Mortality was lowest at hemoglobin concentrations 1 to 3 g/dL above current WHO thresholds, with higher mortality below and above this range, suggesting the current WHO anemia thresholds may not optimally identify individuals at lowest mortality risk.
All-cause mortality showed a U-shaped association with hemoglobin concentration, with the lowest mortality risk occurring 1 to 3 g/dL above the WHO anemia threshold.
Participants with hemoglobin more than 2 g/dL below the WHO threshold had markedly elevated all-cause mortality compared to the reference group.
Even participants with hemoglobin 0 to 1 g/dL above the WHO threshold (borderline-low but not classified as anemic) had elevated all-cause mortality compared to the reference group.
Associations between hemoglobin and mortality were less consistent in women younger than 60 years.
Cardiovascular and cancer mortality showed patterns similar to all-cause mortality at low and borderline hemoglobin concentrations, but the associations attenuated at higher hemoglobin concentrations.
The authors caution that these hypothesis-generating findings should not be used to define new hemoglobin thresholds but may inform clinical interpretation of borderline hemoglobin values.
This research suggests that the long-standing World Health Organization (WHO) definition of anemia—based on hemoglobin thresholds set in 1959—may not perfectly identify the hemoglobin levels associated with the lowest risk of death. In a large UK study of nearly 478,000 adults followed for an average of about 14 years, people whose hemoglobin was 1 to 3 points above the WHO cutoff had the lowest death rates. People below the WHO threshold had substantially higher death rates, and even those just barely above it (within 1 point) had modestly elevated risk compared to those in the optimal range. The relationship between hemoglobin and mortality followed a U-shape: both very low and very high hemoglobin levels were associated with greater mortality than intermediate levels. This pattern held for deaths from cardiovascular disease and cancer as well, though the elevated risk at higher hemoglobin levels was less pronounced for those specific causes. The association was somewhat less clear-cut in women under age 60, possibly because factors like menstruation influence hemoglobin levels in ways that complicate interpretation. This research suggests that clinicians may want to pay closer attention to patients whose hemoglobin is near but not below the current anemia cutoff, as these individuals also appear to face elevated mortality risk. However, the authors are careful to note that this single study—with its limitations, including only one hemoglobin measurement per person and a mostly White population—should not by itself prompt changes to diagnostic thresholds. Rather, it raises important questions that future research should investigate more thoroughly.
Malek Ahmad, John G. F. Cleland, P. Maffia, R. Strawbridge, Naveed Sattar, Jill P. Pell, et al.. (2026). Association Between Blood Hemoglobin Concentration and Mortality by Age and Sex : A Prospective Cohort Study.. Annals of Internal Medicine. https://doi.org/10.7326/ANNALS-26-00057