Sensitivity and specificity of published strategies using urinary creatinine to identify incomplete 24-h urine collection

Objective Although urinary creatinine has been used to identify incomplete 24-h urine in numerous epidemiologic studies, information on its utility is limited. We examined the sensitivity and specificity of several strategies that use creatinine to identify incomplete urine using the p-aminobenzoic acid (PABA) check method as reference. Methods Subjects were 654 female Japanese dietetic students 18–22 y of age. A single 24-h urine sample was collected, with recording of the time of the start and end of the collection period and missing urine volume. Simultaneous administration of PABA was done to assess completeness. The sensitivity and specificity of five strategies derived from the literature that used creatinine to identify incomplete urine were calculated as the proportion of incomplete and complete urine correctly identified, respectively. Results A total of 7.6% of subjects was identified as having incomplete urine by PABA (PABA recovery <85%). This proportion significantly (P < 0.0001) decreased (to 5.5%) after considering self-reported collection time and missing urine volume in the calculation of total urine volume. The sensitivity and specificity of the strategy of Knuimann et al. (incomplete urine = <0.7 of [mmol urinary creatinine × 113]/[21 × kilograms of body weight]) were 0.47 and 0.99, respectively. The corresponding values of other strategies were 0.11–0.22 and 0.57–1.00, respectively. Conclusion At least in well-motivated populations in which the proportion of incomplete urine is presumed to be small, the strategy of Knuimann et al. and consideration of the self-reported collection time and missing urine volume in the estimation of total volume may be useful.