To develop a method to make use of incomplete 24-hour urinary samples in nutritional epidemiology, especially when validating the dietary intake of nitrogen (protein), sodium and potassium.

Urinary data for men and women collected in three different studies were evaluated. The concentration of para-aminobenzoic acid (PABA) in one 24-hour urine sample per person was compared with the concentrations of nitrogen, sodium, potassium and creatinine.

Men and women living in Cambridge, UK and women living in the town of Varberg, Sweden.

In total, this study consists of data from 73 Swedish women (20–50 years of age), 165 UK women (50–65 years) and 75 UK men (55–88 years).

On average four out of 10 people in this study had a PABA recovery below 85%. The linear regression equations for urinary excretion of nitrogen, sodium and potassium in relation to PABA recovery were y = 2.3 + 0.088 * x (r = 0.99), y = 45 + 0.82 * x (r = 0.87) and y = 19 + 0.60 * x (r = 0.93), respectively.

The linear regression equations can be used for adjusting urinary nitrogen, sodium and potassium in urinary collections in cases where the PABA recovery is below 85%. Since it is common to obtain 24-hour urine collections with a PABA recovery below 85%, this method should increase the usefulness of biological markers of food intake in nutritional epidemiological studies and also increase the possibilities to study people that previously have been part of the drop-out group or the group with low motivation and cooperation. It is important to stress that we have not studied the relationship between PABA recovery and various urinary variables below the PABA recovery of 50%. Thus, in a case of PABA recovery below 50%, we do not recommend the use of this method to compensate for incomplete collections.