Key results
The present study demonstrates relationships between urinary biomarkers of hydration in spot samples and in 24 h collections of urine, as well as curvilinear relationships between spot samples and water intake (Fig. 2). The expected urine concentrations of the biomarkers in spot samples for increasing amounts of water consumptions are also reported (Table 2). However, the focus is placed on calculating the sensitivity the specificity for ranges of biomarkers to quantify the water intake, which is relevant information when using urine biomarkers in population studies.
The results show that morning urine and spot urine samples had almost the same ability to discriminate between ranges of daily water intakes, albeit that spot samples were slightly poorer. The osmolality and the creatinine concentration both had a reasonably good capacity to indicate large water intakes (> 3–4 L). Small volumes (< 1.5–2 L) were indicated by highly concentrated urine, with only a few misses, but the sensitivity was hampered due to many “false alarms.” The usefulness of these biomarkers was reduced when the subject was in the process of increasing the daily intake of water. One should note that the 24-h urine collection and the spot samples predicted the water intake during the same day while the morning urine summarized the water consumption for the preceding 24 h.
Urine analysis
Urine sampling for measurement of metabolic waste products has gained recognition as a tool for monitoring body hydration in sports medicine [2–4], whereas its use in other settings, such as in population studies and geriatric hospital care, is rare [11, 12]. Uncertainty about the accuracy of urine sampling in indicating water ingestion in individuals rather than groups has certainly contributed to its limited applicability. The complexity of the issue has been further increased by different modes of sampling.
The present study offers orientating guidance on these issues. The reference data consist of careful registration of all water in liquid and food consumed during a two-week period by healthy hospital workers. The ability of two urine biomarkers — osmolality and creatinine — to indicate the water consumption was tested during periods of both normal and increased fluid intake. Several modes of sampling were used, and ROC curves were employed to study the sensitivity and specificity of each mode to detect the water ingestion in an individual.
ROC curves
The first step was to determine how well measurements of biomarkers in a 24 h collection of urine could predict the urine volume. Both variables were known, and this comparison then served as a reference for methodological errors as well as between-subject and between-measurement variability. The AUC of the ROC curves then averaged 0.88.
The AUC decreased to 0.77 when the same biomarkers were used to discriminate between pre-set arbitrary ranges of water intake; the difference may be due to variable insensible fluid losses, different degrees of fluid retention, and inaccuracies in the quantification of water intake.
Morning urine showed only slightly poorer discriminating ability, but the AUC for osmolality as the biomarker for the 24 h water ingestion still averaged 0.72 (Fig. 1). Surprisingly, the AUC for spot samples, which were collected during the last 3 days of each week, showed the same AUC as the morning samples, with an average of 0.74 for the osmolality (Fig. 3).
Habitual or increased fluid intake
It is now well established that the urine osmolality and the urinary creatinine concentration vary with the habitual intake of water in healthy individuals who do not exercise [6, 7, 10]. A methodological problem is that an increase in ingested volume of 30% was not enough to significantly increase the dilution of the morning urine [10]. The slowness of the kidneys to adapt to such a diet change is reflected by that low biomarker levels most reliably indicated large intakes of water during the first week, while high levels gave a more accurate indication during the second week.
Acute ingestion of water as well as exercise-induced dehydration make the kidneys adapt within hours [13–15] but quite large increases of the water consumption seem to be needed (> 1.5 L per day) [16, 17]. Thus, renal adaptation from conserving to excreting water occurs faster than indicated by the morning urine. For example, intravenous volume loading with 1.5 L of electrolyte-containing fluid over 30 min in elderly males showed that the initial setting of the kidneys to excrete or retain water strongly influences the excreted volume during the first hour, and that some effect remains for at least three hours [18].
Literature
Urine analysis to indicate the water balance in male athletes was initiated in the early 1990s by Armstrong et al. [13]. The focus was on the visual estimation of urine color as compared to laboratory analyses of urine-specific gravity and urine osmolality [3].
Further studies in sports medicine focused on progressive dehydration and included plasma osmolality [3, 4, 14]. Cheuvront et al. found a sensitivity of 90% for urine osmolality and for urine-specific gravity to detect dehydration when defined as a rise in plasma osmolality from baseline [15].
Urine analysis was later extended to a wide age span and both genders during recreational sports and showed that a loss of 1% of the body weight was detectable [19]. Attention was also given to the variability in urine concentration in the population [4] and to concentrated urine as a risk factor for a poor outcome in geriatric care [11, 20, 21].
Attempts have been made to identify a fixed time of the day when measurement of a biomarker in a single urine sample could yield the same result as a 24-h collection. Two studies report that urine osmolality measured in the early or late afternoon correlate best with the osmolality of 24-h collections [7, 8]. A similar evaluation was performed in the present study but designed to show how robust spot samples are to reflect the 24-h excretion of biomarkers during four days in an individual volunteer. The results confirm that afternoon samples reflect the 24-h collection better than morning samples, at least during the period of increased water consumption (Table 3). However, the scatter of the data was still considerable, making it unclear how reliable a single spot sample is to indicate the excretion of biomarker in 24-h urine collections during a four-day period. On the other hand, a spot sample taken in the afternoon seems to be a useful indicator of hydration if larger populations are compared.
The accuracy of spot samples may be more clearly affected by the time after intake of water and food. Liberal water consumption during a meal induces acute diuresis, whereby the biomarkers in the urine become temporarily diluted. A previous report showed that spot samples were 20–30% more diluted than morning samples for 7 h after intake of food or water, while after 7 h they were 5–10% less diluted. By contrast, 24 h collections of urine for biomarker analysis yielded quite similar mean values to those of the spot samples during the first 7 h, while the spot samples taken later showed higher concentrations [10].
Dehydration
The kidneys concentrate the urine over a very wide range [5]. This is the normal adaptation to variations in the daily consumption of water and does not implicate dehydration, as the body fluid volumes are well maintained [10]. Several studies from sports medicine show that a change in urine biomarkers from a known baseline can diagnose acute exercise-induced dehydration [15]. However, relying on a single sample for this purpose in a mixed population is futile, since the inter-individual variation in normal water ingestion is great; the reference point is unknown because the kidneys may have adapted well to a chronically low intake of water. This issue has caused distrust in the ability of urine analysis to diagnose chronic dehydration and low fluid intake [12, 22].
Biomarkers measured in the plasma, such as creatinine and osmolality, may be used to indicate dehydration, but changes develop late and only when the renal water conservation mechanisms are insufficient. Blood sampling also lacks the simplicity of the urine analysis. Hyperosmolality is often used as a sign of “intracellular” dehydration [14, 15], and this suggests that the kidneys are too slow to cope with sudden losses of water. In chronic dehydration, hyperosmolality develops when the capacity to concentrate the urine is exhausted, and this can be due to a low intake of water but also to age-related impairment of the renal concentrating capacity. Therefore, no correlation has been found between serum osmolality and single measurements of urine biomarkers of hydration in the elderly [12, 22, 23]. Chronic dehydration remains a challenging diagnosis that still needs be based on a combination of medical history, clinical signs, and measurements. None of the volunteers in the current presentation was considered dehydrated, based on the hemodynamic measurements, body fluid volumes as measured with bioimpedance, and the fact that they were full-time active hospital workers.
Visual estimation followed by manual recording of the water consumption is a commonly used alternative to urine analysis for monitoring of the fluid intake in sick patients. This approach requires much attention and involves considerable uncertainties. Ingestion of fluid is sometimes simply missed. Moreover, patients visually underestimate, and healthy volunteers commonly overestimate, the amount of fluid that is contained in a glass, while nurses make the best estimates [24]. Lack of knowledge about how to judge the water content of solid food further reduces the value of visual estimates of fluid intake.
Tucker et al. correlated the void frequency with the degree of hydration in male volunteers [25]. They voided 7 (2) times per 24 h during habitual fluid intake (mean, 2.4 L) and 5 (2) times when being hypohydrated (intake 1 L per 24 h). The void frequency was considered a reliable indicator of hydration status despite the apparent overlapping indicated by the two standard deviations. In the present study, the void frequency increased stepwise with the fluid intake up to 3 L per 24 h (Table 2). However, counting the void frequency requires monitoring over 24 h and the idea of relying on a single measurement of a biomarker is lost.
Practical application
No urine analysis provides precise information about the 24 h fluid intake. Any judgment of what represents “good” and “poor” sensitivity and specificity in this setting is up to the reader. The following conclusions represent the views of the author.
High water intakes can be indicated in both morning and spot samples that show osmolality and creatinine in the lowest range (50–400 mosmol/kg and 2–6 mmol/L, respectively). False indications are rare.
The most important information is whether a subject ingests normal (2–3 L) or small (< 1.5–2 L) amounts of water. Urine osmolality in the highest range indicates low fluid intakes with few misses, but false indications (high osmolality but large fluid intake) are common. Therefore, diagnosis should be corroborated by other measurements.
The present results suggest that an investigator should ascertain that a marked change in the fluid intake has not occurred during at least one week preceding the urine sampling.
Strengths and limitations
The strengths of the study include that that all water intake was recorded, i.e., even the water that was contained in food. The volunteers were hospital workers, usually nurses, with experience in following a protocol and reporting. Collection of data was performed in a standardized way. Biochemical analysis was performed within 36 h, even on weekends, by an accredited hospital laboratory. Medical checkups were performed on three occasions during the 2-week study, ensuring hemodynamic stability and that no signs of dehydration or disease appeared [10].
Limitations include a small number of subjects, although each volunteer contributed with 8 morning urine samples that were matched with the same number of 24 h measurements of water intake. The number of spot samples was 10 times higher, which means that several spot samples were associated with the same calculation of the 24 h water intake. However, different data were used for the morning urine and spot sample analyses, which complicates direct comparisons.
As evident from Table 2, the distributions of spot sample data across ranges of water intakes were not balanced. For example, only 4 volunteers contributed the 27 samples associated with fluid intakes < 1.5 L, all of which occurred during normal fluid intake.
Four arbitrary ranges of biomarkers were used. Only three are shown in the figures, but the fourth (osmolality 600–800 mosmol/kg and urine creatinine 9–15 mmol/L) can be inferred from the others.
The indications are not valid in association with strenuous sport activities or profuse sweating.