Comparison of electrical conductivity and freezing point methods for urine osmolality measurement and the influence of urinary biochemical parameters

Introduction

Urinalysis has become a frequently requested test today due to its accessibility and the valuable information it provides. Its demand continues to rise, as it is used not only for evaluating urinary tract pathologies but also for monitoring systemic diseases affecting the urinary system. To address this growing demand, manufacturers of fully automated urine analysers have started to incorporate novel parameters into urinalysis, alongside the traditional ones (pH, specific gravity, erythrocytes, leukocytes, nitrite, protein, glucose, ketone, urobilinogen, and urine sediment analysis), to enhance the detection and monitoring of pathologies [1]. In this context, some fully automated urine analysers can also measure urine osmolality in the tested samples.

Osmolality is the total number of solute particles dissolved in 1 kg of solvent [2]. Urine osmolality plays an important role not only in the diagnosis of disorders associated with impaired renal concentrating ability due to kidney damage but also in conditions related to hormonal dysregulation that alter urine concentration [3], 4]. In addition to its use in diagnosis and differential diagnosis, urine osmolality is also used to assess adequate fluid intake. It serves as a risk factor in the prognosis of chronic kidney disease [5], 6]. However, inaccurate urine osmolality measurements may result in misdiagnosis, inappropriate treatment, or delays in the diagnostic and therapeutic processes. Therefore, measurement techniques employed for urine osmolality and the factors that may affect these methods are essential for ensuring reliable results.

Several methods based on the colligative properties of fluids, such as vapour pressure, boiling point, freezing point, and osmotic pressure, are employed for osmolality measurement [7]. Osmometers utilizing these methods are less commonly available in clinical laboratories than routine urinalysis and biochemistry analysers [8]. Moreover, the need for consumable materials in osmolality measurement poses an additional limitation [2]. Consequently, osmolality calculation formulas have been developed for situations where direct measurement is not feasible [7], 9]. However, significant discrepancies between calculated and measured osmolality values limit the applicability of these formulas [10], 11]. Thus, there remains a need for methods that can provide accurate osmolality results in clinical laboratories without requiring additional consumables or equipment.

Osmolality can also be assessed by the electrical conductivity method, which is predicated on the principle that ions dissolved in a solution contribute to its electrical conductivity. A significant advantage of this technique is its compatibility with integration into fully automated urinalysis platforms. Given its broad accessibility and the absence of additional consumable requirements, this method has recently attracted attention as a potential alternative to classical colligative property-based osmometry [12], 13]. Nevertheless, the factors that may influence urine osmolality results obtained by this method have not yet been fully elucidated.

In the present study, we sought to evaluate the analytical performance of two distinct systems: The Sysmex UF-5000 (Sysmex Corporation, Kobe, Japan), a fully automated urine sediment analyser equipped with an electrical conductivity module, and the OSMOMAT 3000 basic (Gonotec, Berlin, Germany), which operates on the freezing point depression principle. Additionally, we aimed to investigate the extent to which various urinary molecules (total protein, creatinine, microalbumin, urea, glucose, calcium, potassium, sodium, and chloride) may affect the trueness of electrical conductivity-based osmolality measurements.

Materials and methods

Results

The precision of the Sysmex and Osmometer methods was evaluated across four control levels by assessing within-run, between-run, within-day, and between-day variability, as well as total within-laboratory variability. The results are summarized in Table 1. Within-run precision for the Sysmex device exhibited %CV values ranging from 0.3 to 7.8 %, whereas the osmometer demonstrated %CV values between 0.7 and 9.4 %. Between-run precision for Sysmex showed %CV values up to 0.2 %, while the osmometer displayed %CV values up to 7.8 %. Within-laboratory precision ranged from 0.3 to 7.8 % (%CV) for Sysmex and 0.7–12.4 % for the osmometer.

Trueness analysis indicated that, for the osmometer, the assigned value was 290.0, the mean was 310.6±9.3, the bias was 20.6, the 95 % confidence interval (CI) ranged from 11.5 to 29.6, and the relative bias was 7.1 %. For Sysmex, the assigned value was 500.0, the mean was 537.6±10.7, the bias was 37.6, the 95 % CI ranged from 30.4 to 44.7, and the relative bias was 7.5 % (Table 2).

Comparison of measurements between the Osmometer and Sysmex devices included 210 samples for each method. The median values (1st–3rd quartile) were 541 (380–771) for the osmometer and 563.5 (360–802) for the Sysmex. The minimum and maximum values ranged from 82 to 1,250 for the osmometer and 60 to 1,238 for Sysmex, with a p-value of 0.348 from the Wilcoxon signed-rank test (Table 3).

Correlation analysis revealed a strong positive correlation between the two methods (Spearman’s ρ=0.857, p<0.01). Passing–Bablok regression analysis indicated no significant deviation from linearity (p=0.49), with the regression equation: Sysmex=−46.74 (−77.5 to −7.96) + 1.111 (1.038 to 1.187) × Osmometer.

Bland–Altman analysis demonstrated an absolute mean difference of 2.507 with an SD of 151.766 and a SE of 10.029. The 95 % CI for the absolute mean difference ranged from −294.955 to 299.969. For the relative mean difference, the mean was 2.462, with an SD of 32.369, SE of 2.139, and a 95 % CI ranging from −60.981 to 65.905 (Figure 1).

Figure 1:

Bland–Altman plot illustrating the percentage difference of Sysmex and Osmometer measurements as a function of the osmolality of the osmometer. The solid black lines denote the mean difference and allowable bias, while the dashed lines represent the 95 % confidence intervals (CIs).

Regression analyses were conducted separately for the OSMOMAT 3000 basic and Sysmex UF-5000 to identify predictors of urine osmolality. Both models exhibited high explanatory power (OSMOMAT: R=0.9623, Adjusted R²=0.9234; Sysmex: R=0.9458, Adjusted R²=0.8908). For the OSMOMAT 3000 basic, urea nitrogen, creatinine, calcium, potassium, and sodium were significant predictors (all p<0.001), whereas glucose and microalbumin did not contribute meaningfully (p>0.05). Similarly, for the Sysmex UF-5000, sodium, potassium, calcium, and creatinine were significant predictors (all p<0.001), while glucose, urea nitrogen, and microalbumin were not (p>0.05). Overall, both models indicated that sodium, potassium, calcium, and creatinine consistently accounted for a substantial portion of the variance in osmolality results. In contrast, glucose, urea nitrogen, and microalbumin had negligible contributions (Table 4).

Discussion

In our study, we evaluated and compared the analytical performance of the Sysmex UF-5000, which measures osmolality using the electrical conductivity method, and the OSMOMAT 3000 basic, which employs the freezing point depression method. In addition, we investigated the effects of urinary biochemical parameters (total protein, creatinine, microalbumin, urea, glucose, calcium, potassium, sodium, and chloride) on the osmolality results obtained from both devices.

In the precision evaluation of both devices, the within-run, between-run, within-day, between-day, and total %CV values across four concentration levels were found to be lower than the maximum acceptable value of 14.2 % [16]. The total %CV values obtained from the Sysmex UF-5000 were lower at all four levels compared to those from the OSMOMAT 3000 basic. The %CV values of Sysmex UF-5000 and OSMOMAT 3000 basic were similar to those reported in other studies (2–5). However, the OSMOMAT 3000 basic values were higher than the manufacturer’s stated limits of <2 mOsmol/kg SD and <0.5 % CV at concentrations of 400 mOsmol/kg and 1,500 mOsmol/kg, respectively [17].

When the trueness of the methods was evaluated, the %bias values for the Sysmex UF-5000 and OSMOMAT 3000 basic were calculated as 7.5 and 7.1 % respectively. In our study, the %bias values for the Sysmex UF-5000 were consistent with those reported by Yoo et al. [12], but higher compared to the study by Yis et al. [2], Revell et al. [18] reported %bias values for the OSMOMAT 3000 basics that were similar to those obtained in our study. Importantly, in our results, the %bias values of both devices were comparable and well below the predefined acceptable threshold of 16.1 %, supporting the analytical trueness of both methods [16].

In the comparison performed with patient samples, no significant difference was observed between the results obtained from the two devices, and a strong correlation was found. However, Bland–Altman analysis revealed that 26.6 % of the results from the Sysmex UF-5000 differed from those of the OSMOMAT 3000 basic by more than the allowable total error (TEa) of 28.1 %. In our study, the proportion of samples exceeding the allowable total error was higher than that reported by Yis et al. [2], but lower than that of Yoo et al. [12]. Collectively, our findings suggest that while both methods demonstrate acceptable analytical performance, the Sysmex UF-5000 may be more vulnerable to matrix effects in specific patient groups compared with freezing-point osmometry.

These discrepancies observed in the Bland–Altman analysis led us to further investigate the underlying factors. Our multiple regression analysis was performed to investigate the possible effects of urinary biochemical parameters on osmolality. Due to multicollinearity, total protein was excluded in the presence of microalbumin, and chloride in the presence of sodium. The osmolality results obtained from the OSMOMAT 3000 basic (freezing point method) were associated with sodium, potassium, calcium, urea, and creatinine, but not with glucose or microalbumin. Sodium, potassium, and calcium are water-soluble ions that lower the freezing point, a colligative property of solutions [19]. Similarly, water-soluble molecules such as creatinine and urea also affect the freezing point. The influence of creatinine on osmolality measured by the freezing point method has also been demonstrated [20], 21]. While urea is typically included in serum osmolality calculations, creatinine is generally excluded. The effect of creatinine on urine osmolality, despite its negligible role in serum, may be explained by the higher concentrations of creatinine in urine compared to serum. Likewise, glucose is a water-soluble molecule that affects the freezing point and is included in osmolality calculation formulas [10]. However, in our regression analysis, glucose did not affect the results obtained from the OSMOMAT 3000 basic. Under normal conditions, renal tubules reabsorb almost all glucose, leaving very low levels in urine [22]. Therefore, the urine concentrations of glucose in our sample group were low, which may have prevented the expected effect on the freezing point method from being observed.

Microalbumin, as a water-soluble molecule, can lower the freezing point. However, due to its much larger molecular weight compared to smaller molecules such as creatinine and urea, it produces a lower molar concentration in solution at similar mass/volume concentrations [23]. Since freezing point depression depends on molar concentration rather than mass, the effect of microalbumin on the freezing point method is negligible [24]. This finding agrees with prior reports, reinforcing that macromolecules contribute minimally to osmolality when assessed by colligative property-based methods.

To understand the distinct effects of these compounds on the electrical conductivity method, we also conducted multiple regression analysis for the Sysmex UF-5000. Sodium, potassium, calcium, and creatinine were found to be associated, while microalbumin, urea, and glucose were not. Sodium, potassium, and calcium are present in urine as ionized solutes, which increase electrical conductivity and thus influence this method. Creatinine is a zwitterion molecule and would not be expected to affect electrical conductivity [25]. Creatinine can bind protons under acidic conditions, acquiring a positive charge. Although only a small fraction is protonated at physiological pH (∼7.4), the proportion of protonated creatinine increases as the pH decreases, consistent with its chemical behaviour in acidic environments [26]. In our sample set, 72.9 % of pH results obtained by dipstick were <6. Therefore, the relatively high urinary creatinine concentrations combined with low pH may have led to the observed effect of creatinine on electrical conductivity.

A particularly important finding of our study is the differential impact of non-ionic molecules. Both urea and glucose are nonpolar molecules that do not increase electrical conductivity [25], 27]. Although urea affects osmolality, as measured by the freezing point method, it cannot be measured by electrical conductivity. In one study, urea’s effect on osmolality was indirectly demonstrated through urease, which generates ammonia that increases conductivity [25]. Thus, while urea influences the freezing point method, it does not affect electrical conductivity, which may partly explain discrepancies between the two methods. Similarly, glucose affects the freezing point but not conductivity, and significant differences between the two methods have been reported in urine samples with high glucose levels [2]. However, another study demonstrated that glucose decreases conductivity, suggesting a potential negative interference in osmolality measured by this method [27]. We did not observe this negative interference in our study, likely due to the low glucose concentrations in our sample set, as observed in the freezing point method. Studies using samples with higher urinary glucose levels may better demonstrate this effect.

Depending on pH, albumin (microalbumin) is a polar molecule that carries different charges and, therefore, could affect electrical conductivity [28]. However, similar to its lack of effect on the freezing point method in our study, no effect was observed in the electrical conductivity method. This may be attributed to the low molar concentrations of microalbumin in urine. Taken together, our findings suggest that the impact of macromolecules and non-ionic solutes on conductivity-based osmolality is minimal compared to that of small electrolytes and pH-dependent charged species. These results highlight that conductivity-based osmolality primarily reflects the ionic milieu of urine, whereas the contribution of larger or non-ionic molecules is negligible under routine clinical conditions. Future studies including samples with higher concentrations of non-ionic solutes are warranted to clarify their potential impact in specific patient populations.

Conclusions

In this study, we compared the electrical conductivity and freezing point methods for urine osmolality measurement. Both methods demonstrated high reproducibility and had acceptable bias values. Our findings indicated that the electrical conductivity-based method is more sensitive to matrix effects, such as urine ionic composition and pH, than the freezing point method, while the contribution of large or non-ionic molecules to this method is negligible.

From a clinical perspective, conductivity-based osmolality measurements are a suitable and practical alternative for routine use in most outpatient populations. However, in cases where urinary solute composition deviates significantly from the norm, such as in uncontrolled diabetes mellitus, nephrotic syndrome, or intensive care unit (ICU) settings, conductivity results should be interpreted with caution. In these scenarios, freezing point osmometry remains the reference method. Potential differences in paediatric and geriatric populations should also be considered; while conductivity may be a practical screening tool in paediatrics, confirmatory measurements may be needed in elderly patients.

Overall, our results emphasize that electrical conductivity is a reagent-free, efficient alternative for estimating urine osmolality. Future multicentre studies are needed to better define the trueness in patients with various pathophysiological conditions and to fully delineate the clinical applicability of conductivity-based measurements.