Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Anaemia and Mortality in Older People


Bandolier 137 examined the increasing prevalence of anaemia with age six in larger studies, using WHO criteria for anaemia of a haemoglobin of less than 120 g/L for women and less than 130 g/L for men. Anaemia has many causes in older people. The question remains about the consequences. A large cohort study [1] indicates that anaemia may be independently associated with increased mortality.


This was a prospective observational study of risk factors for, and consequences of, cardiovascular disease in older adults in the community. Beginning in about 1990, it recruited people aged 65 years or older in four communities in the USA. Exclusion criteria were being wheelchair bound, treatment for cancer, or living in an institution. At enrolment there was a physical examination and history, together with various tests.

Deaths were identified from databases, and by bi-annual follow up, up to mid-2001. Analysis was by quintile of haemoglobin level (performed separately for men and women), and by use of the WHO criteria for anaemia.


Baseline haemoglobin measurements were available of 5,800 participants, average age 74 years at baseline, with a median follow up of 11.2 years, and 54,000 person years of follow up. Using WHO criteria, 498 were anaemic initially, a prevalence of 8.5%. Anaemia was commoner in black participants (18%) than white (7.0%), but similar in men and women.

Participants in the lowest quintile of haemoglobin were older, more likely to be black, and had more comorbid conditions. The strongest correlates were with low BMI, low activity levels, fair or poor self reported health, frailty, heart failure, stroke, or transient ischaemic attack. Low haemoglobin was also associated with higher creatinine levels, CRP, and fibrinogen, and lower serum albumin and white cell count.

Low haemoglobin was associated with higher mortality (Figure 1), whether using the WHO criteria, or the lowest quintile. Death rates over the 11 years were 57% for those with WHO-defined anaemia but 39% for those without anaemia, with high levels of statistical significance, and for both cardiovascular and non-cardiovascular causes of death.

Figure 1: Mortality in older people with WHO criteria of anaemia, and by quintile of Hb

Compared with the fourth quintile (women with haemoglobin levels of 139 to 144 g/L and men with levels of 151 to 156 g/L), significant increases in mortality were seen for both the lowest and the highest quintiles of haemoglobin, using adjustments for a wide range of possibly confounding characteristics (Table 1).

Table 1: Statistical comparisons

Hazard ratio
(95% CI)
WHO criteria
1.4 (1.2 to 1.6)
Lowest Hb quintile Q1
1.3 (1.2 to 1.5)
1.2 (0.99 to 1.3)
1.0 (0.9 to 1.2)
Highest Hb quintile Q5
1.2 (1.01 to 1.4)
Hazard ratios using adjustment for age, sex, race, and baseline characteristics


The prevalence of anaemia was 8.5%, within the prevalence range seen previously in larger studies (Bandolier 137). This study clearly associates lower haemoglobin levels with increased mortality in a reasonably large population over a long time. The association with higher mortality was particularly strong using the WHO criteria for anaemia for women and men. There may be increased mortality at higher levels of haemoglobin, but that was much less strong. Other studies have found similar association between anaemia and higher mortality in older people, and in specific conditions. This paper is a rich source of references.

There is much we do not know, of course. We do not know the cause of anaemia in these people, and no observational study can determine causality. We can speculate that long-term anaemia can contribute to adverse physiological changes. We can only guess at the moment whether treating the anaemia would be beneficial in older people, though it is in specific conditions.


  1. NA Zakai et al. A prospective study of anemia status, hemoglobin concentration, and mortality in an elderly cohort. Archives of Internal Medicine 2005 165: 2214-2220.

previous or next story