Comment on osmotically inactive ions.

TO THE EDITOR: Nguyen et al. (6) attempted to show the physiological significance of the coefficients of their linear regression fit to the data (6) of Edelman et al. (Table 1) and a similar one of Boling et al. (Table 2) in regard to understanding the physiological factors responsible for dysnatremias. An understanding of this clinical condition is very important, but this paper has a number of technical flaws that render its conclusions invalid. 1) The statistical treatment of the data (Tables 1 and 2) in the Nguyen et al. (6) paper is flawed as first suggested by Ring (8) and confirmed here. Figure 1 shows their Table 1 data (open circles) and my linear regression analysis (solid line) obtained using the Sigma Stat computer program. Both the slope (0.88) and Y-intercept (10.0) are quite different from those found by Edelman et al. and also different than their analysis (6) of Edelman’s data, 0.93 and 1.37, respectively. Similarly, their analysis (6) of the Boling data in Table 2 gave somewhat different results (0.487, 71.54) from my analysis of that data (0.461, 73.8) using Sigma Stat. 2) They suggested (6) that a valid estimate of the osmotically inactive Na and K could be obtained by extrapolating the regression line to the origin. Figure 1 shows this extrapolation and that the region of 95% confidence (bounded by the dotted lines) grows as the extrapolation gets further from the experimental data region. At the y-axis, it lies between 11 and 32 with 95% certainty and includes zero (no osmotically active electrolytes). Furthermore, the intercept of 10 meq/LW is not statistically significant (P 0.37). An analysis of the Boling data gives similar results. 3) The assumption of a linear model was questioned by Ring (8) and confirmed herein (see item 4, below). My validated mathematical model of body fluid and electrolytes (9) shows in Fig. 1 that subtracting (i.e., urine loss) of either a mass of NaCl (long-dash) or NaHCO3 (short-dash) from the normal Na mass produces far different nonlinear decreases in plasma water Na concentration ([Na ]pw), which at best, suggest quite different values if extrapolated. 4) Their present Eq. 2 (Eq. 1, below) to determine [Na ]pw was first derived in 2003 as Eq. 2 (2), under the assumption of osmotic equilibrium (equal osmolarities) in the fluid compartments in the whole body. In this derivation, G/Ø was implicitly equal to one. It was changed to its present form in 2004 (4). It was again changed in 2006 (5) by eliminating the G/Ø term and having a factor g/Ø multiply only the first two sets of terms in Eq. 1 (their Eq. 12, Eq. 2, below). This latter equation is flawed as first suggested by Ring (7), subsequently by Dorrington (1), and now from my analysis (see below). Na pw G