The association between urine osmolarity and the risk of initiating dialysis, a stronger marker for end stage renal disease, has not been previously investigated. Our data thus adds to the existing evidence. The data is in line with Torres et al. suggesting a faster renal function decline in patients with higher urine osmolarity. Our cohort appeared to have the highest urine osmolarity with a median of 510 mosm/L versus a mean of 3686159 mosm/L, and a mean of 270 to 334 mosm/L. It is therefore conceivable that lowering urine osmolarity below a specific threshold might be harmful to the kidney as well. A recent study addressing the association between sodium excretion, an important part of total urine osmolarity, and end-stage renal disease in patients with type 1 diabetes mellitus showed an inverse association with end stage renal disease. Furthermore, a U-shaped curve was described for sodium excretion and mortality, such that subjects with the highest and the lowest sodium excretion had the highest mortality, partly supporting the hypothesis of low urine osmolarity being associated with harmful effects. Interestingly, in the univariate versus the Gomisin-D multivariable analysis the strong effect of impaired renal function seems to obscure the positive correlation of osmolarity with the incidence of ESRD. Adjusting for creatinine clearance reveals this relationship. We conclude that among patients with equal renal function those with higher osmolarity values are more likely to progress to ESRD than those with lower osmolarity values in 
our cohort. As described in the methods, Danshensu estimated urine osmolarity was calculated from urine sodium, potassium and urea concentrations in 24 hour urine samples. These are the dominant solutes in the urine. Other solutes represent less than 10% of total urinary solutes. Thus, estimated urine osmolarity is often used as an approximation of true osmolarity. Regarding the difference between osmolality and osmolarity, it is negligible in our study because the density of urine is very close to that of pure water in the range of values considered here. Other authors have used estimated urine osmolarity in several studies. Several authors have studied urine volume and GFR decline. Opposed to the prevailing view that water is beneficial in CKD, Hebert et al. reported higher GFR decline in higher urine volume quartiles. However, multivariable adjustment seemed to diminish the association. In support of Hebert et al., Wang et al. found a weak, but significant association between higher urine volume and GFR decline. A large study by Clark et al. showed a relationship between higher urine volume and lower rate of eGFR decline, which stayed significant after multivariable adjustment. Another large study reported worse kidney function in individuals with a lower self-reported fluid intake;however, neither urine volume nor urine osmolarity were evaluated. It has been suggested that lower mean baseline GFR in cohorts of Hebert et al. and Wang et al., which is associated with alterations in water metabolism, might explain these paradoxical findings. With falling GFR, the ability to concentrate urine to osmolalities greater than that of plasma is progressively lost.