Biological variation and diagnostic accuracy of dehydration assessment markers.

BACKGROUND Well-recognized markers for static (one time) or dynamic (monitoring over time) dehydration assessment have not been rigorously tested for their usefulness in clinical, military, and sports medicine communities. OBJECTIVE This study evaluated the components of biological variation and the accuracy of potential markers in plasma, urine, saliva, and body mass (B(m)) for static and dynamic dehydration assessment. DESIGN We studied 18 healthy volunteers (13 men and 5 women) while carefully controlling hydration and numerous preanalytic factors. Biological variation was determined over 3 consecutive days by using published methods. Atypical values based on statistical deviations from a homeostatic set point were examined. Measured deviations in body fluid were produced by using a separate, prospective dehydration experiment and evaluated by receiver operating characteristic (ROC) analysis to quantify diagnostic accuracy. RESULTS All dehydration markers displayed substantial individuality and one-half of the dehydration markers displayed marked heterogeneity of intraindividual variation. Decision levels for all dehydration markers were within one SD of the ROC criterion values, and most levels were nearly identical to the prospective group means after volunteers were dehydrated by 1.8-7.0% of B(m). However, only plasma osmolality (P(osm)) showed statistical promise for use in the static dehydration assessment. A diagnostic decision level of 301 plusmn 5 mmol/kg was proposed. Reference change values of 9 mmol/kg (P(osm)), 0.010 [urine specific gravity (U(sg))], and 2.5% change in B(m) were also statistically valid for dynamic dehydration assessment at the 95% probability level. CONCLUSIONS P(osm) is the only useful marker for static dehydration assessment. P(osm), U(sg), and B(m) are valid markers in the setting of dynamic dehydration assessment.