Optimal Hydration Biomarkers: Consideration of Daily Activities

Water intake and hydration status have recently gained attention as one of the many, and potentially manipulable, factors associated with disease development and wellbeing [1,2,3,4,5,6,7]. Researchers have explored a variety of physiological dysfunctions implicated with chronic low water intake and hypohydration, but particular interest resides in the study of obesity and related disease states due to overwhelming prevalence. However, the current lack of a hydration assessment gold standard greatly impedes our attempts to link water intake and negative health outcomes as well as to make public dietary guidelines.

Current hydration biomarkers include blood, saliva, and urine sampling subjected to a variety of quantification methods (osmolality, specific gravity, volume, etc.). Recent work from our laboratory [8] contributed to the ongoing quest to identify optimal selection of hydration biomarkers, but through consideration of the context under which body water loss occurs. We present two scenarios when selecting optimal hydration biomarkers: i) active versus passive body water loss, and ii) single versus serial measurements.

We examined 23 healthy, recreationally active males (age = 22 ± 3 years; mass = 77.3 ± 12.8 kg; height = 179.9 ± 8.8 cm; body fat = 10.6 ± 4.5%). Participants completed one familiarization trial, recorded their complete dietary intake for 3 consecutive days, and underwent two experimental trials. Individual water intake was prescribed for the day prior to experimental trials as an average intake from their 3-day dietary records. Furthermore, participants replicated all food consumption the day prior to experimental trials.

Two experimental trials were conducted in a hot environment (about 36° C, approximately 50% relative humidity) to examine: i) passive dehydration (PAS), where participants simply sat in the hot environment for 5 h, and ii) active dehydration (ACT), where participants cycled in the hot environment at a moderate intensity (about 68% of age-predicted maximal heart rate) for 5 h. Participants did not consume foods or water during these trials. Upon each percent body mass loss, participants provided blood, saliva, and urine samples. Researchers measured blood serum osmolality (S_osm), salivary osmolality (V_osm), urine osmolality (U_osm), volume (U_vol), and specific gravity (U_sg).

We uncovered that the context under which body water losses occur does influence the ability of hydration biomarkers to assess dehydration. Largely, we found during PAS that progressive and large changes were observed more so in urinary compared to serum and salivary variables. Conversely, we found during ACT that serum and salivary markers responded more precisely to body water losses than did urinary variables (fig. 1). These findings are likely explained through autonomic regulation and hemoconcentration during heat and exercise exposure [9], which concentrates the extracellular compartment, primary saliva [10], and diverts cardiac output away from the kidneys [11,12,13]. In support of these findings, we were able to compare mean differences between PAS and ACT trials at 1% body mass loss, and found that even mild dehydration elicited statistically and physiologically meaningful differences between the two methods of body water loss (table 1). In essence, these findings indicate that exercise exaggerated changes in S_osm and V_osm, and that urinary markers best represented water losses during PAS.

Statistical tools for measuring diagnostic accuracy and reference change values (RCV) permitted evaluation of hydration biomarker utility with availability of single versus serial measurements, respectively.

In order to evaluate hydration biomarker utility with a single measurement, we operationally defined a dehydrated state as 2% body mass loss, as this degree of body water loss typically provokes physiological and psychological dysfunction [5,14,15,16,17]. Diagnostic accuracy statistics included: i) receiver operating characteristic (ROC) to define criterion values for each diagnostic test, ii) area under the ROC curve (AUC) to summarize diagnostic accuracy, and iii) sensitivity, specificity, and positive likelihood ratio (+LR) to assist in clinical decision making. The best combination of sensitivity (the probability of a true positive) and specificity (the probability of a true negative) allows the practitioner to use the ROC-derived criterion value with confidence to diagnose dehydration with the respective hydration biomarker. Great practicality exists in the +LR, as high ratios indicate that true positives are much more likely than false positives.

Diagnostic accuracy was conducted for ACT only, as PAS resulted in an insufficient sample size greater than 2% body mass loss. We found that V_osm and S_osm most accurately assessed dehydration (table 2), which supports our observation that these variables responded more drastically to body water loss than urinary variables during ACT. The +LR for S_osm (table 3) tells the practitioner that, e.g., a value of 303 mOsm·kg^-1 occurs 4.88 times more often in a dehydrated than sufficiently hydrated individual who has been exercising.

RCV statistics evaluate the utility of a test with serial measurement. This requires preliminary measures of analytical and biological variation, which includes coefficients of variation for the analytical device (CV_A), and interindividual (CV_I) and intraindividual (CV_G) variation. The derived RCV (table 4) in this investigation indicates a meaningful change (expressed as a percentage or unit) in the hydration biomarker indicative of a dehydrated state. Subsequent analyses included the index of individuality (II) which indicates usefulness for population-based reference intervals, where a value greater than 1.4 signifies high usefulness and less than 0.6 suggests low usefulness. The index of heterogeneity (IH) assists in validating the RCV, which, in our investigation, was an IH greater than 2.

Due to the nature of RCV statistics, the values reported are irrespective of the method through which body water loss occurred (active vs. passive). We found that only three hydration biomarkers demonstrated valid RCV (IH > 2): S_osm, U_sg, and body mass loss. For example, a change in serial U_sg measurement of 0.007 would indicate a dehydrated state according to our findings (table 4). Furthermore, only U_vol had a high enough II considered useful for population-based reference intervals, while it did not result in a valid RCV. This implies good utility for interpretation of hydration state among healthy populations based on typical variation, but not for serial measurement with acute dehydration, respectively.

This investigation elucidated the need to consider previous or concurrent activities, and the frequency of sample measurement when selecting hydration biomarkers. Largely, we found that S_osm and V_osm during ACT in the heat, and urinary markers during PAS in the heat, were the most suitable biomarkers given the context or means of dehydration. When only single measurements are available during ACT, S_osm and V_osm established the greatest efficacy. Regardless of the method of body water loss, S_osm, U_sg, and body mass loss resulted as the most valuable serial measurement dehydration indices. In essence, the context surrounding body water loss should dictate selection of hydration biomarkers to appropriately identify hydration status before research examining the influence of chronic water intake on health outcomes progresses.

The authors declared no conflict of interest.