Undernutrition is often suspected in patients when serum albumin or prealbumin levels are low. We asked whether these measures are indeed low in undernourished people if no inflammatory illness is present.

We did a systematic review to identify otherwise healthy subjects who were severely nutrient-deprived due to poor access to food or unwillingness to eat. We excluded children and pregnant women. We tabulated available measures of nutrient intake, anthropometry, serum albumin and prealbumin, and, when available, changes in these measures during nutritional intervention.

In otherwise healthy subjects, serum albumin and prealbumin levels remained normal despite marked nutrient deprivation until the extremes of starvation, that is, body mass index <12 or more than 6 weeks of starvation.

In these otherwise healthy subjects, serum albumin and prealbumin levels are not “markers of nutritional status.” The “markers” failed to identify subjects with severe protein-calorie malnutrition until extreme starvation. That is, they failed to identify healthy individuals who would benefit from nutrition support, becoming abnormal only when starvation was already obvious. In contrast, serum albumin and prealbumin levels are known to fall promptly with injury or illness regardless of nutrient intake. They are negative acute-phase reactants. When these measures are low in sick patients, this cannot be assumed to reflect nutritional deprivation. Decisions about nutrition support should be based on evidence of meaningful benefit from this treatment rather than on assessment of “nutritional markers.”