Excessive fluid intake as a novel cause of proteinuria

We are currently conducting a longitudinal population cohort study in the community of Walkerton, Ontario, to determine health outcomes after the municipal water supply was contaminated with Escherichia coli O157:H7 and Campylobacter in 2000.1 Our study cohort of 4496 people represents about 50% of the community, and their demographic characteristics are similar to those of the whole community.2

As part of our initial assessment in 2002, we measured urine protein levels in 24-hour collections of urine from 2253 adults who later attended a follow-up clinic annually between 2003 and 2005. We measured the protein levels with a Vitros 950 autoanalyzer (Ortho-Clinical Diagnostics Inc., Rochester, New York) using the pyrocatechol dye procedure,3 with a coefficient of variation for the low control below 3.5% and a lower limit of sensitivity of 0.05 g/L. Of the 2253 people screened, 1861 had urine creatinine levels within the sex-adjusted reference limits.4^,5 They comprised 1199 who had neither proteinuria nor polyuria, 445 who had proteinuria without polyuria, 4 who had polyuria without proteinuria and 213 who had proteinuria with polyuria. Unexpectedly, in this last group, who had proteinuria (protein excretion > 0.20 g in 24 hours), polyuria (urine volume > 2.9 L in 24 hours) and seemingly normal kidney function (creatinine clearance rate > 60 mL/min per 1.73 m²),4 we found 100 participants (4.4% of the community sample) who had no explanation for their proteinuria. These 100 people had no evidence of kidney disease, diabetes mellitus, hemolytic uremic syndrome or psychogenic polydipsia by self-report or medical chart audit. None had received lithium or diuretics,6 and only 28 had hypertension.

The 100 participants had urine protein concentrations from 0.07 to 0.15 g/L, well above our lower limit of detection. In the 24-hour collections of urine, the mean amount of protein excreted was 0.43 g (standard deviation [SD] 0.21 g, 95% confidence interval [CI] 0.36–0.49 g), and the mean volume of urine was 3.7 L (SD 1.2 L, 95% CI 3.3–4.1 L). This group was similar to the overall cohort of 2253 adults in terms of age (mean 47 [range 15–72] and 47 [range 15–92] years, respectively), sex (68% and 61% women), history of hypertension (28.0% and 35.2%) and gastroenteritis (61.0% and 60.4%).

We attempted to discern the cause of the unexplained proteinuria and polyuria. Of the 100 people, 63 agreed to confirmatory 24-hour urine collection to measure protein excretion, followed by a urine osmolality measurement after overnight water deprivation. The 24-hour confirmatory urine samples had a mean protein content of 0.40 g (SD 0.29 g, 95% CI 0.30–0.49 g) and a mean volume of 3.6 L (SD 1.8 L, 95% CI 3.0–4.2 L). After the overnight water-deprivation test, we found that 39 of the 63 participants had a urine osmolality of more than 450 mOsm/kg. Of the 24 whose urine was not concentrated to this level, 17 returned after another voluntary overnight water-deprivation test to undergo 8 hours of observed water deprivation. By the end of the 8 hours, we found that the urine was concentrated to a level greater than 450 mOsm/kg in all 17 cases (Table 1), which meant that diabetes insipidus was unlikely. The baseline characteristics did not differ significantly between the 56 people who participated in the complete testing (39 who required only 1 test plus 17 who completed 2 tests) and the 44 people who did not participate in the complete testing.

The 56 participants were then asked to voluntarily reduce their total daily fluid intake to fewer than 8 large glasses (< 2.0 L/d) for 1 week, on the last day of which they provided another 24-hour urine sample. The mean urine volume following the voluntary water-reduction intervention decreased to 1.81 L in 24 hours (95% CI 1.54–2.07 L), and the amount of protein excreted decreased to a mean of 0.16 g in 24 hours (95% CI 0.12–0.20 g) (Table 2). The mean urine creatinine level did not change significantly from the baseline level before or after the intervention (Table 2). The relation between urine volume and protein excretion was similar for the 56 participants regardless of whether they had had symptoms of gastroenteritis at the time of the water contamination in 2000.

When asked about their fluid consumption, the 56 participants indicated that they were drinking large volumes of fluid because they perceived it to be a healthy lifestyle choice. The water contamination and resultant temporary switch to bottled water may have led some to assume a habit of overconsumption. However, most said that their overconsumption preceded the water contamination. Although our participants did not fit the classic description of the compulsive water drinker, their attenuated concentration response did.7

Our unexpected finding of a 4.4% rate of proteinuria associated with polyuria in seemingly healthy people stemmed from observations that had been made during a community-based screening program. The reversible nature of the proteinuria and polyuria after reduction of the total fluid intake to less than 2 L/d makes any explanation other than voluntary excessive fluid intake unlikely. In a study of the effect of water loading on urinary albumin excretion in 18 healthy volunteers, Viberti and colleagues8 noted that water loading was associated with a short-lived but significant increase in urine albumin levels. We did not characterize the proteinuria, nor do we know at this time whether the reversible proteinuria associated with large fluid intake in these otherwise healthy people could affect their kidney function in the long term.9^–11 If the proteinuria was largely tubular in origin, owing to tubular washout, rather than glomerular in origin, one might presume that progressive kidney injury would be less likely.9 Until such data are available from our longitudinal study, it may be advisable to discourage otherwise healthy people from consuming large volumes of fluid.