO cut-off.

For over a decade, there has been much debate, but no consensus, among experts regarding the need to revise the WHO criteria for anaemia in older adults. Some investigators have suggested the decline in haemoglobin that occurs commonly in later decades might warrant lower haemoglobin limits for older adults (Nilsson-Ehle et al, 2000), while others submit the WHO criteria are appropriate given the increased mortality associated with haemoglobin levels below the WHO cut-offs (Izaks et al, 1999; Milman et al, 2008). Still others have suggested that a haemoglobin cut-off higher than that established by WHO might be appropriate in light of data that suggested low-normal haemoglobin is associated with functional (Chaves et al, 2002) and cognitive (Deal et al, 2009) decline. However, as these studies have also demonstrated that high haemoglobin concentrations are associated with mortality in older adults (Chaves et al, 2002; Culleton et al, 2006; Denny et al, 2006), a higher haemoglobin cut-off has not been widely accepted. Some investigators have proposed to include comorbidities (Chaves et al, 2004) when assessing whether the haemoglobin concentration is appropriate, while others have indicated that an ‘optimal’ haemoglobin concentration (Milman et al, 2008), rather than a population-based normal, may be most important. This approach is similar to current recommendations regarding target low density lipoprotein cholesterol levels in patients with heart disease compared to those without risk factors (Boehringer & Darden, 2006).

Gathering sufficient data to determine rates of morbidity and mortality at various haemoglobin concentrations based on age, race, sex and comorbidities, are critical next steps for ensuring the appropriate clinical care for every patient. With better data concerning the range of haemoglobin at which these patient groups are significantly impacted, the haematology community could more effectively engage primary care providers to coordinate referrals, ensure appropriate treatment, and optimize inclusion and exclusion criteria for clinical trials.

Epidemiological studies cannot attribute a causal role to anaemia in the development of functional decline in older adults. It is possible that anaemia is merely a sensitive marker of systemic physiological decline. It is also possible that the pathological process that causes anaemia independently causes other hallmarks of ageing, such as decreased vascular tone, decreased pulmonary function, decreased muscle mass and cognitive decline, such that improving anaemia would not have any impact on functional status. Therefore clinical trials of interventions that aim to improve haemoglobin must also improve functional outcomes before such treatments should gain wide acceptance in medical practice.

Current standard of care

The epidemiological evidence in the preceding section indicates that the WHO haemoglobin cut-off for anaemia may not be optimal for older adults. While alternative cut-offs have been proposed, they have not been widely accepted. Until that time, the clinical evaluation of elderly patients who present with anaemia should proceed in the same way as would be approached in younger adults. Recently, two separate studies have outlined similar criteria to effectively discriminate between possible aetiologies of anaemia in older adults (Table II). Serum vitamin B12 levels, folate, and iron studies should be evaluated to rule out deficiencies and thyroid function tests should be used to rule out hypothroidism. The patient’s clinical history should be sought to evaluate for anaemia of inflammation or chronic disease and patients should be evaluated for signs and symptoms of chronic infections or autoimmune disease. When available, reticulocyte haemoglobin (CHr) <28 pg is a sensitive indicator of functional iron deficiency because it measures the haemoglobin content of the erythrocytes most recently produced (Brugnara, 2003). Soluble transferrin receptor (sTfR) is produced when the transferrin receptor is shed after erythroblast haemoglobin production is complete. sTfR has been used in combination with serum ferritin (sFt) to discriminate between anaemia of inflammation (sTfR/log sFt <0·8) and iron deficiency (sTfR/log sFt >1·5) (Brugnara, 2003). Serum and urine protein electrophoresis should be performed to rule out plasma cell dyscrasias; however, given that monoclonal gammopathy of undetermined significance (MGUS) is common in this population, other signs of progressive disease, such as hypercalcaemia, lytic bone lesions, renal dysfunction, increased clonal plasma cells in the bone marrow (>5%), or rising monoclonal proteins should be sought before anaemia is attributed to a plasma cell dyscrasia. Renal dysfunction [estimated glomerular filtration rate (eGFR) < 30 ml/min] should be excluded as a cause of anaemia. Erythropoietin levels can be accurately measured and although they can predict responsiveness to erythroid stimulating agent (ESA) therapy in patients with myelodysplastic syndrome (MDS), they do not have a clear role in the diagnostic work up of anaemia in older adults. Interestingly, there appears to be no correlation between erythropoietin level and haemoglobin in patients with unexplained anaemia (Ferrucci et al, 2007; Artz & Thirman, 2011; Price et al, 2011; Waalen et al, 2011). Megaloblastosis (MCV > 100 fl) or other cytopenias (absolute neutrophil count <1·2 × 109/l or platelet count <120 × 109/l) should prompt bone marrow evaluation to rule out a primary bone marrow disorder, such as aplastic anaemia or MDS. Patients with severe microcytosis or microcytic anaemia without iron deficiency should be evaluated for haemoglobinopathies.

Table II.   Evaluation for underlying causes of anaemia.
  1. Adopted from (Artz & Thirman, 2011) and (Price et al, 2011).

  2. TSH, thyroid-stimulating hormone; SPEP, serum protein electrophoresis; UPEP, urine protein electrophoresis; eGFR, estimated glomerular filtration rate; MCV, mean corpuscular volume; RBC, red blood cell; HIV, human immunodeficiency virus; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; AI, anaemia associated with inflammation; UAE, ‘unexplained’ anaemia of the elderly.

Vitamin B12 deficiencyVitamin B12 < 200 ng/lEvaluation of methylmalonic acid levels or a treatment trial may be considered for a vitamin B12 level of 200 – 300 ng/l
Excluded if not corrected by 3 month trial of vitamin B12
Folate deficiencyFolate level < lower limit of normalExcluded if not corrected by 3 month trial of folate
Iron deficiencySerum ferritin <50 μg/lExcluded if not corrected by 3 month trial of iron therapy or there is evidence of bleeding, h. pylori infection or coeliac disease
Thyroid dysfunctionTSH < 0·1 mcU/ml or >10 mcU/mlThere is some debate as to ‘normal’ TSH level in older adults
Anaemia of chronic inflammation/disease1) Auto-immune diseases
2) Chronic infection or recent acute infection
Absent a diagnosis of autoimmune disease (systemic lupus erythematosis, rheumatoid arthritis, inflammatory bowel disease, etc.) or documented chronic infection (HIV, viral hepatitis, tuberculosis, osteomyelitis)
Low serum iron and elevations of CRP, ESR can be found in both AI and UAE and do not distinguish the two states
Plasma Cell DyscrasiaMonoclonal protein on SPEP/UPEPMonoclonal gammopathy < 10 g/l, absence of bone lesions on skeletal survey, normal calcium, <5% light chain restricted plasma cells, and no other evidence of progressive disease suggests that plasma cell dyscrasia is not cause of anaemia
Renal insufficiencyeGFR <30 ml/minThere is some evidence that more moderate renal insufficiency (eGFR 31–60 ml/min) may be associated with UAE
Myelodysplastic SyndromeBone marrow evaluation using the World Health Organization criteria (Vardiman et al, 2009)Bone marrow examination should be considered if MCV ≥ 100 fl, a platelet count <120 × 109/l, or a neutrophil count <1·2 × 109/l or if patient prior chemotherapy, radiation therapy, an abnormal peripheral blood smear
Thalassaemia traitMCV < 80 fl, normal RBC count, replete iron0 stores and appropriate ethnic group. Elevated Hb A2 (>3·5%)Patients with a documented decline of Hb of 15 g/l from baseline should be evaluated for other causes of anaemia
Androgen DeprivationHormonal therapy prior 12 months 
AlcoholMore than 3–4 drinks per dayAlcohol can cause anaemia by multiple mechanism including nutrient deficiency, iron deficiency from gastrointestinal bleeding/gastritis or direct marrow toxicity

Therapy for anaemia in the elderly should be directed at the underlying causes. However anaemia will remain unexplained in a significant number of patients. There is currently no established standard of care to treat anaemia in these patients, although clinical trials examining this question are underway. In the next section, we will review the putative causes of anaemia in older adults and the progress in defining them.

Putative causes of anaemia in older adults

While the scale of epidemiological studies usually prevents an exhaustive laboratory work-up of anaemia for most subjects, some studies have employed common biochemical indices to sufficiently distinguish between nutrient deficiencies, anaemia associated with inflammation (AI), and other ‘unexplained’ aetiologies (Guralnik et al, 2004; Ferrucci et al, 2007; Waalen et al, 2011). Below, we focus the bulk of our discussion on anaemia resulting from unexplained aetiologies. Given that there is considerable literature concerning the diagnosis and treatment of anaemia resulting from nutrient deficiencies or chronic disease states that has been reviewed elsewhere (Carmel et al, 2003; den Elzen et al, 2010; Goodnough et al, 2010; Roy, 2010), we will only review these topics as they apply specifically to older adults.

Unexplained anaemia in the elderly

Despite high quality biochemical data in epidemiological studies such as NHANES III (Guralnik et al, 2004), InCHIANTI (Aging in the Chianti area; Ferrucci et al, 2007), and the Scripps-Kaiser study (Waalen et al, 2011), a large percentage of anaemia in older adults could not be attributed to any clear cause and was therefore referred to as ‘unexplained’ anaemia of the elderly (UAE). A limitation of these epidemiological studies is that subjects do not return to the clinic for follow-up examinations or bone marrow evaluation. However, a few prospective studies have emerged recently from active outpatient anaemia clinics indicating that a significant proportion of anaemia in older adults remains unexplained after rigorous diagnostic evaluation (Tettamanti et al, 2010; Artz & Thirman, 2011; Price et al, 2011).

UAE becomes a ‘diagnosis of exclusion’ that varies greatly from one study to another depending upon the specific criteria chosen (Table III). In order for our understanding of anaemia in the elderly to advance, a consensus must be reached in the exclusion criteria necessary to arrive at a diagnosis of UAE. The data available (Guralnik et al, 2004; Ferrucci et al, 2007; Tettamanti et al, 2010; Artz & Thirman, 2011; Perlstein et al, 2011; Price et al, 2011; Waalen et al, 2011) are conflicting regarding the impact of age, serum erythropoietin (Epo) concentration, inflammation, or comorbidities on UAE. Differences in inclusion and exclusion criteria among these studies may contribute importantly to these inconsistencies. Still, most studies agree UAE is mild, with median haemoglobin concentrations higher than iron deficiency anaemia (IDA) and more akin to AI (Guralnik et al, 2004; Ferrucci et al, 2007; Artz & Thirman, 2011; Price et al, 2011; Waalen et al, 2011). We propose that the criteria selected by Artz and Thirman (2011) and Price et al (2011), outlined in Table II, be adopted by the field as a consensus definition of UAE if none of the listed conditions can be applied.

Table III.   Summary of findings in UAE subjects.
Study locationUAE% of all anaemiaReferencesAge range (years)Unique exclusion criteriaEpoInflammationOther findings
  1. UAE, ‘unexplained’ anaemia of the elderly; AI, anaemia associated with inflammation; Hb, haemoglobin concentration; Epo, erythropoietin; IDA, iron deficiency anaemia; MDS, myelodysplastic syndrome; CRP, C-reactive protein; IL-6, interleukin-6; ESR, erythrocyte sedimentation rate; eGFR, estimated glomerular filtration rate; GDF15, growth differentiation factor 15.

InCHIANTI36·8%Ferrucci et al (2007)81 ± 8NoneLow (adjusted for age, sex and Hb)CRP/IL-6 lower than non-anaemic controlsvitaminB12/folate anaemia also have Epo lower than controls
Low lymphocyte count
Osteoarthritis and Parkinsons over represented
Chicago anaemia referral43·7%Artz and Thirman (2011)≥65, median 76AI diagnosis requires active inflammatory disorder
Bone marrow-confirmed haematological malignancy
Low compared to IDA (adjusted for Hb)47% CRP >3 mg/l21% macrocytic or pancytopenic
39·7% of UAE overlap with AI by serum Ferritin
Stanford Hospital and Paolo Alto VA35%Price et al (2011)77·8 ± 7·2AI diagnosis requires active inflammatory disorder
Bone marrow-confirmed haematological malignancy
Low compared to MDS and IDAESR, Hepcidin, Ferritin higher than non-anaemic controls52% of UAE have eGFR 30–60 ml/min/1·73 m2
Salute e Anaemia26·4%Tettamanti et al (2010)Not specifiedNoneNot determinedNot determined1/3 macrocytic, leucopenic or thrombocytopenic
Scripps/Kaiser19·5%Waalen et al (2011)74 ± 5·5AI diagnosis requires active inflammatory disorderHigh compared to controls, low compared to IDANo change in IL-6 or HepcidinHigh GDF15 correlates with creatinine low testosterone

We will review some of the common hypotheses concerning the biological basis of UAE below. Many hypotheses concerning UAE overlap with concepts that are an integral part of our understanding of ageing biology in general (Fig. 1).

Figure 1.

 Possible causes of anaemia in older adults. Erythroid prognitors differentiate from haematopoietic stem cells (HSCs) in the context of a complex microenvironment. Systemic effects of acute or sub-clinical inflammation, erythropoietin (Epo) production and micronutrient availability all have the potential to negatively impact erythroid maturation and haemoglobin production. Additionally, defects intrinsic to HSCs, such as telomere length, DNA integrity and epigenetic marks, have the potential to impair maturation along the erythroid lineage. RBC, red blood cell; ROS, reactive oxygen species; Hepc, hepcidin antimicrobial peptide.


Epo is the primary factor promoting the survival of proliferating erythroid precursors. Thus, it is a central regulator of erythropoiesis. Epo levels have been shown to increase with age in individuals with a haemoglobin concentration of at least 140 g/l in the Baltimore Longitudinal Study of Ageing (BLSA) (Ershler et al, 2005). These data suggest that the efficiency of Epo signalling to promote erythropoiesis declines with age. Within the context of UAE, there is considerable agreement that Epo levels are low or relatively low when compared to IDA (Ferrucci et al, 2007; Artz & Thirman, 2011; Price et al, 2011; Waalen et al, 2011). However, Epo response seems to be uniquely sensitive to iron deficiency, which complicates using IDA as a control group. This unique sensitivity to iron deficiency is probably due to the requirement of iron for stabilization of the prolyl-hydroxylase, which promotes degradation of hypoxia inducible factor (HIF), a critical transcription factor required for expression of Epo (Peyssonnaux et al, 2008).

Outside of the concern regarding the comparison to IDA, there is ambiguity as to whether Epo levels in subjects with UAE are higher (Waalen et al, 2011) or lower (Ferrucci et al, 2007) than non-anaemic controls, individuals with AI, or individuals with vitamin B12 or folate deficiencies. The observation that Epo production is impaired in UAE subjects, when compared to subjects with IDA, is similar to the ‘blunted Epo response’ described for rheumatoid arthritis patients, which suggested Epo is not produced sufficiently in keeping with the degree of anaemia (Baer et al, 1987; Hochberg et al, 1988). This could be the result of aberrant hypoxia sensing (Agarwal & Prchal, 2008), including the reduced activity of HIF that occurs with age (Frenkel-Denkberg et al, 1999; Rivard et al, 2000). Further studies designed to assess the sensitivity of HIF activity in individuals with UAE may improve our understanding of the hypoxia-Epo axis in UAE.

An alternative description of the ‘blunted Epo response’ in paediatric cancer patients is that erythroid progenitors fail to respond appropriately to available erythropoietin (Corazza et al, 1998). Given that Epo increased with age to maintain normal haemoglobin in healthy older adults of the BLSA, this might indicate that a similar inefficiency in the erythroid progenitor response to Epo occurs with age. This deficiency in the erythoid progenitor compartment, especially if it is not met with increased Epo production, may participate in the development of UAE. We will discuss the hypothesized deficiencies intrinsic to erythroid progenitors in more detail when considering the role of the stem cell (see below).

The pro-inflammatory state of ageing

Many patients with UAE display features of inflammation. A subclinical pro-inflammatory state independent of disease is widely accepted as one of the basic biological processes underlying ageing (Roubenoff et al, 1998; Franceschi, 2007; Adler et al, 2008). Though the mechanisms driving inflammation with age remain unclear, ageing-related processes that result in the accumulation of reactive oxygen species (ROS), the accumulation of fat mass, or the accumulation of senescent cells activate nuclear factor kappa B (NFκB). NFκB is the transcription factor central to activation of the innate immune response and induces expression of numerous pro-inflammatory cytokines (Hanada & Yoshimura, 2002).

Classically, inflammation is characterized by the cardinal signs: rubor (redness), tumour (swelling), calor (heat), and dolor (pain). The inflammation associated with ageing is more protean in its signs and has been inferred from sensitive laboratory tests. The erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) are two commonly used tests for systemic inflammation. CRP has been used as a marker for low-grade inflammation in the development of successful clinical trials for the treatment of ageing-related disorders, such as cardiovascular disease (Ridker et al, 2008) and diabetes (Fleischman et al, 2008), which underscores its validity as a systemic marker of mild inflammation in older adults.

Other important mediators of inflammation used primarily for research include the pro-inflammatory cytokines interleukin-6 (IL-6), tumour necrosis factor alpha (TNFα), and interferon gamma (IFNγ). IL-6 induces the expression of Hepcidin antimicrobial peptide (Hepc) (Nemeth et al, 2004a), a critical mediator of AI, which decreases serum iron availability (Nemeth et al, 2004b) and correlates with serum ferritin (Ganz et al, 2008).

Artz and Thirman (2011) restricted the diagnosis of AI to individuals with diagnosed chronic inflammatory conditions. In doing so, they found 47% of their UAE subjects with no diagnosed inflammatory condition had CRP > 3 mg/l and 39·7% of their UAE subjects had low serum iron concentrations despite an elevated serum ferritin for the group overall. Similarly, Price et al (2011) found increased ESR, Hepc, and ferritin in their UAE subjects with no diagnosed inflammatory condition when compared to non-anaemic controls, but did not find IL-6 levels to be significantly higher than the control group of older adults. Ferrucci et al (2010) found mildly increased IL-6 and a trend toward increased Hepc among InCHIANTI UAE subjects. These features of inflammation, though low-grade, overlap considerably with the clinical features of AI (Weiss & Goodnough, 2005).

Obesity is a common finding in older adults. In younger cohorts, obesity has been associated with hypoferraemia (Yanoff et al, 2007; Cepeda-Lopez et al, 2011), and increased Hepc expression (Bekri et al, 2006) similar to the AI phenotype. Despite observed iron sequestration, anaemia is not a routine feature of the younger obese cohorts that have been investigated (Yanoff et al, 2007; Ausk & Ioannou, 2008; Cepeda-Lopez et al, 2011). Whether iron-restricted erythropoiesis in the context of obesity may place additional stress on the aged erythron and promote anaemia is a question that remains unanswered.

The special case of renal disease: not just Epo

Chronic diseases, such as diabetes and hypertension, are common in older adults and are associated with renal decline. Most studies of UAE exclude patients with severe chronic kidney disease (eGFR < 30 ml/min/1·73 m2 or creatinine ≥ 123·76 μmol/l). However, Price et al (2011) found half of the UAE participants in their study had stage 3 renal disease (eGFR between 30 and 60 ml/min/1·73 m2). Waalen et al (2011) did not stratify anaemic participants by eGFR (Waalen et al, 2011). However, they found that increased concentrations of the stress-induced serum cytokine Growth Differentiation Factor 15 (GDF15), which is produced, in part, by erythroid precursors (Tanno et al, 2010), but also tracks with kidney injury (Zimmers et al, 2005; Lajer et al, 2010) positively correlated with creatinine level. While inflammation is not overt in this group, the presence of an underlying chronic disease state (renal insufficiency), inappropriately low Epo production, and the presence of GDF15, an indicator of impaired response to Epo, certainly overlap with features of AI.

Endocrine dysfunction

As with other physiological systems, the function of the endocrine system declines with age (Makrantonaki et al, 2010). Low testosterone has been associated with the development of anaemia in older men and women (Ferrucci et al, 2006) and specifically with UAE (Waalen et al, 2011). Androgens have long been known to promote erythropoiesis (Shahidi, 1973; Rhoden & Morgentaler, 2004), though the mechanism driving this effect is unknown. Oestrogen has also been shown to stimulate telomerase activity (Bayne et al, 2007). Telomerase is the enzyme that maintains genomic integrity by protecting chromosome ends throughout the process of mitosis. This is a critical feature of all cells, but especially those with high replicative capacity, such as haematopoietic stem cells (HSCs) and erythroid progenitors. By inducing telomerase activity, oestrogen would be expected to maintain genomic stability and, therefore, cellular viability. Additionally, hypothyroidism can also cause anaemia, but it is unknown if the mild increase in thyroid-stimulating hormone observed in ageing (Laurberg et al, 2011) contributes to UAE.

Marrow toxic drugs

Both prescription and non-prescription drugs can contribute to anaemia in older patients due to direct marrow toxicity from the drugs. Alcohol is probably the most common non-prescription offender, while a large number of prescription drugs can cause anaemia: chemotherapy, immunosuppressive drugs such as azathioprine, or mycophenolayte, drugs that inhibit folate metabolism, drugs against human immunodeficiency virus or other viral diseases, as well as idiosyncratic reactions. In these cases a careful history to determine whether the onset of anaemia coincided with a new medicine can be highly suggestive. Proving that a medication is causing anaemia, however usually requires discontinuation of the drug followed by observation.

Stem cell ageing

HSC function has been shown to decline with age and may contribute to anaemia in the elderly. The two hallmark features of HSC function – their ability to self-renew and to differentiate into mature blood lineages – are both affected by the ageing process. Loss of self-renewal in aged HSC has been demonstrated by a declining ability of HSCs from aged mice to reconstitute haematopoiesis after transplantation (Sudo et al, 2000; Chambers et al, 2007). Despite a decline in HSC function, phenotypically defined stem cells appear to increase in number in the marrow of aged animals (Harrison, 1983) implying that the loss of bone marrow HSC function is due to qualitative loss of HSC function rather than a decline in HSC numbers. These changes are thought mostly to be intrinsic to HSC and include accumulated DNA damage, telomere shortening and epigenetic changes (Gazit et al, 2008). Transplantation studies of aged HSC into younger mice appear to only partially rescue HSC function, however, suggesting that both intrinsic and extrinsic factors play a role in HSC dysfunction with ageing (Kamminga et al, 2005).

HSC can be made to prematurely ‘age’ through serial transplantation experiments where bone marrow from one recipient animal is transplanted into secondary or tertiary recipients and beyond. This leads to a progressive loss of self-renewal. Transplantation is thought accelerate replicative senescence by requiring transplanted cells to rapidly proliferate to repopulate the recipient host. Mathematical models have suggested that this type of HSC exhaustion would take several life-times under homeostatic conditions and would not be relevant to normal ageing, but could contribute to anaemia in subset of patients who experienced extensive haematopoietic demands over a lifetime (Glauche et al, 2011).

Inflammation and resulting changes in the haematopoietic micro-environment are thought to additionally contribute to HSC dysfunction with age. NFκB signalling was shown to be activated in aged HSC (Chambers et al, 2007). Furthermore, IFN signalling, TNFα and toll-like receptor have all been shown to regulate HSC function (Baldridge et al, 2011).

Aged mouse bone marrow demonstrates a lineage bias towards myeloid and against lymphoid development. This is thought to result from an expansion of myeloid biased HSC (Beerman et al, 2010), a process that may be mediated by common ageing-related mechanisms, such as cytokine involvement (Challen et al, 2010) or mitochondrial dysfunction (Norddahl et al, 2011). Most studies of HSC function in anaemia have not rigorously evaluated HSC function largely because of technical challenges. The CD45.1/CD45.2 congenic system, which enables detailed evaluation of mouse blood and marrow chimerism by flow cytometry, is not useful for evaluation of erythrocytes that do not express CD45. Furthermore, DNA-based techniques cannot be used in enucleated erythrocytes.

The links between human HSC dysfunction and the development of anaemia in the elderly remain speculative. Ageing in human HSC appears to be associated with accumulated DNA damage, increased ROS, methylation changes and telomere dysfuction (Chambers & Goodell, 2007; Gazit et al, 2008; Calado & Young, 2009; Bocker et al, 2011). Studies of ageing on human HSC are limited, and self-renewal cannot be tested in serial transplantation. Transplantation into immunocompromised mice has been used as a surrogate for tranplantation into humans. These studies have revealed a similar defect in lymphopoiesis in aged human marrow but no clear myeloid bias (Kuranda et al, 2011), which is contrary to the phenotype that has been seen in mouse models (Waterstrat & Van Zant, 2009).

MDS is a heterogeneous haematopoietic disease that preferentially affects older adults (Greenberg et al, 2002). Characterized primarily by cytopenias, sub-classes of the disease carry differential risks of malignant transformation. Molecular features of MDS, including aberrant mitochondrial function and short telomeres (Young, 2010), reflect the biology classically associated with ageing in general. Sub-clinical MDS or very early MDS can present with isolated normocytic anaemia as an initial sign and should be a consideration in all patients with UAE. Patients with macrocytosis (MCV > 100 fl) of unknown cause or other cytopenias who decline a bone marrow examination should be considered suspicious for MDS. Bone marrow studies often do not reveal significant dysplasia or cytogentic abnormalities to warrant a diagnosis of MDS (Price et al, 2011). Improved diagnostic testing may allow reclassification of these patients from UAE to MDS and is an area of active research.

Nutrient deficiencies

NHANES III attributed about 1/3 of the anaemia in adults over 65 years of age to nutrient deficiencies based on serum biochemical data. In this large epidemiological study, a confirmatory trial of the deficient nutrient was not feasible to refine the ‘diagnosis’. Approximately half of the anaemia in this category (16·6% of all anaemia) resulted from iron deficiency, while the remaining was attributed to vitamin B12, and folate deficiency (Guralnik et al, 2004). The results from the InCHIANTI study were quite similar, with 16·7% of all anaemia attributed to iron deficiency and 10·5% attributed to vitamin B12 or folate deficiency (Ferrucci et al, 2007). Older adults with IDA should be screened for gastrointestinal blood loss with a fecal occult blood test or endoscopy that might attribute losses to gastrointestinal lesions (Rockey & Cello, 1993). Similarly, malabsorption from gastrointestinal disease, or a strict vegan diet are the most likely causes for vitamin B12 deficiency (Carmel et al, 2003). Folate deficiency can be caused by malnutrition, particularly in institutionalized individuals without access to fresh fruits or vegetables; certain drugs which interfere with folate metabolism (phenytoin, methotrexate, sulfasalazine, triamterene, pyrimethamine, trimethoprim); and malabsorption (Andres et al, 2008). Malabsorption can be caused by autoimmune gastritis (evaluate for gastrin and antiparietal antibodies), atrophic gastritis, Helicobacter Pylori infection, coeliac disease, inflammatory bowel disease, or bariatric surgery (Hershko et al, 2006; Fernandez-Banares et al, 2009; Rashtak & Murray, 2009). It is important to note that true vitamin B12 and folate deficiencies are relatively rare in older adults (Carmel et al, 2003; Artz & Thirman, 2011; Price et al, 2011). Additionally, despite the fact that serum micronutrient levels may be below normal, the anaemia may be the result of another pathological process independent of the micronutrient deficiency (Carmel et al, 2003).

Chronic inflammation or disease

NHANES III attributed about 1/3 of the anaemia in adults over 65 years of age to chronic inflammation. Based on the common physiological changes that occur in the context of infection, AI was defined by anaemia, low serum iron concentration and no other evidence of iron deficiency (Guralnik et al, 2004). This broad, but appropriate, characterization alleviated the need for defining a disease or source of inflammation for each individual represented in the study. However, as reviewed above, there is some overlap with UAE and this broad definition of AI.

While AI has classically been associated with inflammation resulting from the innate response to pathogens, other chronic disease states that weren’t previously appreciated as inflammatory diseases are now considered to have some features of AI (Roy, 2010) including chronic kidney disease (Adamson, 2009) and chronic heart failure (Ezekowitz et al, 2003; van der Putten et al, 2008; Parikh et al, 2011). Of the NHANES III individuals with AI, about 1/3 (approximately 12% of all anaemia) were diagnosed with anaemia of chronic renal failure and 12% (approximately 4% of all anaemia) were diagnosed with congestive heart failure (Guralnik et al, 2004).

Hepc is the primary regulator of serum iron availability (Ganz & Nemeth, 2009). While clearly elevated in the serum of individuals with acute inflammation (Nemeth et al, 2003; Roecker et al, 2005; Sihler et al, 2010), the evidence for Hepc levels in older adults with AI is not clear. A small study in the anaemic elderly in California found that Hepc was increased in a group of male patients with AI when compared to young or aged non-anaemic controls (Lee et al, 2008), but this was not the case for females. In contrast, Hepc was found to be decreased in subjects with AI from the large InCHIANTI study when compared to non-anaemic older adults and comparable to Hepc concentrations in older adults with vitamin B12 or folate deficiency anaemia (Ferrucci et al, 2010). The findings in the InCHIANTI study may indicate that AI in older adults is more closely related to the direct effects of pro-inflammatory cytokines on erythroid progenitors, Epo production, and marrow response to Epo (Weiss & Goodnough, 2005) than it is on iron availability regulated by Hepc.

The problem with the broad definition of AI for older adults is that the anaemia is often associated with a chronic disease state that is managed rather than a pathogen that can be treated. AI associated with a pathogen or autoimmune disorder will reverse with antimicrobial treatment or immunosuppression to treat the underlying disease, but conditions such as diabetes (Symeonidis et al, 2006), chronic kidney disease (van der Putten et al, 2008; Adamson, 2009) or chronic heart failure (Le Jemtel & Arain, 2010; Parikh et al, 2011) are intractable and managed rather than cured. As a result, many older adults with AI in the context of these diseases have anaemia that persists despite the known risk of morbidity and mortality.

Treatment strategies and research opportunities

The need for clinical trials in older patients with anaemia

In the previous sections, we have reviewed the prevalence of anaemia in older adults and its association with declining measures of cognition and physical performance, as well as the association with an increased risk of death. Though anaemia is correlated with these critically important outcomes, we have also reviewed the literature, which demonstrates that anaemia in older adults is often mild, can be multifactorial or unexplained, and is usually found in the background of other chronic conditions. Only with an improved understanding of the pathophysiology of anaemia in the elderly, will we be able to refine diagnostic criteria to reduce the numbers of patients in whom anaemia remains unexplained and target appropriate therapies at the underlying pathophysiology.

Anaemia in older adults may remain under-recognized and under-diagnosed because of the paucity of clinical data demonstrating safe and effective treatment strategies for this population. Older adults are largely under-represented in clinical trials, therefore the existing data for the treatment of anaemia may not be applicable to older patients. Clinical trials in anaemic older adults will be required to build an appropriate evidence base concerning the safety and efficacy of anaemia treatment. A recent randomized, double-blind, placebo-controlled trial of epoietin alpha in adults over 65 with chronic anaemia demonstrated improvements in haemoglobin and self-reported measures of fatigue and quality of life (Agnihotri et al, 2007). This study clearly demonstrated that trials in anaemic older adults can be successfully designed and safely executed, but more trials are needed. Because recruitment of older adults for clinical trials can be difficult, regional and national cooperatives that can share human subjects research infrastructure are sure to excel in the development of such trials.

The criteria used to define anaemia in the elderly will be critical for interventional trials, as subtle differences in inclusion and exclusion criteria used to date have somewhat confounded our understanding of the pathogenesis of anaemia in older adults. Whether to simplify these criteria to capture the broadest clinical picture of anaemia and focus on functional outcomes, or whether to continue to stratify subgroups of subjects, so as to test the efficacy of a specific intervention in a specific subset of anaemic older adults will depend on the hypothesis being tested.

The most informative trials will be designed for dual outcomes assessing haemoglobin concentration and functional measures. Health-related quality of life instruments, such as the Short Form-36 Health Survey (SF-36) (Ware & Sherbourne, 1992) and the Functional Assessment of Chronic Illness Therapy-Anaemia (FACIT-An) (Cella, 1997), have demonstrated sensitivity to haemoglobin concentration in the context of cancer-related anaemia (Lind et al, 2002), chronic kidney disease (Alexander et al, 2007) and ageing (Thein et al, 2009). These self-reported measures are a relatively convenient way to collect important data on physical performance without significant burden to older adults.

Objective physical performance measures can be ascertained in older adults within the context of an office visit, but require more intensive participation from research subjects. Single measures or summary measures of grip strength, standing balance, timed walk tests, and timed tests to rise from a chair are just some examples of outcomes that have been used effectively to document the inverse relationship between anaemia and physical performance in observational studies of older adults [reviewed in (Chaves, 2008)].

Immediate opportunities for clinical trials

Blood transfusion is the most direct method for increasing haemoglobin concentration in any individual. However, the use of such a precious resource for individuals with mild to moderate anaemia may be difficult to espouse. Furthermore, the risks of repeated blood transfusions, such as infection, transfusion reactions and iron overload do not make regular transfusion therapy an attractive option for patients with chronic anaemia. Blood transfusions for anaemic older adults are most commonly accepted in the peri-operative period for individuals with haemoglobin <80 g/l, but have not been shown to improve survival for older adult hip fracture patients without coronary artery disease and with haemoglobin >80 g/l (Crosby, 2002). Data from the United States Centers for Medicare and Medicaid Services indicated that older adults receiving tranfusions were more likely to be over 85 years of age, to be African- or Mexican-American, to have a low body mass index (BMI), to smoke, or to have a history of cancer, diabetes mellitus, end stage renal disease, heart disease or lung disease than adults who did not receive a transfusion (Rogers et al, 2011).

A number of therapeutic interventions are currently available that could be considered for trials to treat anaemia in older adults. ESAs were widely used in individuals with mild to moderate anaemia before March 2007 when the United States Food and Drug Administration issued a strong warning concerning their use. Recent data suggests ESAs may increase the risk of thrombosis and cancer progression [reviewed in (Barbera & Thomas, 2010)] – two adverse events for which older adults are already at increased risk. A better understanding of the factors that put patients at risk for the development of these adverse events will be necessary before an appropriate trial of available ESAs can be developed in anaemic older adults. Additionally, more data is necessary to determine whether anaemic older adults are truly Epo-deficient, as the existing data we have discussed above are inconclusive.

Intravenous iron has been used successfully to improve haemoglobin response to ESAs in dialysis patients (Coyne et al, 2007) and to improve functional capacity in anaemic older adults with heart failure (Anker et al, 2009). While intravenous iron may improve haemoglobin response in older adults with functional iron deficiency or an iron-dependent blunted erythroid response to available Epo, the inhibitory effects of intravenous iron on HIF transcriptional activity of Epo could, conversely, suppress Epo production and provide no benefit to older adults (Agarwal & Prchal, 2008).

While ESAs and iron would be expected to have direct effects on the erythron, agents that address systemic dysfunction, such as the mild pro-inflammatory state of ageing, might also improve erythropoiesis. Salsalate, a non-acetylated salicylate that is a potent inhibitor of NFκB, has been used successfully to treat low grade inflammation and glycaemic response in obese subjects with dysglycaemia (Fleischman et al, 2008). The proven safety profile of this particular intervention, coupled with its success in an ageing-related disease population makes it a worthy candidate for trials in anaemic older adults.

Similarly, statins are safe and widely used to manage hyperlipidaemia. Recently, rosuvastatin was shown to reduce CRP and cardiovascular events in adults without hyperlipidaemia (Ridker et al, 2008). The anti-inflammatory properties and broad use of statins suggests they might be effective against other, inflammation-related pathologies. Importantly, observational trials have demonstrated myalgia and myopathy may occur in 5–10% of individuals using statins (Bruckert et al, 2005; Nichols & Koro, 2007) suggesting this particular intervention may not be tolerated by some older adults.

Finally, lenalidomide has been used successfully to prevent transfusions in older adults with low risk MDS. As a possible regulator of NFκB, this immunomodulatory drug (and the new drugs in its class) may also be an appropriate intervention for clinical trials directed at subsets of anaemic older adults (Zeldis et al, 2011).

Towards a better diagnosis – eliminating ‘unexplained’ anaemia

Clinical trials and additional prospective studies are also needed to facilitate collection of biological samples for new translational research opportunities. State of the art methods are available to test very specific hypotheses related to anaemia in older adults. High sensitivity enzyme-linked immunosorbent assays and core facilities that can perform them are available for biomarker development in the serum and plasma of appropriate anaemic cohorts and their controls. Primary cell culture expansion strategies and flow cytometry techniques using well-defined markers of human stem cell and erythroid progenitors are also available for analysis of bone marrow aspirates. Further, the measurement of telomere length in leucocytes and bone marrow cells using flow cytometry-based assays is now commercially available and might reveal a subset of patients in whom short telomeres play a role in anaemia. Finally, high resolution genomic approaches demonstrated that even patients with low risk MDS have numerous genetic changes which predict clinical outcome (Starczynowski et al, 2008). We hypothesize that such genomic approaches might have utility in discriminating subclinical MDS from UAE, allowing early intervention that could alter the course of their disease.


The diagnosis of anaemia in community-dwelling older adults is often complicated by the presence of multiple comorbid conditions, but should be adequately addressed in light of the significant association of anaemia, morbidity and mortality in this population. Diagnosis requires persistence, yet a significant proportion of cases of anaemia in older adults remains unexplained. There are few evidence-based approaches to its treatment, if unexplained. The international community of haematologists needs to address this problem with better guidelines regarding healthy haemoglobin targets for diverse groups of older adults; with high-quality interventional trials aimed to improve physical performance in anaemic older adults; and by addressing gaps in our knowledge of the molecular pathogenesis of anaemia in the elderly with the most advanced methods available.


AAM and CNR wrote the paper. AAM was supported for this work by a Clinician Scientist Award from the Johns Hopkins University School of Medicine. CNR was supported for this work by the American Society for Hematology Scholars Award, and NIH grant DK082722. The authors would also like to thank the members of the Partnership for Anaemia Clinical and Translational Trials in the Elderly (PACTTE) for their leadership and thoughtful discussions in regard to the problem of unexplained anaemia.

Conflicts of interest

Dr Roy maintains a sponsored research agreement with Celgene Corporation which funds a pilot study concerning the anaemia of inflammation.