Critically ill patients are at high risk for dysnatremia because of the nature of the acute illness and to lack of free access to water (1–3). Up to one third of patients have dysnatremia at intensive care unit (ICU) admission (2, 4, 5), and another third develop dysnatremia in the ICU (3, 6). However, the prevalence of dysnatremia at ICU admission varies greatly according to the definition used (2, 4, 7, 8).
Dysnatremia is known to adversely affect physiologic functions. Hypernatremia has been associated with peripheral insulin resistance, impaired hepatic lactate clearance, decreased left ventricular contractility, and various neuromuscular manifestations (9–12). Similarly, neurological abnormalities may develop in patients with profound hyponatremia (13, 14). In several studies, mild-to-severe dysnatremia was associated with excess mortality in critically ill patients after adjustment for confounders (2, 3, 5, 6, 13, 15). The association between serum sodium concentration and mortality followed a U shape, with greater severity of hyponatremia or hypernatremia being associated with higher mortality rates (2, 13, 15). However, the existing data cannot serve to indicate whether the dysnatremia-mortality association reflects a direct effect of dysnatremia, an association between dysnatremia and underlying comorbidities or reason for ICU admission, or both (13). Evaluating the influence of dysnatremia correction on mortality in patients with mild-to-severe dysnatremia may help to resolve this issue.
We hypothesized that dysnatremia correction and/or the rate of dysnatremia correction might eliminate the association between dysnatremia and mortality. The primary objective of this observational database study in critically ill patients with dysnatremia at ICU admission was to determine whether dysnatremia correction by day 3 influenced day 28 mortality. The secondary objective was to assess the potential influence of the rate of serum sodium correction on day 28 mortality.
Study design and data source
We conducted an observational study on a prospective multicenter database (OutcomeRea; www.outcomerea.org) fed by 18 French ICUs. The database prospectively collects data on daily disease severity, iatrogenic events, and nosocomial infections. Each year, each ICU includes a random sample of at least 50 patients who have ICU stays longer than 24 h. Each ICU chooses to obtain this random sample by taking either consecutive admissions to selected ICU beds throughout the year or consecutive admissions to all ICU beds for 1 month.
Study population and definitions
The study was approved by the institutional review board of the Clermont-Ferrand University Hospital and was performed in accordance to the tenets of the Declaration of Helsinki. We included consecutive patients who were older than 18 years at ICU admission and were entered into the database between January 2005 and November 2012. Patients with less than two serum sodium measurements or ICU stays shorter than 72 h were excluded.
The study patients were classified into three groups based on whether serum sodium at ICU admission was normal, defined as between 135 and 145 mmol/L (16), elevated, or decreased. Mild, moderate, and severe hyponatremia were defined as serum sodium less than 135 mmol/L but greater than or equal to 130, less than 130 but greater than or equal to 125, and less than 125 mmol/L, respectively (6). Mild, moderate, and severe hypernatremia were defined as serum sodium greater than 145 but less than or equal to 150, greater than 150 but less than or equal to 155, and greater than 155 mmol/L, respectively (6).
Persistent dysnatremia was any degree of dysnatremia on both day 1 (the day of ICU admission) and day 3. Patients with dysnatremia on day 1 but normal natremia on day 3 were classified as having corrected dysnatremia. Intensive care unit–acquired dysnatremia was the development of any degree of dysnatremia between day 1 and day 3 in patients with normal natremia on day 1 (ICU admission).
The serum sodium correction rate (mmol/L per day) was calculated as follows:
Hypernatremia: [(serum sodium on day 1 – serum sodium on day 3)/time in hours between the two measurements] × 24
Hyponatremia: [(serum sodium on day 3 – serum sodium on day 1)/time in hours between the two measurements] × 24
Data were collected daily by senior physicians and/or specifically trained study monitors in the participating ICUs. For each patient, these investigators entered the data into an electronic case report form using data capture software (RHEA; OutcomeRea) then imported all records into the OutcomeRea database. All codes and definitions were established before study initiation. The data quality checking procedure has been described elsewhere (17). The following information was recorded: age and sex, admission category (medical, scheduled surgery, or unscheduled surgery), and origin (home, ward, or emergency department). Severity of illness was evaluated on the first ICU day using the Simplified Acute Physiology Score II (SAPS II), Sequential Organ Failure Assessment (SOFA) score, and Logistic Organ Dysfunction (LOD) score (18–20). Knaus Scale definitions were used to record preexisting chronic organ failures, including respiratory, cardiac, hepatic, renal, and immunological dysfunctions (21).
Quality of the database
For most of the study variables, the data capture software immediately ran an automatic check for internal consistency, generating queries that were sent to the ICUs for resolution before incorporation of the new data into the database. In each participating ICU, data quality was checked by having a senior physician from another participating ICU review a 2% random sample of the study data on alternate years. A 1-day-long data capture training course held once a year was open to all OutcomeRea investigators and study monitors. All qualitative variables used in the analyses had κ coefficients >0.8, and all variables had inter-rater coefficients in the 0.67 to 1 range, indicating good to excellent reproducibility.
Categorical variables are described as number (%) and continuous variables as median (interquartile range, IQR). For comparisons, we used the chi-square test for categorical data and the Mann-Whitney U test for continuous data.
Factors associated with mortality 28 days after ICU admission by univariate analysis were entered into a multivariate logistic regression model using a forward selection procedure. Factors yielding values of P < 0.05 by univariate analysis were included in the initial multivariate model into a multiple logistic regression model and retained in the final model when associated with values of P < 0.05. In addition to variables related to dysnatremia and serum sodium correction, variables tested in the model were sex, age older than 64 years, SOFA score, symptom of shock at admission, underlying immunodeficiency, fluid therapy more than 50 mL/kg at ICU admission, and vasoactive drugs at ICU admission. Linearity of the logit of quantitative variables was assessed using cubic splines; these variables were left as continuous covariates or transformed into dummy or ordered variables, as appropriate. Two-by-two clinically sound interactions were tested. Dysnatremia correction by day 3 was then introduced into the final model to assess its effect on day 28 mortality.
Last, the influence of the serum sodium correction rate (in millimolars per day) on day 28 mortality was assessed using a logistic regression model adjusted for the previously identified factors independently associated with day 28 mortality.
Values of P < 0.05 were considered significant. Analyses were performed using SAS 9.3 software (SAS Institute, Cary, NC).
Of the 11,073 patients entered into the database during the study period, 7,067 fulfilled our inclusion criteria (Fig. 1). At admission (day 1), 1,830 patients (25.9%) had moderate-to-severe hyponatremia, including 1,187 patients (16.8%) with mild hyponatremia, 435 patients (6.2%) with moderate hyponatremia, and 208 patients (2.9%) with severe hyponatremia. In addition, 634 patients (9.0%) had mild-to-severe hypernatremia including 480 patients (6.8%) with mild hypernatremia, 92 patients (1.3%) with moderate hypernatremia, and 62 patients (0.9%) with severe hypernatremia (Fig. 1).
Overall, characteristics of patients (Table 1) with dysnatremia compared with those with normal serum sodium were older age; female predominance; and greater disease severity as assessed by the SAPS II, SOFA score, or need for life-supporting treatment. In addition, patients with dysnatremia were more often admitted for a medical condition or surgical emergency than patients with normal serum sodium concentration. Overall, 19.0% patients had underlying diabetes mellitus (n = 1,341), including 805 patients without dysnatremia at ICU admission (17.5%), 394 patients with hyponatremia at day 1 (21.5%), and 142 patients with hypernatremia at ICU admission (22.4%) (P = 0.001). Main diagnoses at ICU admission are reported in Table S1 (see Table, Supplemental Digital Content 1, at http://links.lww.com/SHK/A201).
Outcomes according to dysnatremia correction
Correction of hyponatremia by day 3 was achieved in 1,019 patients (55.7% of patients with any degree of hyponatremia), including 36 who developed hypernatremia by day 3 (Fig. 1). Hypernatremia was corrected by day 3 in 393 (62%) patients with any degree of hypernatremia, including one who developed hyponatremia by day 3. Of 4,603 patients with normal serum sodium on day 1, 449 (9.8%) developed ICU-acquired hyponatremia and 263 (5.7%) developed ICU-acquired hypernatremia by day 3.
Day 28 mortality was 17.5% overall, 18.7% in patients with hyponatremia on day 1 (day 28 mortality of 18.0%, 21.6%, and 16.4% in patients with mild, moderate, and severe hyponatremia, respectively) and 24.6% in patients with hypernatremia (day 28 mortality of 24.2%, 23.9%, and 29.0% in patients with mild, moderate, and severe hypernatremia, respectively) on day 1.
Crude day 28 mortality according to serum sodium changes is reported in Figure 1. When entered into a logistic regression model (Table 2), persistent hyponatremia and persistent hypernatremia on day 3 were independently associated with day 28 mortality (odds ratio [OR], 1.30; 95% confidence interval [95% CI], 1.06 – 1.60; and OR, 1.91; 95% CI, 1.40 – 2.60; respectively). Similarly, ICU-acquired hypernatremia by day 3 was independently associated with day 28 mortality (OR, 1.58; 95% CI, 1.17 – 2.15). Conversely, factors not associated with significant differences in adjusted day 28 mortality compared with patients with normal serum sodium on day 1 were hyponatremia corrected by day 3 (OR, 0.88; 95% CI, 0.72 – 1.07), ICU-acquired hyponatremia by day 3 (OR, 1.07; 95% CI, 0.80–1.44), and hypernatremia corrected by day 3 (OR, 0.98; 95% CI, 0.74 – 1.31).
Serum sodium correction rate
Median correction rate between day 1 (ICU admission) and day 3 was 2.58 mmol/L per day (IQR, 0.67 – 4.55) overall, 2.62 mmol/L per day (IQR, 0.77 – 4.68) in patients with hyponatremia on day 1, and 2.44 mmol/L per day (IQR, 0.58 – 4.30) in patients with hypernatremia on day 1. A higher serum sodium correction rate was associated with lower day 28 mortality in both the crude analysis (OR per mmol/L per day, 0.97; 95% CI, 0.94 – 1.00; P = 0.04) and the adjusted analysis (OR per mmol/L per day, 0.93; 95% CI, 0.90 – 0.97; P = 0.0003) (Table 3). The logit linearity assumption was confirmed.
In the subgroup of patients with moderate-to-severe dysnatremia, findings were similar after adjustment for confounders (OR per mmol/L per day, 0.87; 95% CI, 0.82 – 0.95; P < 0.0001).
However, serum sodium correction rate during shorter period (between admission day 1 and day 2) had no influence on adjusted day 28 mortality (OR per mmol/L per day, 0.99; 95% CI, 0.97 – 1.00; P = 0.11).
This large multicenter cohort study suggests that early dysnatremia correction after ICU admission is associated with a lower mortality when compared with patients with persistent dysnatremia. In addition, our results suggest that higher serum sodium correction rates during the 48 h after ICU admission may be associated with better survival.
Recent studies demonstrated that dysnatremia was a common abnormality associated with decreased survival (2, 3, 5, 6, 13, 15). Hyponatremia was far more common than hyponatremia: up to 40% of all hospitalized patients had hyponatremia at admission (22), whereas only 2.5% had moderate-to-severe hypernatremia (23). Thirst is the main mechanism that prevents hypernatremia but is effective only in patients having free access to water. As a result, advanced age and critical illness are risk factors for hypernatremia (14, 23–25). Hypernatremia has been associated with insulin resistance, impaired hepatic lactate clearance, decreased left ventricular contractility, and various neuromuscular abnormalities (9, 26–28). Severe neurological manifestations occur in patients with acute profound hyponatremia (10, 16). In several studies in patients admitted to wards or the ICU, dysnatremia was associated with excess mortality after adjustment for illness severity or case mix (2, 3, 5, 6, 13, 15). In keeping with these findings, dysnatremia was both common and associated with higher day 28 mortality in our study.
Although previous studies consistently found an association between dysnatremia and mortality, there is no evidence to date for a causal relationship between these two events (2, 13, 15). Biological arguments confer plausibility to the dysnatremia-mortality association, but the available data come from cohort studies that did not evaluate the clinical consequences of dysnatremia (2, 3, 5, 6, 13, 15). Thus, the association between dysnatremia and mortality may reflect either a direct deleterious effect of dysnatremia or one or more associations between dysnatremia and unmeasured variables such as case mix, underlying comorbidities, medication use, and reason for admission. However, in our study, the early correction of dysnatremia completely eliminated the excess mortality associated with this abnormality: day 28 mortality in patients whose dysnatremia was corrected within 48 h was not significantly different from that in patients with normal serum sodium concentrations. Furthermore, the serum sodium correction rate was associated with better day 28 survival after adjustment for confounders. These findings, after adjustment for patients’ severity changes during the first 3 days, and together with the dose effect (greater excess mortality in patients with greater dysnatremia severity) might suggest a direct link between dysnatremia and excess mortality (2, 13, 15).
Treatment of dysnatremia is well codified (11). Water restriction in patients with hypo-osmotic hyponatremia and correction of water deficit in patients with hypernatremia are the cornerstones of this electrolyte disturbance management (10, 16, 26). These treatments should be used along with specific treatments related to severity (hypertonic saline in patients with severe symptomatic hyponatremia) or mechanism (correction of extracellular space fluid disorders, eviction of involved drugs, correction of associated calcium or potassium disorder, or pharmacological inhibition of antidiuretic hormone) of dysnatremia (10, 16, 26). Interestingly, recent studies underlined participation of inadequate hypertonic solute infusion on ICU-acquired hypernatremia development (1, 3). Last, although formulae have been proposed to predict changes in serum sodium concentrations, these formulae are poorly calibrated for individual prediction (12, 29). Thus, regular measurements of serum sodium and electrolyte balances remain mandatory to correct dysnatremia, assess efficacy of treatment, and avoid complications of infusion therapy (12).
Our study has several limitations. First, the variables available in our database did not allow us to evaluate the causes of dysnatremia. In particular, we had no information on fluid balance or diuretic therapy before ICU admission. Similarly, confounding variables such as glucose or triglyceride level or explanatory variables such as osmolarity are lacking from our database. In addition, our study lacked the statistical power needed to detect potential adverse neurological effects of rapid dysnatremia correction because only 36 patients had correction rates higher than 12 mmol/L per day. Such effects have usually been reported in patients with profound dysnatremia and a correction rate higher than 12 mmol/L per day (30). Consequently, and in keeping with previous recommendations, we believe that the correction rate, for most of the patients, should not exceed 12 mmol/L per day. Last, our study was not designed to specifically assess the prognostic influence of dysnatremia or serum sodium correction rate in patients with neurological conditions. In these patients, minimal brain volume variations induced by serum sodium changes might lead to dramatic modifications in intracranial pressure or cerebral perfusion (31, 32). Studies are needed to assess this point.
Our findings confirm the high prevalence of dysnatremia at ICU admission and the excess mortality associated with this metabolic disturbance, and they indicate that dysnatremia correction is independently associated with survival, with the effect being greater with faster correction rates of up to 12 mmol/L per day. These results might suggest a direct link between dysnatremia and excess mortality in ICU patients.
The authors thank A. Wolfe, MD, for helping with this manuscript.
This study was performed on behalf of the OutcomeRea study group.
1. Lindner G, Kneidinger N, Holzinger U, Druml W, Schwarz C: Tonicity balance in patients with hypernatremia
acquired in the intensive care unit
. Am J Kidney Dis 54: 674–679, 2009.
2. Funk G-C, Lindner G, Druml W, Metnitz B, Schwarz C, Buaer P, et al.: Incidence and prognosis
of dysnatremias present on ICU admission. Intensive Care Med 36: 304–311, 2010.
3. Hoorn EJ, Betjes MGH, Weigel J, Zietse R: Hypernatraemia in critically ill patients: too little water and too much salt. Nephrol Dial Transplant 23: 1562–1568, 2008.
4. Bennani S-L, Abouqal R, Zeggwagh A-A, Madani N, Abidi K, Zekraoui A, Kerkeb O: [Incidence, causes and prognostic factors of hyponatremia
in intensive care]. Rev Med Interne 24: 224–229, 2003.
5. Lindner G, Funk G-C, Schwarz C, Kneidinger N, Kaider A, Schneeweiss B, Kramer L, Druml W: Hypernatremia
in the critically ill is an independent risk factor for mortality. Am J Kidney Dis 50: 952–957, 2007.
6. Darmon M, Timsit J-F, Francais A, Nguile-Makao M, Adrie C, Cohen Y, Garrousete-Orgeas M, Goldgran-Toledano D, Dumenil AS, Jamali S, et al.: Association between hypernatraemia acquired in the ICU and mortality: a cohort study. Nephrol Dial Transplant 25: 2510–2515, 2010.
7. Stelfox HT, Ahmed SB, Zygun D, Khandwala F, Laupland K: Characterization of intensive care unit
following cardiac surgery. Can J Anaesth 57: 650–658, 2010.
8. Stelfox HT, Ahmed SB, Khandwala F, Zygun D, Shahpori R, Laupland K: The epidemiology
of intensive care unit
-acquired hyponatraemia and hypernatraemia in medical-surgical intensive care units. Crit Care 12: R162, 2008.
9. Kozeny GA, Murdock DK, Euler DE, Hono JE, Scanlon PJ, Bansal VK, Vertuno LL: In vivo
effects of acute changes in osmolality and sodium concentration on myocardial contractility. Am Heart J 109: 290–296, 1985.
10. Adrogué HJ, Madias NE: Hyponatremia
. N Engl J Med 342: 1581–1589, 2000.
11. Pokaharel M, Block CA: Dysnatremia in the ICU. Curr Opin Crit Care 17: 581–593, 2011.
12. Lindner G, Schwarz C, Kneidinger N, Kramer L, Oberbauer R, Druml W: Can we really predict the change in serum sodium levels? An analysis of currently proposed formulae in hypernatraemic patients. Nephrol Dial Transplant 23: 3501–3508, 2008.
13. Darmon M, Diconne E, Souweine B, Ruckly S, Adrie C, Azoulay E, Clec’h C, Garrouste-Orgeas M, Schwebel C, Goldgran-Toledano D, et al.: Prognostic consequences of borderline dysnatremia: pay attention to minimal serum sodium change. Crit Care 17: R12, 2013.
14. Macdonald NJ, McConnell KN, Stephen MR, Dunnigan MG: Hypernatraemic dehydration in patients in a large hospital for the mentally handicapped. BMJ 299: 1426–1429, 1989.
15. Sakr Y, Rother S, Ferreira AMP, Ewald C, Dünish P, Riedemmann N, Reinhart K: Fluctuations in serum sodium level are associated with an increased risk of death in surgical ICU patients. Crit Care Med 41: 133–142, 2013.
16. Kumar S, Berl T: Sodium. Lancet 352: 220–228, 1998.
17. Laupland KB, Zahar J-R, Adrie C, Minet C, Vésin A, Goldgran-Tolendano D, Azoulay E, Garrouste-Orgeas M, Cohen Y, Schwebel U, et al.: Severe hypothermia increases the risk for intensive care unit
-acquired infection. Clin Infect Dis 54: 1064–1070, 2012.
18. Timsit J-F, Fosse JP, Troché G, De Lassence A, Alberti C, Garrouste-Orgeas M, Bornstain C, Adrie C, Cheval C, Chevret S: Calibration and discrimination by daily Logistic Organ Dysfunction scoring comparatively with daily Sequential Organ Failure Assessment scoring for predicting hospital mortality in critically ill patients. Crit Care Med 30: 2003–2013, 2002.
19. Le Gall JR, Lemeshow S, Saulnier F: A new Simplified Acute Physiology Score (SAPS II) based on a European/North American multicenter study. JAMA 270: 2957–2963, 1993.
20. Vincent JL, Moreno R, Takala J, Willatts S, De Mendonça A, Bruining H, Reinhart CK, Suter PM, Thijs LG: The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intensive Care Med 22: 707–710, 1996.
21. Knaus WA, Draper EA, Wagner DP, Zimmerman JE: APACHE II: a severity of disease classification system. Crit Care Med 13: 818–829, 1985.
22. Sedlacek M, Schoolwerth AC, Remillard BD: Electrolyte disturbances in the intensive care unit
. Semin Dial 19: 496–501, 2006.
23. Palevsky PM, Bhagrath R, Greenberg A: Hypernatremia
in hospitalized patients. Ann Intern Med 124: 197–203, 1996.
24. Long CA, Marin P, Bayer AJ, Shetty HG, Pathy MS: Hypernatraemia in an adult in-patient population. Postgrad Med J 67: 643–645, 1991.
25. Borra SI, Beredo R, Kleinfeld M: Hypernatremia
in the aging: causes, manifestations, and outcome. J Natl Med Assoc 87: 220–224, 1995.
26. Adrogué HJ, Madias NE: Hypernatremia
. N Engl J Med 342: 1493–1499, 2000.
27. Lenz K, Gössinger H, Laggner A, Druml W, Grimm G, Scheeweiss B: Influence of hypernatremic-hyperosmolar state on hemodynamics of patients with normal and depressed myocardial function. Crit Care Med 14: 913–914, 1986.
28. Bratusch-Marrain PR, DeFronzo RA: Impairment of insulin-mediated glucose metabolism by hyperosmolality in man. Diabetes 32: 1028–1034, 1983.
29. Liamis G, Kalogirou M, Saugos V, Elisaf M: Therapeutic approach in patients with dysnatraemias. Nephrol Dial Transplant 21: 1564–1569, 2006.
30. Sterns RH, Cappuccio JD, Silver SM, Cohen EP: Neurologic sequelae after treatment of severe hyponatremia
: a multicenter perspective. J Am Soc Nephrol 4: 1522–1530, 1994.
31. Munar F, Ferrer AM, de Nadal M, Poca MA, Pedraza S, Sahuquillo J, Garnacho A: Cerebral hemodynamic effects of 7.2% hypertonic saline in patients with head injury and raised intracranial pressure. J Neurotrauma 17: 41–51, 2000.
32. Tseng M-Y, Al-Rawi PG, Pickard JD, Rasolu FA, Kirkpatrick PJ: Effect of hypertonic saline on cerebral blood flow in poor-grade patients with subarachnoid hemorrhage. Stroke 34: 1389–1396, 2003.