Risk factors for undercorrection in severe hyponatremia: a post-hoc analysis of the SALSA trial
Article information
Abstract
Background
Hyponatremia undercorrection (i.e., nonoptimal correction) is insufficient to resolve cerebral edema and is associated with poor clinical outcomes. This study aimed to identify risk factors of undercorrection under controlled hypertonic saline treatment.
Methods
This post-hoc analysis of the prospective randomized controlled SALSA (Efficacy and Safety of Rapid Intermittent Correction Compared With Slow Continuous Correction With Hypertonic Saline) trial included 172 patients with glucose-corrected serum sodium (sNa) ≤125 mmol/L after excluding six without follow-up sNa measurements. Undercorrection was defined as sNa increases of <5 or <10 mmol/L within 24 or 24–48 hours, respectively, or failure to reach 130 mmol/L within 24–48 hours.
Results
The mean age was 72.8 years; the mean sNa level was 118.2 mmol/L. Undercorrection occurred in 15.1% of patients (26/172). These patients received more hypertonic saline (932 mL vs. 486 mL, p < 0.001) and had lower urine output over 48 hours (2,647 mL vs. 4,117 mL, p = 0.004). Risk factors included higher urine osmolality and serum calcium and creatinine levels, and lower body mass index, systolic blood pressure, white blood cell count, and albumin levels. Although most etiologic categories were not independently associated with undercorrection, thiazide use was inversely associated with undercorrection in the multivariable analysis. Undercorrection did not increase mortality but was associated with prolonged hospitalization (19.6 days vs. 7.8 days, p = 0.01).
Conclusion
Despite protocol-based hypertonic saline treatment, undercorrection occurred in 15% of patients and was associated with patient-specific risk factors. These findings emphasize the need for individualized sodium correction strategies to optimize patient outcomes.
Introduction
Hyponatremia is the most prevalent electrolyte imbalance in hospitalized patients, with an incidence of 14%–42%. It is associated with increased mortality and prolonged hospital stays [1–3]. Clinical manifestations range from mild symptoms, such as fatigue, nausea, vomiting, headache, gait disorder, and confusion, to severe symptoms, such as seizures, coma, and brain hypoxia [4–7].
Hypertonic saline remains the cornerstone of treatment for symptomatic hyponatremia, and correction protocols are designed to prevent complications [8,9]. The primary treatment concern has been overcorrection, which can lead to osmotic demyelination syndrome (ODS) [10]. However, inadequate sodium correction (undercorrection) carries a significant risk. Failure to achieve sufficient correction may result in persistent cerebral edema, prolonged hospitalization, and poor clinical outcomes [11–15]. Despite its importance, undercorrection remains poorly characterized and lacks a standardized definition or clear management strategies.
While some guidelines and studies suggest defining undercorrection as a serum sodium (sNa) level increase of <5 mmol/L within 24 hours, <10 mmol/L within 48 hours, or failure to reach 130 mmol/L within 48 hours [1,15,16], few prospective studies have evaluated its clinical impact [1,17–19]. Undercorrection occurs in 20%–41% of patients treated with hypertonic saline, yet the reason for its occurrence remains unclear [20,21]. These concerns warrant further investigation into the predictors and clinical impact of undercorrection.
The SALSA (Efficacy and Safety of Rapid Intermittent Correction Compared With Slow Continuous Correction With Hypertonic Saline) trial previously focused on mitigating overcorrection risks [22,23]. However, the prevalence of undercorrection, despite adherence to standardized protocols, suggests an unmet clinical need. This post-hoc analysis of the SALSA trial aimed to identify the risk factors for undercorrection in patients with severe hyponatremia and assess its clinical impact, particularly regarding hospitalization duration and symptom resolution. Identifying these factors may help optimize sodium correction strategies and improve patient outcomes.
Methods
Study design and population
This study was a post-hoc analysis of the SALSA trial, a prospective, randomized controlled trial investigating hypertonic saline treatment strategies for symptomatic hyponatremia. Among 178 eligible patients with glucose-corrected sNa levels of ≤125 mmol/L at enrollment who were admitted to Seoul National University Bundang Hospital, SMG-SNU Boramae Medical Center, and Hallym University Dongtan Sacred Heart Hospital between August 2016 and August 2019, six patients were excluded because of missing follow-up sNa values, resulting in a final analysis of 172 patients. The detailed study protocol has been described previously (ClinicalTrials.gov; NCT02887469) [24]. This post-hoc analysis was approved by the Institutional Review Boards of three centers: Seoul National University Bundang Hospital (No. B-2504-964-103), SMG-SNU Boramae Medical Center (No. 40-2025-29), and Hallym University Dongtan Sacred Heart Hospital (No. 2025-03-007). The study adhered to the ethical guidelines of the 1975 Declaration of Helsinki. Written informed consent was obtained from all participants.
Definition of undercorrection
Undercorrection was defined using the following criteria: sNa level increase <5 mmol/L within 24 hours; sNa level increase of <10 mmol/L or failure to reach 130 mmol/L within 24–48 hours.
These thresholds were based on previously published guidelines and expert consensus regarding the optimal correction rates for hyponatremia [1,18,20,21,25].
Outcome measures and data collection
Clinical data, including baseline patient demographics, alcohol consumption information, laboratory findings, urine studies, and hospital courses, were collected. Regardless of the type of alcohol consumed, chronic alcoholism was defined as the consumption of at least four and three glasses of alcohol per day by males and females, respectively. Comorbidities, such as hypertension, diabetes mellitus, congestive heart failure, and cancer, were identified by screening for I10–I15, E10–14, I11.0, I13.0, I13.2, I50, and C codes based on the International Classification of Diseases, 10th Revision. We also investigated self-reported and verified histories of antihypertensive and antidiabetic medication use.
Hyponatremia was defined based on the biochemical severity; “severe (profound) hyponatremia” referred to sNa levels of ≤125 mEq/L [1,18]. The clinical presentation at the initial detection of hyponatremia was used to categorize the clinical manifestations of hyponatremia as moderate or severe. Moderate symptoms included nausea, drowsiness, headache, general fatigue, and malaise [25]. Severe symptoms included vomiting, stupor, seizures, and coma, defined as a Glasgow Coma Scale score of 8 [1]. Somnolence was defined as drowsiness or stupor. The underlying cause of the hyponatremia was identified using a systematic diagnostic approach that incorporated the patient’s history, physical examination, and laboratory findings.
Previously published protocols outlined two methods for administering hypertonic saline: rapid intermittent bolus (RIB) and slow continuous infusion (SCI) [24,26]. Hypertonic saline was administered using a protocolized strategy tailored to symptom severity. Patients assigned to the RIB group received weight-based boluses of 3% saline (2–4 mL/kg, or 100–200 mL when body weight was unknown) over 20–40 minutes, with repeat boluses at each monitoring point until the predefined correction targets were achieved. Those in the SCI group received 3% saline at 0.5–1.0 mL/kg/hr initially, with the infusion rate subsequently increased, maintained, or discontinued according to serial sodium measurements. In both groups, adjustments were guided by standardized correction limits (targeting 5–9 mmol/L at 24 hours and 10–17 mmol/L at 48 hours or achievement of symptom resolution). Details of the treatment guidelines for both groups can be found in Supplementary Methods (available online) and Supplementary Fig. 1 (available online). The initial infusion rate was determined based on the severity of the hyponatremia symptoms. The treatment goals were to raise the sNa level by 5–9 mEq/L and alleviate symptoms within the first 24 hours and to increase the sNa level by 10–17 mEq/L or to ≥130 mEq/L, along with symptom relief, within the first 48 hours. sNa levels were measured every 6 hours over 2 days using indirect ion-selective electrodes at the following centers using the specified equipment: Seoul National University Bundang Hospital (AU5800 [Beckman Coulter] and Dimension Vista 1500 [Siemens Healthineers]), SMG-SNU Boramae Medical Center (Modular DP [Roche Diagnostics] and Unicel DxC 800 [Beckman Coulter]), and Hallym University Dongtan Sacred Heart Hospital (AU5800 [Beckman Coulter]).
The primary outcomes were undercorrection and independent risk factors. Other clinical outcomes related to undercorrection included the length of hospital stay, in-hospital mortality, and symptom resolution.
Statistical analysis
Baseline characteristics were compared using the chi-square test for categorical variables and the t test or Mann-Whitney U test for continuous variables, depending on the distribution. Univariate logistic regression was performed for each variable to determine the significant risk factors associated with the undercorrection of hyponatremia. Multivariate logistic regression was performed to identify the risk factors for undercorrection, adjusting for clinically relevant variables such as urine osmolality, baseline sodium level, volume status, and hypertonic saline infusion method. To compare the trajectories of plasma sodium levels between patients with and without undercorrection, we applied a linear mixed-effects model with time, group, and their interaction as fixed effects and subject as a random effect. Variables with p-values <0.05 were retained in the multivariate model using backward selection. Odds ratios (ORs) and 95% confidence intervals (CIs) for hyponatremia undercorrection were calculated after stepwise adjustments for multiple confounders. Variables with p-values <0.05 were considered statistically significant. All analyses were performed using IBM SPSS version 29.0 (IBM Corp.) and R version 4.3.3 (R Foundation for Statistical Computing).
Results
Study population
The study enrolled 178 patients with symptomatic hyponatremia and glucose-corrected sNa levels of ≤125 mmol/L between August 24, 2016, and August 21, 2019. Six patients without sNa values were excluded, resulting in a final analysis of 172 patients. The mean age was 72.8 ± 12.3 years, and 45.3% were male. The most common causes of hyponatremia were thiazide diuretic use (n = 52, 30.2%), syndrome of inappropriate antidiuresis (SIAD; n = 49, 28.5%), adrenal insufficiency (n = 43, 25.0%), non-renal sodium loss with decreased extracellular fluid (ECF) volume (n = 39, 22.7%), and increased ECF volume (n = 17, 9.9%). Among the patients, 41 (23.8%) had malignancy, and five (2.9%) had a history of chronic alcoholism. Moderate and severe symptoms were observed in 132 (76.7%) and 40 (23.3%) patients, respectively. At admission, the mean sNa and serum potassium levels were 118.2 ± 5.1 and 4.0 ± 0.8 mmol/L, respectively. The serum and urine osmolality at admission were 251.1 ± 21.9 and 424.3 ± 162.0 mOsm/kg, respectively.
Incidence and risk factors for hyponatremia undercorrection
Undercorrection occurred in 15.1% of patients (26/172). The patients were divided into the undercorrection and no undercorrection groups, and their baseline characteristics are summarized in Table 1. Patients with undercorrection were more likely to be male, have a lower body mass index (BMI), and exhibit differences in comorbid conditions, including higher rates of malignancy. A lower systolic blood pressure (126.3 ± 16.7 mmHg vs. 141.2 ± 26.3 mmHg, p = 0.01), lower serum albumin level (3.5 ± 0.5 g/dL vs. 4.0 ± 0.6 g/dL, p < 0.001), lower total cholesterol level (129.7 ± 34.8 mg/dL vs. 159.5 ± 45.5 mg/dL, p = 0.001), and higher urine osmolality (529.8 ± 216.7 mOsm/kg vs. 405.0 ± 142.6 mOsm/kg, p = 0.01) were associated with a higher likelihood of undercorrection. Most etiologic categories, including SIAD, adrenal insufficiency, and volume status abnormalities, were not independently associated with undercorrection. However, thiazide use was inversely associated with undercorrection in multivariable analysis. Consistent with this finding, thiazide-associated hyponatremia was less frequent in the undercorrection group than in the no undercorrection group (7.7% vs. 34.2%, p = 0.01). At both 24 and 48 hours, the undercorrection group received larger cumulative volumes of hypertonic saline than the no undercorrection group (24 hours: 559.5 mL vs. 367.6 mL; 48 hours: 932.4 mL vs. 486.6 mL; both p < 0.001) (Fig. 1). Additionally, the 48-hour urine output was significantly lower in the undercorrection group than in the no undercorrection group (2,647.6 mL vs. 4,117.8 mL, p = 0.004) (Fig. 2). In the linear mixed-effects model evaluating longitudinal changes in sNa, sodium levels increased significantly over time in both groups (time effect: estimate = 0.218 mmol/L per hour, p < 0.001). Baseline differences between the undercorrection and no undercorrection groups were not significant (group effect, p = 0.99). However, the time-by-group interaction was strongly significant (estimate = –0.104 mmol/L per hour, p < 0.001), indicating that the rate of sodium increase over time was significantly slower in the undercorrection group. These findings are consistent with the visual trajectories shown in Fig. 3.
Baseline characteristics and progression according to undercorrection
Cumulative 3% saline volume administered over time in patients with and without undercorrection.
Comparison of 48-hour urine output between patients with and without undercorrection.
Plasma sodium change over time in patients with and without undercorrection.
LMM, linear mixed model.
Univariate and multivariate logistic regression analyses were performed to identify independent risk factors for undercorrection (Table 2). Multivariate logistic regression analysis identified lower BMI (OR, 0.75; 95% CI, 0.59–0.97; p = 0.03), systolic blood pressure (OR, 0.95; 95% CI, 0.91–0.99; p = 0.01), serum albumin level (OR, 0.06; 95% CI, 0.01–0.55; p = 0.01), and white blood cell count (OR, 0.70; 95% CI, 0.53–0.92; p = 0.01) as significant risk factors for undercorrection, after adjusting for all variables. Higher serum creatinine levels (OR, 4.96; 95% CI, 1.27–19.30; p = 0.02), serum calcium levels (OR, 12.12; 95% CI, 1.85–79.16; p = 0.01), and urine osmolality (OR, 1.01; 95% CI, 1.00–1.01; p < 0.01) were also risk factors for undercorrection. Baseline symptoms were independently associated with a lower risk of undercorrection, likely reflecting more aggressive early correction in symptomatic patients. However, the mode of hypertonic saline administration (bolus vs. continuous infusion) was not significantly associated with undercorrection (p = 0.08). These findings suggest that patient-specific factors, rather than the treatment modality or hyponatremia etiology, play a significant role in determining the sodium correction response.
Risk factors for undercorrection by univariate and multivariate logistic regression
In an additional exploratory analysis, the total NASK (hypoNatremia, Alcoholism, Severe symptoms, and hypoKalemia) score was inversely associated with undercorrection (OR, 0.83; 95% CI, 0.69–0.99; p = 0.04). However, its discriminatory performance for undercorrection was limited (area under the receiver operating characteristic curve, 0.39; 95% CI, 0.29–0.48; p = 0.02) (Supplementary Table 1, Supplementary Fig. 2; available online).
Clinical significance of undercorrection
The presence of undercorrection within 24 or 48 hours was associated with a significantly prolonged hospital stay (Table 3; Supplementary Fig. 3, available online). Patients without undercorrection had a mean hospital stay of 7.8 days (standard deviation [SD], 7.5), whereas those with undercorrection had a mean stay of 19.6 days (SD, 30.2) (p = 0.01). Despite the differences in sodium correction patterns (Fig. 3), the symptom persistence and mortality did not differ significantly between the two groups (Tables 3, 4).
Clinical outcome by undercorrection status
Symptom persistence at 24 and 48 hours
Discussion
In this post-hoc analysis of the SALSA trial, we investigated the incidence, risk factors, and clinical impact of undercorrection in patients with severe hyponatremia treated with protocolized hypertonic saline therapy. Despite adherence to treatment guidelines, undercorrection occurred in 15.1% of patients. Patients who experienced undercorrection required significantly higher cumulative volumes of hypertonic saline over 24 and 48 hours and demonstrated lower urine output, suggesting impaired sodium responsiveness. Undercorrection was associated with several clinical and biological characteristics, including a higher urine osmolality and serum calcium and creatinine levels, as well as a lower BMI, systolic blood pressure, white blood cell count, and serum albumin concentration. Undercorrection was clinically associated with prolonged hospitalization; however, no significant increase in mortality was observed.
The incidence of undercorrection in our study (15.1%) was lower than that reported in previous studies, which ranged from 20% to 41% in patients receiving hypertonic saline therapy [20,21]. Differences in patient populations, hyponatremia severity, and treatment protocols may account for this discrepancy. Prior research has highlighted the role of impaired free water clearance—due to persistent antidiuretic hormone (ADH) activity or renal dysfunction—as a key barrier to successful sodium correction [10,27]. The proportion of thiazide users was significantly lower in the undercorrection group compared to that in the no undercorrection group (7.7% vs. 34.2%, p = 0.01). This finding suggests that patients with thiazide-induced hyponatremia may exhibit a more favorable sodium response to hypertonic saline, which is potentially driven by the underlying volume depletion and natriuresis [1]. One possible explanation is that thiazide-induced volume depletion leads to compensatory activation of the renin-angiotensin-aldosterone system, which enhances sodium retention upon saline administration [28,29]. Additionally, previous studies have indicated that natriuresis in thiazide users may facilitate a more predictable correction pattern of sNa levels than other causes of hyponatremia, such as SIAD [30]. However, further studies are needed to confirm whether thiazide withdrawal alone is responsible for this response or whether the underlying physiological differences play a role.
The mechanisms underlying undercorrection primarily involve two pathophysiological processes: impaired free water clearance and altered ECF dynamics. Impaired free water clearance plays a pivotal role. High urine osmolality, reflecting persistent ADH activity, indicates a reduced ability to excrete free water, even with hypertonic saline administration [8,31]. Elevated serum creatinine levels further suggest that renal dysfunction impairs the ability of the kidneys to eliminate excess water and respond appropriately to sodium loading [32,33]. Although hypercalcemia is traditionally associated with polyuria, elevated serum calcium levels may paradoxically impair sodium correction by inducing nephrogenic resistance to ADH and disrupting the handling of tubular water. In addition, altered ECF dynamics contribute significantly. Lower serum albumin levels and a reduced BMI are markers of malnutrition and decreased oncotic pressure, both of which can limit the effective vascular volume and impair sodium distribution [34,35]. Similarly, lower systolic blood pressure may reflect underlying hypovolemia, which triggers compensatory ADH secretion, exacerbates free water retention, and hinders sodium correction [14,32,36,37]. Collectively, these findings suggest that undercorrection results from the cumulative effects of impaired water excretion and dysregulated fluid distribution, which act synergistically to limit the efficacy of hypertonic saline therapy. This underscores the complexity of electrolyte management in patients with multifactorial risk factors and highlights the need for individualized correction strategies. These risk factors may assist clinicians in identifying patients who are likely to correct more slowly, allowing earlier adjustments such as initiating slightly higher initial infusion intensity within protocol limits, shortening the interval of sodium monitoring, or reassessing free water clearance earlier in the course of treatment. Integrating these individualized modifications into standardized hypertonic saline protocols may help reduce the likelihood of persistent undercorrection.
Importantly, the clinical relevance of undercorrection extends beyond biochemical targets. In our study, patients with undercorrection required substantially higher cumulative volumes of hypertonic saline yet exhibited a slower rise in sNa and a significantly prolonged hospital stay (7.8 days vs. 19.6 days, p = 0.01). This pattern suggests that undercorrection reflects reduced sodium responsiveness and impaired free water clearance rather than insufficient treatment intensity. From a healthcare utilization perspective, undercorrection may therefore represent a form of treatment inefficiency, leading to prolonged monitoring, repeated dose adjustments, and extended hospitalization despite adherence to standardized correction protocols. However, whether undercorrection directly contributes to prolonged hospitalization or instead reflects underlying clinical complexity remains uncertain. Patients in the undercorrection group had lower serum albumin levels and a higher prevalence of malignancy, factors that may independently contribute to longer hospital stays. Accordingly, future studies with larger cohorts should adjust for these comorbid conditions to clarify the independent impact of undercorrection on hospitalization duration.
Despite these differences in sodium correction patterns and hospital stay durations, symptom persistence at 24 and 48 hours did not differ significantly between the groups. The proportion of patients with persistent symptoms at 24 hours was 16.6%, with no significant between-group differences (13.0% vs. 17.2%, p = 0.79). Similar results were observed at 48 hours (8.7% vs. 7.5%, p = 0.69). The dissociation between biochemical correction and clinical improvement suggests that factors beyond sNa normalization, such as underlying frailty, comorbid conditions, and neurological status, substantially influence symptom recovery. Although the between-group differences in the in-hospital mortality were not statistically significant, the undercorrection group had a higher proportion of deaths (7.7% vs. 2.1%, p = 0.17), suggesting a potential trend toward worse outcomes. This highlights the need for early identification of at-risk patients and close monitoring of the sodium trajectory to optimize correction strategies, particularly among those with impaired renal function and persistent water retention tendencies. In an exploratory stratified analysis based on absolute 24-hour sodium correction rate (Supplementary Tables 2 and 3, available online), no consistent association between correction velocity and mortality was observed. Mortality rates were low across all correction-speed strata and did not demonstrate a monotonic relationship with increasing or decreasing correction speed. Hospital duration differed significantly according to the 24-hour sodium correction rate strata (p = 0.02), with the very slow correction group showing the longest hospital stay. Early symptom persistence at 24 hours appeared more frequently in the very fast correction group. However, these patterns were not uniform and did not persist at 48 hours. Given the small number of events and limited sample size within extreme correction strata, these findings should be interpreted descriptively. Collectively, these results suggest that correction velocity alone may not be a primary determinant of short-term clinical outcomes, and that underlying patient-specific pathophysiological factors likely play a more substantial role. Although the total NASK score showed an inverse association with undercorrection in an exploratory supplementary analysis, this finding should not be interpreted as validation of the NASK score for predicting undercorrection. The NASK score was originally derived for overcorrection, and its inverse association with undercorrection may simply reflect that baseline factors predisposing to rapid correction are, in aggregate, associated with a lower likelihood of insufficient correction.
This study has several limitations. First, the relatively small number of patients experiencing undercorrection limits statistical power and may reduce the generalizability of our findings, particularly for subgroup or stratified analyses. Second, the study population primarily consisted of patients with moderate-to-severe hyponatremia; and therefore, the results may not be directly applicable to patients with milder disease. Third, this study is a post-hoc analysis of a randomized controlled trial that was not originally designed to evaluate undercorrection. Therefore, causal interpretation should be made cautiously. However, as the parent SALSA trial was a prospective randomized controlled trial with protocolized treatment and predefined sodium monitoring, key variables were systematically collected, and missing data were minimal, supporting the internal validity of this analysis. Finally, although undercorrection was defined based on guideline-recommended correction targets, no universally accepted definition exists, and alternative thresholds may yield different clinical associations. Future prospective studies with larger and more diverse cohorts are warranted to validate these findings and explore individualized correction protocols that balance the risks of both undercorrection and overcorrection.
Despite these limitations, our study has several strengths that contribute to the field of hyponatremia management. First, while most of the studies on hyponatremia have focused on the dangers of overcorrection, data on undercorrection, including its incidence, predictors, and clinical consequences, remain scarce. This study provides novel insights into this underexplored area using prospective data from a randomized controlled trial. Second, analyzing a cohort of patients uniformly treated with protocolized hypertonic saline enabled the isolation of patient-specific factors that influenced the treatment response. This methodological consistency strengthens the validity of our findings by minimizing the confounding effects of treatment variability and allowing for clearer identification of clinical predictors associated with undercorrection.
Traditionally, the management of hyponatremia has focused on preventing overcorrection to mitigate the risk of ODS. However, our findings emphasize that undercorrection represents a significant clinical concern and is associated with a suboptimal treatment response and delayed recovery. As patients with undercorrection require higher cumulative doses of hypertonic saline and exhibit impaired sodium responsiveness, clinicians should closely monitor sodium trajectories and proactively adjust the therapy. A more individualized correction approach that ensures safe but sufficient sodium level elevations may help optimize clinical outcomes while minimizing complications. Future studies should focus on refining treatment strategies that address both extremes of correction to ensure safer and more effective management of severe hyponatremia.
Supplementary Materials
Supplementary data are available at Kidney Research and Clinical Practice online (https://doi.org/10.23876/j.krcp.25.373).
Notes
Conflicts of interest
All authors have no conflicts of interest to declare.
Funding
The research was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (2016R1C1B1015687 and RS-2021-NR061495).
Data sharing statement
The data presented in this study are available from the corresponding author upon reasonable request.
Authors’ contributions
Conceptualization, Data curation: GY, SK, SHB
Funding: SK, SHB
Formal analysis, Methodology: GY, SHB
Investigation, Project administration: JWS, JRK, YKO, YHJ, SK, SHB
Software, Validation, Visualization: GY
Supervision: SHB
Writing–original draft: GY, SHB
Writing–review & editing: All authors
All authors read and approved the final manuscript.