Dysnatremia, its correction, and mortality in patients undergoing continuous renal replacement therapy: a prospective observational study

Data on patients undergoing CRRT were obtained prospectively from a cohort from the tertiary referral center (Seoul National University Hospital). The inclusion criteria were as follows: CRRT was needed due to severe AKI; and the patient’s age was 18 years old or older. Accordingly, a total of 669 patients were enrolled at the start of CRRT from May 2010 to September 2013. All of the patients who participated in the present study provided informed consent and agreed with serial examination of serum sodium in adherence to the study protocol. We excluded patients who were previously diagnosed with end-stage renal disease or who received dialysis before enrollment (n = 97). If the patients underwent CRRT more than once (n = 3), only the first experience was counted as a single case. Consequently, 569 patients were analyzed for the present study. Flow diagram for this process is shown in Additional file 1.

Clinical parameters, such as age, sex, weight, cause of AKI, dialysis dose, the need for mechanical ventilation, the use of vasoactive drugs, and underlying chronic kidney disease, were recorded at the start of CRRT. The causes of AKI were divided into sepsis, surgery, nephrotoxin, and others. Information on comorbidities was collected and was presented as Charlson Comorbidity Index [9]. The Acute Physiology and Chronic Health Evaluation (APACHE) ІІ score was calculated to quantitatively assess each patient’s status [10]. Serum sodium, creatinine, and albumin were measured (Toshiba-200FR, Toshiba, Tokyo, Japan), and the urine output data during the first 2 h of CRRT were recorded. Serum sodium was additionally measured at 24 and 72 h after CRRT. Fluid balance before evaluation of later sodium levels (24 and 72 h) was monitored and expressed as follows: fluid balance = total input – total output. There were no missing data for any of the variables. Corrected sodium values for serum glucose were used in the study analyses, based on the following formula: corrected sodium = measured sodium + 0.016 × (serum glucose – 100) [11]. The onsets of dysnatremia were presented as acute (duration < 48 h), chronic (≥ 48 h), or unknown duration. If dysnatremia was identified, it was treated using CRRT with diluted or concentrated substitution fluid to control the speed of correction [12, 13].

The primary outcome was 30-day all-cause mortality after starting CRRT. The mortality data were obtained from the national database of Statistics Korea; all Koreans have a resident registration number and death events are recorded and stored at a national level.

The study protocol complies with the Declaration of Helsinki and received full approval from the institutional review board at the Seoul National University Hospital (no. 1309-018-518). No extramural funding was used to support this work. The authors are solely responsible for the design and conduct of this study, all study analyses and drafting and editing of the paper.

All of the analyses and calculations were performed using STATA software (version 12.0, StataCorp LP, College Station, Texas, USA). The data are presented as the mean ± standard deviations for continuous variables and as proportions for categorical variables. Based on variable distributions using histograms, the variables with non-normal distributions are expressed as the median (interquartile ranges). The chi-squared test was used to compare the categorical variables. Comparisons between normally distributed continuous variables were performed using an analysis of variance or a post-hoc analysis of least significant difference, according to the number of comparison groups. Comparisons between non-normally distributed continuous variables were performed by either the Kruskal-Wallis test or the Mann–Whitney U test, depending on the number of comparison groups. The dose–response relationship (in particular, the non-linear relationship) between sodium and mortality was initially explored by a restricted cubic spline analysis. Due to the U-shaped relationship, we divided the patients into five groups as follows: normonatremia (135–145 mmol/L), mild hyponatremia, moderate to severe hyponatremia, mild hypernatremia, and moderate to severe hypernatremia. The mild and moderate to severe dysnatremia groups were determined based on the median sodium value within each hypo- and hyper-natremia group. Survival curves were drawn using the Kaplan-Meier method. To compare survival curves between groups, the log-rank test was initially applied. The odds ratios (ORs) and 95 % confidence intervals (CIs) for 30-day mortality were calculated using the logistic regression model after adjustment of all covariates (as provided in Table 1). A P value less than 0.05 was considered significant.

The baseline characteristics of the patients are shown and compared between groups with respect to serum sodium level in Table 1. For the 569 subjects, the mean age was 63 years. Except for one Caucasian subject, all other patients were Asian. Approximately half received CRRT because of sepsis. The ranges of sodium levels in dysnatremia groups were as follows: mild hyponatremia, 131.1–134.9 mmol/L; moderate to severe hyponatremia, 115.4–131.0 mmol/L; mild hypernatremia, 145.1–148.4 mmol/L; and moderate to severe hypernatremia, 148.5–166.0 mmol/L. The subjects were followed for a median of 19 days (4 to 193 days).

The AKI causes differed among sodium groups. Hyponatremia groups had higher scores for the Charlson Comorbidity Index than the normonatremia group. The moderate to severe hypernatremia group had higher APACHE II scores than the normonatremia group. The serum creatinine and albumin levels were similar among the sodium groups, except for a higher creatinine level in the mild hyponatremia group. Treatment with hypertonic saline was used in 8 patients, of whom 7 had moderate to severe hyponatremia and 1 had mild hyponatremia. Hypotonic saline was used in 13 patients with mild hypernatremia and 16 patients with moderate to severe hypernatremia.

A total of 57.5 % of the subjects died within 30 days after starting CRRT. Figure 1 shows the relationship between baseline sodium level and the predicted probability of mortality. The relationship between these variables appeared to be U-shaped. Based on this observation, we divided the patients into 5 groups, from hyponatremia to hypernatremia, and we used the normonatremia group as the reference for all of the analyses. The crude 30-day mortality rate was as follows: 50.2 % of those with normonatremia, 62.5 % with mild hyponatremia, 65.8 % with moderate to severe hyponatremia, 51.0 % with mild hypernatremia, and 70.0 % with moderate to severe hypernatremia. Kaplan-Meier survival curves (Fig. 2) show that the survival rates were significantly different (P = 0.016 among the 5 groups, P = 0.024 among the Fig. 1a groups, and P = 0.051 among the Fig. 1b groups). The risks of 30-day mortality according to sodium group are shown in Table 2. In univariate analysis, both hypo- and hypernatremia groups, with the exception of the mild hypernatremia group, had greater ORs for mortality than the normonatremia group. This trend remained significant irrespective of covariates' effects. Similar to these results, the correlation between sodium levels and ORs of mortality after adjustment for all of the covariates was U-shaped (Fig. 3). After initiating CRRT, 542 and 457 patients survived at 24 and 72 h, respectively, and their sodium levels were included in the subsequent analyses. However, sodium levels at 24 and 72 h were not associated with mortality risk (Table 3). Additionally, when adjusted for the sodium level at baseline, the later sodium levels did not show a significant correlation with mortality (all Ps > 0.05).

We further analyzed the risk of mortality according to the change in sodium over 24 or 72 h. The number of 30-day death cases for each group is shown in Additional file 2. First, the non-linear relationships between adjusted ORs for mortality and changes in sodium levels were explored (Fig. 4). In the groups with baseline normonatremia or hyponatremia, the change in sodium over 24 or 72 h did not show an association with the risk of mortality (Fig. 4a to d). In the group with baseline hypernatremia (Fig. 4e and f), a greater decrease in sodium seemed to incur a reduced risk of mortality. However, this trend was shown only in the change over 24 h and not over 72 h. In particular, comparison of mortality rates within three sodium groups, i.e., normonatremia to normonatremia, hyponatremia to hyponatremia, and hypernatremia to hypernatremia, showed no significant difference (Additional file 3). Moreover, when adjusted for the sodium level at baseline, the change in sodium levels did not have any significance with mortality (all Ps > 0.05). These results indicated that sodium change and correction of dysnatremia did not change the mortality risk.