The sensitivity and specificity of the Goldberg cut-off for EI:BMR for identifying diet reports of poor validity

To explore the specificity and sensitivity of the Goldberg cut-off for EI:BMR for identifying diet reports of poor validity as compared with the direct comparison of energy intake with energy expenditure measured by doubly-labelled water.

Twenty-two studies with measurements of total energy expenditure by doubly-labelled water (EE), basal metabolic rate (BMR) and energy intake (EI) provided the database (n=429). The ratio EI:EE provided the baseline definition of under- (UR), acceptable- (AR) and over-reporters (OR), respectively EI:EE <0.76, 0.76-1.24 and >1.24. Four strategies for identifying under- and over-reporters using the Goldberg cut-off were explored. Sensitivity of the cut-off was calculated as the proportion of UR correctly identified and specificity as the proportion of non-UR correctly identified.

UR, AR and OR (by EI:EE) were 34, 62 and 4% respectively of all subjects. When a single Goldberg cut-off for the physical activity level (PAL) of 1.55 was used, for men and women respectively the sensitivity was 0.50 and 0.52 and the specificity 1. 00 and 0.99. Using a cut-off for higher PAL traded specificity for sensitivity. Using the cut-off for a PAL of 1.95, sensitivity was 0. 76 and 0.85 and the specificity 0.87 and 0.78 for men and women respectively. Using cut-offs for mean age-sex specific PAL did not improve sensitivity. When subjects were assigned to low, medium and high activity levels and cut-offs for three different PALs used, sensitivity improved to 0.74 and 0.67 without loss of specificity (0. 97 and 0.98), for men and women respectively. If activity levels for men were applied to the womens' data, sensitivity improved to 0.72.

To identify diet reports of poor validity using the Goldberg cut-off for EI:BMR, information is needed on each subject's activity level.