Abstract
Objective
Dysnatremia is common in critically ill children due to disruption of hormonal homeostasis. Children with brain injury are at risk for SIADH, cerebral salt wasting and sodium losses due to externalized ventricular drain (EVD) placement. We hypothesized that among pediatric intensive care unit (PICU) patients managed with an EVD, hyponatremia is common, hyponatremia is associated with seizures and in-hospital mortality, and greater sodium fluctuations are associated with in-hospital mortality.
Design
Retrospective observational study
Setting
Tertiary care PICU
Patients
All pediatric patients treated in the PICU with an EVD from January 2005 to December 2009. Patients were identified by searching the physician order entry database for EVD orders. Hyponatremia was defined as the minimum sodium during patients’ EVD time and was categorized as mild (131-134 meq/L) or moderate-severe (≤130 meq/L). Magnitude of sodium fluctuation was defined as the difference between a patient’s highest and lowest sodium during the time in which an EVD was in use (up to 14 days). Seizure was defined as a clinically evident convulsion during EVD presence. A priori confounders were age, history of epilepsy, and EVD indication. Multivariable regression was performed to test the association between sodium derangements and outcomes.
Interventions
None.
Measurements and Main Results
Three hundred eighty patients were eligible. One hundred nine (29%) had mild hyponatremia, and 30 (8%) moderate-severe hyponatremia. Twenty eight (7%) patients had a seizure while hospitalized. Eighteen patients died (5%) prior to discharge. Survivors had a median daily sodium fluctuation of 1 [0, 5] vs non-survivors 9 [6, 11], (p< 0.001) and a median sodium fluctuation of 5 meq/L [interquartile range 2, 8] vs non-survivors 15 meq/L [9, 24] (p<0.001) during EVD management. After controlling for a priori covariates and potential confounders, hyponatremia was not associated with an increased odds of seizures or in-hospital mortality. However, greater fluctuations in daily sodium (OR 1.38, 95% CI (1.06, 1.8)) and greater fluctuations in sodium during EVD management were associated with increased odds of in-hospital mortality (OR 1.59, 95% CI (1.2, 2.11)).
Conclusion
Hyponatremia was common in PICU patients treated with EVDs but not associated with seizures or in-hospital mortality. Greater sodium fluctuations during EVD management were independently associated with increased odds of in-hospital mortality.
Keywords: hyponatremia, externalized ventricular drain, children, seizures, dysnatremia, pediatrics, mortality
Introduction
Dysnatremia is common in critically-ill pediatric patients due to disruption of hormonal homeostasis (1). Poor outcomes in pediatric patients with extreme sodium derangements developing in the inpatient setting have been reported (2). Hyponatremia is a particularly well-known complication in children with central nervous system disease following neurosurgical procedures due to syndrome of inappropriate antidiuretic hormone (SIADH), cerebral salt wasting, adrenal insufficiency, and iatrogenic causes (diuretics and hypotonic fluids) (3-5). Hyponatremia has been associated with seizures, cerebral edema, encephalopathy and higher morbidity (6, 7). Due to the risk for these and other complications, neurosurgical patients represent an increasing percentage of patients managed in pediatric intensive care units (PICU) (8, 9).
Many of these patients also have externalized ventriculostomy drains (EVD) to manage intracranial hypertension and hydrocephalus associated with traumatic brain injury (TBI), intracranial hemorrhage (ICH), intracranial infections and brain tumors. While patients with EVDs are monitored closely for the associated risks of dislodgement, malfunction, hemorrhagic complications and infections, reported pediatric data is sparse regarding the associations of EVDs with hyponatremia and the important clinical effects of hyponatremia, such as seizures and worse outcomes (10-13). In adult ICU patients, both hyponatremia and large sodium fluctuations are associated with increased hospital mortality (14). However, it is currently unknown whether these electrolyte derangements are associated with mortality in children who are managed with EVDs.
The aims of this study were to identify factors associated with development of hyponatremia and large sodium fluctuations in pediatric patients with EVDs and to evaluate whether these sodium derangements were associated with seizures and in-hospital mortality. We hypothesized that hyponatremia would be common in patients managed with an EVD, and would be associated with increased prevalence of seizures and increased in-hospital mortality. Furthermore, we hypothesized that patients who had large fluctuations in sodium during the time in which they had an EVD would have increased odds of in-hospital mortality.
Materials and Methods
We conducted a retrospective observational study in children with EVDs admitted to the PICU at The Children’s Hospital of Philadelphia between January 1, 2005 and December 31, 2009. This dataset was previously published evaluating the association of EVDs and infection (15). This study was approved by the hospital Institutional Review Board and the requirement for obtaining informed consent was waived.
Patients were identified by searching the physician order entry database for EVD orders. Inclusion criteria included admission to the PICU and the presence of an EVD. The medical records of eligible patients were examined. Study data were collected and managed using Research Electronic Data Capture (REDCap) hosted at The Children’s Hospital of Philadelphia. REDCap is a secure, web-based application designed to support data capture for research studies (16).
Variables collected included baseline demographics (age, gender, race, and weight), EVD indication, history of epilepsy, admission electrolytes, history of diuretic use on admission, clinical seizures during hospitalization, presence of intracranial pressure monitoring, days EVD was present, and survival to hospital discharge. During the study period, there was no standard protocol for EVD management, and it was left to the discretion of the attending neurosurgeon. Daily values were recorded for up to the first 14 days of EVD presence including ventriculostomy output, highest and lowest serum sodium, total daily fluid intake and output, daily volumes of intravenous (IV) 3% saline, 0.9% saline, 0.45% saline, 0.22% saline, total parenteral nutrition (TPN), and administration of enteral sodium chloride supplements. Based on these values, each patient had a daily fluid salt concentration administered calculated.
Based on the lowest serum sodium measured with the EVD in place, hyponatremia was defined as: mild (131-134 meq/L), moderate (125-130 meq/L) and severe (<125 meq/L). (17) Due to small numbers in the severe hyponatremia group, moderate and severe hyponatremia were grouped together for the analysis. Hypernatremia was defined as the patient’s maximum sodium > 145 meq/L. Maximum daily sodium fluctuation was defined as the largest change in sodium in a 24 hour period (highest - lowest sodium (meq/L)) during the first 14 days of the patient’s management with an EVD. Maximum sodium fluctuation during EVD management was defined as the difference between the highest and lowest sodium for a patient during EVD management. Seizure was defined as a clinical convulsion during EVD presence. EEG monitoring for subclinical seizures was not routinely performed during the study period.
Summary statistics are reported as median and interquartile ranges [IQR: 25th percentile, 75th percentile] for continuous data and counts and proportions for categorical data. The association of variables and mortality was performed using Wilcoxon rank sum or Kruskal Wallis for continuous data and chi-squared or Fishers exact test for categorical data. Logistic regression was used to test the association between hyponatremia groups and hospital mortality or PICU seizures and the magnitude of sodium fluctuation and hospital mortality. A priori confounders were EVD indication and history of epilepsy when assessing the association of sodium groups with seizures. Covariates that had a p-value < 0.2 on univariate analysis were sequentially introduced into the model and assessed for possible confounding. Variables introduced into the model that did not change the coefficient of the exposure variable by at least 10% were removed as the likelihood of confounding was low. For analysis of the exposures that were ordinal variable, the likelihood ratio test was performed after addition of each variable if the p-value was <0.05 then the variable remained in the model. If these variables did not affect the association between the exposure and outcome, they were removed from the final model. All statistics were performed using Stata 10 (College Station, Tx).
Results
Three hundred eighty patients were eligible for evaluation. One hundred thirty nine (37%) patients were hyponatremic during EVD presence (Na <135 meq/L): 109 (29%) had mild hyponatremia (Na 131-134 meq/L), and 30 (8%) had moderate or severe hyponatremia (≤130 meq/L)). Two hundred forty one (63%) did not have hyponatremia. Five (1.3%) patients were receiving diuretics on admission. Hyponatremic patients were younger (p= 0.015) and more often received TPN (p<0.001), intravenous three percent saline and oral sodium supplementation (p< 0.001) (Table 1a). Patients with hyponatremia were more likely to have meningitis, traumatic brain injury or spontaneous intracranial hemorrhage (p=0.038). Patients with severe or moderate hyponatremia had EVD in place longer (p<0.001) and had their lowest sodium level later in their EVD course (p=0.027). There was no difference in maximal daily cerebrospinal fluid loss between hyponatremia groups (p=0.38). Only twenty five (7%) patients were hyponatremic (Na< 135 meq/L) on PICU admission.
Table 1.
Moderate and Severe Hyponatremia | Mild Hyponatremia | No Hyponatremia | |||
---|---|---|---|---|---|
Total | Na <130 meq/L | Na 131-134 meq/L | Na >=135meq/L | p-value | |
n =380 | n =30 (8) | n =109 (29) | n =241 (63) | ||
Demographics | |||||
Age (years) | 6.5 [2.1, 13.9] | 3.5 [1.3, 8.7] | 5 [ 1.9, 11.9] | 7.8 [2.3, 14.3] | 0.015 |
Sex | 0.49 | ||||
Male | 169 (44) | 16 (53) | 45 (41) | 108 (45) | |
Female | 211 (56) | 14 (47) | 64 (59) | 133 (55) | |
PRISM score at 24 hours | 319 | 4 [0,9] | 3 [0,5] | 0 [0,5] | 0.01 |
Ventriculostomy Indication | 0.038 | ||||
Tumor | 122 (32) | 7 (23) | 39 (36) | 76 (32) | |
TBI | 17 (5) | 3 (18) | 1 (1) | 13 (5) | |
Spontaneous ICH | 23 (6) | 1 (4) | 9 (8) | 13 (5) | |
Ventricular shunt Externalization/other | 209 (54) | 16 (53) | 58 (53) | 135 (56) | |
Meningitis | 9 (2) | 3 (10) | 2 (2) | 4 (2) | |
Preexisting Epilepsy | 73 (19) | 6 (20) | 21 (19) | 46 (19) | 0.99 |
Admission Sodium | 138 [137, 141] | 136.5 [132,139] | 138 [136,140] | 139 [138, 141] | <0.001 |
Maximum Daily CSF Output (ml) | 279 [168, 374] | 314 [215, 406] | 256 [177, 368] | 282 [142, 346] | 0.38 |
Maximum Sodium Level | 141 [139, 144] | 142 [138, 150] | 140 [138, 143] | 141 [139, 144] | 0.03 |
Magnitude of Maximum Daily Sodium Fluctuation (meq/L) | 2 [0,6] | 8[5, 11] | 4 [0,8] | 0 [0,4] | <0.001 |
Magnitude Sodium Fluctuation | 5 [2.5, 9] | 16 [11, 24] | 7 [4,11] | 3 [1,6] | <0.001 |
Hospital day of lowest Na | 2 [1, 4.5] | 3 [2, 7] | 2 [1,5] | 2 [1,4] | 0.027 |
Interventions | |||||
Normal Saline | 377 (99) | 30 (8) | 109 (29) | 238 (63) | 0.65 |
Three percent Saline | 56 (15) | 11 (37) | 21(19) | 24 (10) | <0.001 |
Quarter NormalSaline | 4 (1) | 1 (3) | 1 (1) | 2 (1) | 0.35 |
Half NormalSaline | 18 (5) | 4 (13) | 3 (3) | 11 (5) | 0.069 |
Total Parenteral Nutrition | 33 (9) | 12 (40) | 9 (8) | 12 (5) | <0.001 |
Oral Sodium Supplementation | 27 (7) | 13 (43) | 12 (11) | 2 (1) | <0.001 |
Days with ventriculostomy | 6 [4,9] | 11 [6, 14] | 7 [4, 11] | 6 [3,8] | <0.001 |
b. The magnitude of patients’ maximum daily sodium fluctuation (difference between highest and lowest sodium each day) and the magnitude of patients’ sodium fluctuation (difference between highest and lowest sodium) during the first 14 days managed with an externalized ventricular drain (EVD). | ||||
N =380 | Magnitude of Maximum Daily Sodium Fluctuation (meq/L) Median [IQR] | Magnitude of sodium fluctuation (meq/L) Median [IQR] | ||
p-value | p-value | |||
Sex | 0.49 | 0.17 | ||
Male | 0 [2,6] | 6 [3,10] | ||
Female | 0 [2,6] | 5 [2,9] | ||
Ventriculostomy Indication | <0.001 | <0.001 | ||
Tumor | 3 [1,7] | |||
TBI | 10 [8,11] | 19 [10, 26] | ||
Spontaneous ICH | 7 [4,8] | 9 [7, 19] | ||
Ventricular Shunt Externalization /other | 0 [0,2] | 4 [2,7] | ||
Meningitis | 4 [0,7] | 9 [6,11] | ||
Preexisting Epilepsy | 0.03 | 0.65 | ||
Yes | 0 [0, 5] | 5 [2,11] | ||
No | 2 [0, 6] | 5 [3,9] | ||
ICP>20 (n=154) | 0.002 | <0.001 | ||
Yes | 4 [0,8] | 10 [6, 19] | ||
No | 3 [0,7] | 6[3,11] | ||
Interventions | ||||
Normal Saline | 0.74 | 0.42 | ||
Yes | 2 [0,6] | 5 [3, 9] | ||
No | 0 [0, 11] | 0 [0, 29] | ||
Three Percent Saline | <0.001 | <0.001 | ||
Yes | 8 [6, 11] | 17 [9.5, 24] | ||
No | 1 [0,4] | 5 [2,7] | ||
Quarter Normal Saline | 0.012 | 0.007 | ||
Yes | 8 [6.5, 11.5] | 15.5 [14. 20.5] | ||
No | 2 [0,6] | 5 [2, 9] | ||
Half Normal Saline | 0.11 | 0.018 | ||
Yes | 6.5 [0,14] | 12.5 [3, 24] | ||
No | 2 [0,6] | 5 [2, 9] | ||
Total Parenteral Nutrition | <0.001 | <0.001 | ||
Yes | 7 [5,9] | 15 [9, 23] | ||
No | 1 [0,5] | 5 [2, 8] | ||
Oral sodium supplementation | <0.001 | <0.001 | ||
Yes | 5 [2,9] | 12 [7, 15] | ||
No | 1 [0,5] | 5 [2, 8] | ||
ICP > 20 (n=164) | <0.001 | <0.001 | ||
Yes | 7 [3,10] | 10 [6, 19] | ||
No | 4 [0,8] | 6 [3, 11] |
CSF: cerebral spinal fluid, ICH: Intracranial hemorrhage, ICP: Intracranial pressure, IQR: interquartile range, ml: milliters, Na: sodium, meq/L: milliequivalents per Liter, TBI: traumatic brain injury, median [IQR]; n(%)
The median maximum daily sodium fluctuation during EVD management (n=380) was 5 [2.5, 9] meq/L. (Table 1b) Twenty percent of patients were hypernatremic (Na >145 meq/L) during their management with an EVD. Patients with traumatic brain injury or more severe hyponatremia had a greater magnitude of sodium fluctuation during EVD management (Table 1a and 1b). Patients who received intravenous quarter normal saline, half normal saline, three percent saline or TPN (standard sodium content 3-4meq/kg/day, not normal saline) or had a clinical seizure also had greater magnitudes of sodium fluctuations (Table 1b). The percentage of patients receiving each type of sodium containing fluid on each day of EVD management is presented in Table 2. There was no difference in the concentration of sodium containing fluids administered on any day compared by hyponatremia groups. However, patients who died were less likely to receive isotonic fluids on days 3, 5, 11 and 14. Quarter normal and half normal saline were infrequently administered.
Table 2.
DAY of EVD | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
N | 380 | 372 | 349 | 300 | 243 | 209 | 173 | 151 | 115 | 100 | 83 | 70 | 59 | 48 |
Percent of patients Receiving Each Sodium Containing Fluid By day of Management | ||||||||||||||
Three percent saline | 35 (9) | 34 (9) | 30 (8) | 29 (10) | 24 (10) | 12 (6) | 11 (6) | 10 (7) | 7 (6) | 6 (6) | 7 (8) | 6 (9) | 3 (5) | 2 (4) |
Quarter Normal Saline | 1 (0.3) | 2 (0.5) | 1 (0.3) | 1 (0.3) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 1 (1.4) | 0 (0) | 0 (0) |
Half Normal Saline | 10 (2.6) | 5 (1.3) | 2 (0.6) | 4 (1.3) | 4 (1.6) | 8 (3.8) | 5 (2.9) | 3 (2) | 1 (0.9) | 2 (2) | 1 (1.2) | 0 (0) | 0 (0) | 0 (0) |
Normal Saline | 366 (96) | 331 (89) | 295 (85) | 238 (79) | 195 (80) | 164 (79) | 135 (78) | 116 (77) | 86 (75) | 71 (71) | 61 (74) | 47 (67) | 41 (70) | 45 (94) |
TPN | 9 (2.4) | 10 (2.7) | 19 (5) | 21 (7) | 23 (10) | 30 (14) | 18 (10) | 12 (8) | 11 (10) | 11 (11) | 8 (10) | 8 (11) | 6 (10) | 4 (8) |
oral sodium | 3 (0.8) | 7 (1.9) | 9 (2.6) | 11 (3.7) | 9 (3.7) | 7 (3.3) | 9 (5.2) | 8 (5.3) | 8 (7) | 9 (9) | 9 (11) | 7 (10) | 9 (15) | 7 (15) |
Alive (N) | 352 | 319 | 279 | 232 | 193 | 163 | 135 | 112 | 80 | 67 | 58 | 47 | 39 | 29 |
Alive | 154 [154,154] | 154 [154,154] | 154 [154,154] | 154 [154,154] | 154 [154,154] | 154 [154,154] | 154 [154,154] | 154 [154,154] | 154 [154,154] | 154 [154,154] | 154 [154,154] | 154 [154,154] | 154 [154,154] | 154 [154,154] |
Dead (N) | 19 | 19 | 18 | 15 | 14 | 13 | 10 | 10 | 10 | 9 | 7 | 7 | 6 | 4 |
dead | 154 [154,208] | 154 [129, 182] | 154 [154,204] | 154 [122,234] | 154 [108, 216] | 154 [94,224] | 154 [87, 236] | 154 [87, 236] | 154 [89, 217] | 154 [100, 187] | 171 [154, 304] | 154 [77, 310] | 128 [77, 295] | 115 [77, 154] |
p-value | 0.14 | 0.47 | 0.047 | 0.1 | 0.016 | 0.95 | 0.69 | 0.54 | 0.59 | 0.78 | 0.01 | 0.47 | 0.43 | 0.009 |
Eighteen patients who received three percent saline during EVD management had hyponatremia (Na<135) on the day of three percent administration. Three of these patients received hypotonic saline or TPN in the 72 hours prior to initiation of three percent saline administration.
Twenty eight patients (7%) had a seizure. Seizures were more frequent in patients with hyponatremia (Na <135 meq/L) (11% vs 5%, p=0.05) (Table 3); however, severity of hyponatremia was not associated with seizures (p=0.058). Furthermore, patients with a sodium > 145 were more likely to have a seizure (p=0.044). Patients with clinical seizures were younger (p=0.025), more likely to have pre-existing epilepsy (p=0.002) and to have received TPN (p =0.013).
Table 3.
No Seizure | Seizure | p-value | survivors | non-survivors | p-value | |
---|---|---|---|---|---|---|
n= 352 | n =28 | n=361 | n=19 | |||
Sodium group | 0.058 | 0.11 | ||||
>= 135 meq/L | 228 (65) | 13 (46) | 231 (64) | 10 (53) | ||
131-134 meq/L | 99 (28) | 10 (36) | 104 (29) | 5 (26) | ||
<130 meq/L | 25 (7) | 5 (18) | 26 (7) | 4 (21) | ||
Sodium Fluctuation | 5 [2,9] | 7.5 [4.5, 13.5] | 0.007 | 5 [2,8] | 15 [9,24] | <0.001 |
Maximum Daily Sodium Fluctuation | 2 [0,6] | 4 [0,8] | 0.077 | 1 [0,5] | 9 [6,11] | <0.001 |
Hypernatremia (sodium > 145 meq/L) | 65 (18) | 10 (36) | 0.044 | 63 (17) | 12 (63) | <0.001 |
Maximum ICU Na | 141 [139, 144] | 143 [139, 148] | 0.061 | 140 [139, 143] | 148 [144, 160 | <0.001 |
Age (years) | 6.7 [2.2, 14] | 2.4 [0.9, 10] | 0.025 | 2 [0.7, 14] | 6 [2.2,12.4] | 0.63 |
Sex | 0.314 | 0.79 | ||||
Male | 154 (44) | 15 (54) | 160 (44) | 89(47) | ||
Female | 198 (56) | 13 (46) | 201 (56) | 10 (53) | ||
Ventriculostomy Indication | 0.008 | 0.06 | ||||
Tumor | 115 (33) | 7 (25) | 117 (32) | 5 (26) | ||
TBI | 16 (5) | 1 (4) | 17 (5) | 0 (0) | ||
Spontaneous ICH | 20 (6) | 3 (11) | 20 (6) | 3 (16) | ||
Externalization of VPS/other | 196 (57) | 13 (46) | 200 (55) | 9(47) | ||
Meningitis | 5 (1) | 4 (14) | 7 (2) | 2 (11) | ||
Preexisting Epilepsy | 61 (17) | 12 (43) | 0.002 | 67 (19) | 6 (32) | 0.16 |
Admission Sodium | 138 [137, 141] | 138.5 [136, 140.5] | 0.69 | 138 [137,141] | 139 [136,142] | 0.59 |
Maximum Daily CSF Output | 282 [161, 373] | 261 [193, 375] | 0.99 | 284 [165, 375] | 226 [170, 367] | 0.37 |
ICP > 20 | 50 (33) | 5 (38) | 0.69 | 52 (34) | 3 (30) | 0.81 |
Days with ventriculostomy | 6 [4,10] | 6.5 [4, 12.5] | 0.27 | 6 [4,10] | 9 [4,15] | 0.09 |
Seizure in hospital | 24 (7) | 4 (21) | 0.04 | |||
Interventions | ||||||
Normal Saline | 349 (99) | 28 (100) | 1.00 | 358 (99) | 19 (100) | 1.00 |
Three Percent Hypertonic Saline | 49 (14) | 7 (25) | 0.111 | 46 (13) | 10 (53) | <0.001 |
Quarter Normal Saline | 3 (1) | 1 (4) | 0.27 | 2 (1) | 2 (11) | 0.013 |
Half Normal Saline | 16 (5) | 2 (7) | 0.63 | 15 (4) | 3 (16) | 0.054 |
Total Parenteral Nutrition | 27 (8) | 6 (21) | 0.013 | 25 (7) | 8 (42) | <0.001 |
Oral sodium Supplementation | 24 (7) | 3 (11) | 0.44 | 26 (7) | 1 (6) | 1.00 |
CSF: cerebrospinal fluid, ICH: intracranial hemorrhage, ICP: intracranial pressure, ICU: intensive care unit, meq/L: milliequivalents
Nineteen patients (5%) died. There was no difference in survival by severity of hyponatremia (p=0.11), however, non-survivors had a significantly greater maximum daily sodium fluctuation (9 meq/L [6, 11] vs 1 meq/L [0, 5], p<0.001) and a significantly greater maximum sodium fluctuation: 15 meq/L [9, 24] vs 5 meq/L [2, 8] (p<0.001) during EVD management. Non-survivors were more likely to have seizures (21% vs 7%, p= 0.04), and were more likely to receive quarter normal saline, half normal saline, three percent saline, or TPN.
After controlling for a priori covariates and potential confounders, severity of hyponatremia subgroup was not associated with seizures or in-hospital mortality (Table 4a). However, the maximum daily sodium fluctuation and the magnitude of sodium fluctuation during the first 14 days of EVD management were associated with increased in-hospital mortality (Table 4b).
Table 4.
Mortality | Seizures | |||||
---|---|---|---|---|---|---|
OR | 95% Confidence Interval | p-value | OR | 95% Confidence Interval | p-value | |
Sodium > =135 meq/L | REF | REF | REF | REF | REF | REF |
131-134 meq/L | 0.55 | 0.16, 1.9 | 0.35 | 1.7 | 0.69, 4.3 | 0.25 |
<130 meq/L | 0.52 | 0.08, 3.2 | 0.48 | 2.6 | 075, 9.4 | 0.13 |
b. Adjusted odds of mortality by for every increase in 3 of maximum daily change in sodium after controlling for three percent administration and total parental nutrition and ventriculostomy indication (TPN. meq/L: milliequivalents per liter) and change in sodium during EVD management after controlling for ventriculostomy indication, lowest sodium group, three percent and total parental nutrition (TPN). | ||||||
In-hospital mortality | ||||||
OR | 95% Confidence Interval | p-value | ||||
Maximum daily Change in Sodium (by 3 meq/L) | 1.38 | 1.06, 1.8 | 0.016 | |||
Change in Sodium (by 3 meq/L) | 1.59 | 1.2, 2.11 | 0.001 |
We performed a sensitivity analysis specifically evaluating the 164 patients who underwent ICP monitoring during EVD management. The association between maximal daily sodium fluctuation and mortality persisted (OR 1.55, 95% CI [1.12, 2.15]), after adjusting for ICP > 20 and EVD indication. ICP> 20 and EVD indication (tumor, TBI, ICH, and VPS) were not associated with mortality in this model. However, patients who had an EVD placed for meningitis had a higher odds of death (OR 8.28, 95%CI [1.16, 58.8], p= 0.035).
Discussion
In this large cohort of children treated with EVDs, greater sodium fluctuations were associated with increased odds of hospital mortality, even after controlling for hypertonic and hypotonic saline administration, initial hyponatremia and EVD indication. EVD associated hyponatremia occurred in 37% of children, but was not associated with seizures or in-hospital mortality. Risk factors for hyponatremia included receiving TPN, having longer duration of EVD management and younger age.
In our cohort, greater fluctuations in sodium over time were associated with increased odds of death, however, admission sodium and severity of hyponatremia during hospitalization were not associated with mortality. These data are important for three reasons. First, understanding mortality associated risk factors and the implications of management decisions allow clinicians to be thoughtful in the frequency of monitoring sodium levels, the choice of type of fluid administered (hypotonic vs. isotonic), and the degree to which patients’ sodium levels are permitted to fluctuate. Secondly, by identifying sodium fluctuations as an independent risk factor for hospital mortality, further exploration of physiologic causes and pathophysiologic impact may help clinicians delineate ways to identify at risk patients earlier with the hopes of decreasing mortality. Thirdly, hyponatremia did not increase the risk for seizures or mortality, suggesting that the EVD may have been protective perhaps by decreasing the risk of cerebral edema.
In our study, the magnitude of patients’ sodium fluctuation during EVD management was significantly associated with in-hospital mortality. The median sodium fluctuation of survivors was 5 meq/L; however, patients who died prior to discharge had a median sodium fluctuation of 15 meq/L, significantly higher than survivors. These differences persisted despite controlling for the severity of a patient’s hyponatremia, underlying neurologic diagnosis and administration of hypertonic and hypotonic intravenous fluids. Patients treated with EVDs represent a specific pediatric ICU population who have underlying neurologic injury or potential disruption of cerebral and hormonal homeostasis.
To evaluate the impact of more rapid sodium changes we evaluated the association of maximum daily sodium change for each patient and found an association with mortality. The greater change in sodium over a 24 hour period may impose more risk to the patient because of cerebral fluid shift that occur rapidly. In response to a hypertonic extracellular space, brain cells have the ability to generate idiogenic osmoles to maintain osmotic balance with the extracellular space without increasing intracellular ion concentrations to a level that may impact enzymatic pathways. Rapid shifts in extracellular sodium concentrations can lead to intracellular fluid shift, cellular edema and potentially worse outcomes. While more rapid and large sodium shifts may be due to underlying conditions, we included EVD indication in our analysis to account for underlying brain injury.
Several studies have found an independent association of hypernatremia with increased mortality after pediatric traumatic brain injury. After controlling for exogenous administration of mannitol and hypertonic saline, the association between hypernatremia and mortality persisted (18, 19). These studies did not specifically look at the change in sodium over time. It is unclear whether hypernatremia or large sodium fluctuations are influencing mortality.
Sakr et al examined the impact of dysnatremia in a very large adult surgical population (10,923 patients) and similar to our results, found that the median sodium fluctuation during the first week of hospitalization was 5 meq/L (14). Furthermore, they also found that a greater sodium fluctuation was associated with increased odds of in-hospital mortality. Our study extends these observations to a pediatric population, all of whom had an underlying condition requiring an EVD. The association between sodium fluctuations and in hospital mortality may be due to underlying severity of illness, severity of hormonal dysregulation, or adverse neurologic effects associated with correction of hyponatremia or hypernatremia.
More than one third of children in our study had documented hyponatremia, however, only 7 % of patients had a sodium < 135 meq/L on admission. The majority of these patients had mild hyponatremia, and only one percent (4 patients) had a documented serum sodium less than 125 meq/L. Hyponatremia has been well described in children following brain tumor resection, cranial vault remodeling and endoscopic third ventriculostomy with prevalence ranging from (8-30%) (3, 4, 20, 21). A case series of three patients reported EVD associated severe dehydration (22). In a series of children who suffered traumatic brain injury, severe hyponatremia < 125 meq/dL was associated with increased mortality (18). Williams et al. evaluated 2,343 patients who underwent brain tumor resection and found hyponatremia occurred in 8.7% of patients (3). The authors noted that hyponatremia was associated with presence of an EVD. The prevalence of hyponatremia was substantially lower in their cohort than in our population and is likely in part due to the authors using an administrative database and ICD-9 codes for hyponatremia and SIADH rather than laboratory values per se. The variability in the prevalence of hyponatremia is likely associated with the underlying condition, fluids administered, frequency of serum sampling, presence of cerebral salt wasting or SIADH and losses from an EVD. The prevalence of hyponatremia in our cohort was higher than in a large study of adult ICU patients (14); however, the populations were inherently distinct with all of our patients having a neurologic indication for EVD placement and the adult population representing patients > 18 years of age of whom more than 70% had cardiothoracic or gastrointestinal surgery, and only 16% had neurosurgery.
Interestingly, when we compared patients who lived and died, the concentration of (mEq/L) sodium in intravenous fluids administered to these groups significantly differed at several points in time (Table 2). While the median concentration was similar between groups, patients who died received a wider range of sodium concentrations (84- 310 mEq/L/day). This observation is interesting because it has been well documented that children who receive hypotonic maintenance fluids are more likely to develop hyponatremia and severe hyponatremia (23). Hospitalized children are at higher risk for elevations in arginine vasopressin, especially when administered hyponatremic fluids (24). Our calculation of total daily administered sodium concentration was reflective of all intravenous fluids administered. It is possible that patients who died received a wider range of sodium concentrations because of required therapies, such as hypertonic saline or total parenteral nutrition, or that the total number of patients who died was so small that these wide ranges were more apparent because of the sample size. Regardless, the wide body of literature supporting the risks associated with intravenous hypotonic fluids should caution against their use in this at risk population.
After controlling for ventriculostomy indication, age and pre-existing epilepsy, hyponatremia was not associated with seizures or hospital mortality. Several small pediatric studies have described the association of hyponatremia and seizures in patients who have neurologic illness and surgery (4, 21). The lack of association with seizures may be due to frequent serum sodium evaluations, prompt treatment of hyponatremia in the PICU or the presence of an EVD drain that may have prevented a rapid rise in intracranial pressure due to CSF drainage.
In our population, admission serum sodium levels were significantly lower in patients who had moderate hyponatremia during their EVD management; however, admission sodium was not associated with in-hospital mortality. Previous studies have evaluated admission hyponatremia as a risk factor for in-hospital mortality. In a large retrospective evaluation of 13 adult ICUs, mild hyponatremia on ICU admission was not independently associated with 30-day mortality (OR 1.08 (0.95-1.22)), but moderate and severe hyponatremia were [moderate: OR 1.18 (1.002, 1.4); severe (OR1.27 (1.01, 1.6)] (25). An even larger study of 151,486 patients in 77 adult medical and surgical ICUs showed that any hyponatremia on admission was independently associated with increased mortality (14, 26). It is not clear whether the worse outcomes were due to the effects of more severe hyponatremia or that more severe hyponatremia was associated with greater severity of the underlying illness. Nevertheless, these populations are distinct from our cohort in age, underlying medical condition and presence of EVD which may explain the differences in our findings.
Our study had several limitations. This data was collected retrospectively and therefore the documented sodium levels were determined by variable routine clinical care and not at set time intervals. We also did not have data regarding the hemodynamic status of patients and ICP measurements were only available on a subset of patients. Magnitude of sodium fluctuations were evaluated as absolute numbers without directionality. While this cohort is large for a pediatric center, our overall sample size was small for patients with more severe hyponatremia, and therefore we may have been underpowered to show a difference between patients with and without seizures, especially in the moderate and severe hyponatremia groups. Systematic continuous EEG monitoring was not performed during this time and therefore only clinical convulsions were documented and subclinical seizures may have been missed. Lastly, because these data were collected retrospectively one cannot conclude causality between larger maximum daily sodium changes and mortality. Further prospective evaluation would need to be performed to determine if this association persists when accounting for other potential confounders, such as the underlying diagnosis, the severity of illness and hemodynamic perturbations.
Conclusion
Hyponatremia is common in children treated with EVDs, but is not associated with increased odds of seizures or in-hospital mortality. Larger magnitude of sodium fluctuation is independently associated with increased in-hospital mortality. Further characterization of the etiologies and pathophysiologic effects of sodium fluctuations on the risk of mortality is warranted.
Acknowledgments
Dr. Topjian is funded by NIH K23-NS075363
Dr. Huh is funded by NICHD R01 HD061963
Dr. Abend is funded by NINDS NS049453
Dr. Friess K08-NS064051
References
- 1.Moritz ML, Ayus JC. Preventing neurological complications from dysnatremias in children. Pediatr Nephrol. 2005;20:1687–1700. doi: 10.1007/s00467-005-1933-6. [DOI] [PubMed] [Google Scholar]
- 2.Dunn K, Butt W. Extreme sodium derangement in a paediatric inpatient population. J Paediatr Child Health. 1997;33:26–30. doi: 10.1111/j.1440-1754.1997.tb00986.x. [DOI] [PubMed] [Google Scholar]
- 3.Williams C, Simon TD, Riva-Cambrin J et al. Hyponatremia with intracranial malignant tumor resection in children. J Neurosurg Pediatr. 2012;9:524–529. doi: 10.3171/2012.1.PEDS11465. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Hardesty DA, Kilbaugh TJ, Storm PB. Cerebral salt wasting syndrome in post-operative pediatric brain tumor patients. Neurocrit Care. 2012;17:382–387. doi: 10.1007/s12028-011-9618-4. [DOI] [PubMed] [Google Scholar]
- 5.Hardesty DA, Sanborn MR, Parker WE et al. Perioperative seizure incidence and risk factors in 223 pediatric brain tumor patients without prior seizures. J Neurosurg Pediatr. 2011;7:609–615. doi: 10.3171/2011.3.PEDS1120. [DOI] [PubMed] [Google Scholar]
- 6.Al-Zahraa Omar F, Al Bunyan M. Severe hyponatremia as poor prognostic factor in childhood neurologic diseases. J Neurol Sci. 1997;151:213–216. doi: 10.1016/s0022-510x(97)00133-0. [DOI] [PubMed] [Google Scholar]
- 7.Smith DM, McKenna K, Thompson CJ. Hyponatraemia. Clin Endocrinol (Oxf) 2000;52:667–678. doi: 10.1046/j.1365-2265.2000.01027.x. [DOI] [PubMed] [Google Scholar]
- 8.Bell MJ, Carpenter J, Au AK, et al. Development of a pediatric neurocritical care service. Neurocrit Care. 2009;10:4–10. doi: 10.1007/s12028-008-9061-3. [DOI] [PubMed] [Google Scholar]
- 9.Spentzas T, Escue JE, Patters AB, et al. Brain tumor resection in children: neurointensive care unit course and resource utilization. Pediatr Crit Care Med. 2010;11:718–722. doi: 10.1097/PCC.0b013e3181d907fa. [DOI] [PubMed] [Google Scholar]
- 10.Berger A, Weninger M, Reinprecht A, et al. Long-term experience with subcutaneously tunneled external ventricular drainage in preterm infants. Childs Nerv Syst. 2000;16:103–109. doi: 10.1007/s003810050022. discussion 110. [DOI] [PubMed] [Google Scholar]
- 11.Anderson RC, Kan P, Klimo P, et al. Complications of intracranial pressure monitoring in children with head trauma. J Neurosurg. 2004;101:53–58. doi: 10.3171/ped.2004.101.2.0053. [DOI] [PubMed] [Google Scholar]
- 12.Woernle CM, Burkhardt JK, Bellut D, et al. Do iatrogenic factors bias the placement of external ventricular catheters?--a single institute experience and review of the literature. Neurol Med Chir (Tokyo) 2011;51:180–186. doi: 10.2176/nmc.51.180. [DOI] [PubMed] [Google Scholar]
- 13.Ngo QN, Ranger A, Singh RN, et al. External ventricular drains in pediatric patients. Pediatr Crit Care Med. 2009;10:346–351. doi: 10.1097/PCC.0b013e3181a320cd. [DOI] [PubMed] [Google Scholar]
- 14.Sakr Y, Rother S, Ferreira AM, et al. Fluctuations in serum sodium level are associated with an increased risk of death in surgical ICU patients. Crit Care Med. 2013;41:133–142. doi: 10.1097/CCM.0b013e318265f576. [DOI] [PubMed] [Google Scholar]
- 15.Topjian AA SA, Pabalan AA, Clair A, Kilbaugh TJ, Abend NS, Berg RA, Heuer GG, Storm PB, Huh JW, Friess SH. Risk factors associated with infections and need for permanent cerebrospinal fluid diversion in pediatric intensive care patients with externalized ventricular drains. Neurocrit Care. 2014 doi: 10.1007/s12028-013-9946-7. in press. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Harris PA, Taylor R, Thielke R et al. Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42:377–381. doi: 10.1016/j.jbi.2008.08.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Darmon M, Timsit JF, Francais A, et al. Association between hypernatraemia acquired in the ICU and mortality: a cohort study. Nephrol Dial Transplant. 2010;25:2510–2515. doi: 10.1093/ndt/gfq067. [DOI] [PubMed] [Google Scholar]
- 18.Bahloul M, Chaari A, Chabchoub I, et al. Outcome analysis and outcome predictors of traumatic head injury in childhood: Analysis of 454 observations. J Emerg Trauma Shock. 2011;4:198–206. doi: 10.4103/0974-2700.82206. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Alharfi IM, Stewart TC, Kelly SH, et al. Hypernatremia is associated with increased risk of mortality in pediatric severe traumatic brain injury. J Neurotrauma. 2013;30:361–366. doi: 10.1089/neu.2012.2410. [DOI] [PubMed] [Google Scholar]
- 20.Cladis FP, Bykowski M, Schmitt E, et al. Postoperative hyponatremia following calvarial vault remodeling in craniosynostosis. Paediatr Anaesth. 2011;21:1020–1025. doi: 10.1111/j.1460-9592.2011.03605.x. [DOI] [PubMed] [Google Scholar]
- 21.Lang SS, Bauman JA, Aversano MW, et al. Hyponatremia following endoscopic third ventriculostomy: a report of 5 cases and analysis of risk factors. J Neurosurg Pediatr. 2012;10:39–43. doi: 10.3171/2012.4.PEDS1222. [DOI] [PubMed] [Google Scholar]
- 22.Simpson S, Yung M, Slater A. Severe dehydration and acute renal failure associated with external ventricular drainage of cerebrospinal fluid in children. Anaesth Intensive Care. 2006;34:659–663. doi: 10.1177/0310057X0603400502. [DOI] [PubMed] [Google Scholar]
- 23.Wang J, Xu E, Xiao Y. Isotonic versus hypotonic maintenance IV fluids in hospitalized children: a meta-analysis. Pediatrics. 2014;133:105–113. doi: 10.1542/peds.2013-2041. [DOI] [PubMed] [Google Scholar]
- 24.Moritz ML, Ayus JC. Hospital-acquired hyponatremia--why are hypotonic parenteral fluids still being used? Nat Clin Pract Nephrol. 2007;3:374–382. doi: 10.1038/ncpneph0526. [DOI] [PubMed] [Google Scholar]
- 25.Darmon M, Diconne E, Souweine B, et al. Prognostic consequences of borderline dysnatremia: pay attention to minimal serum sodium change. Crit Care. 2013;17:R12. doi: 10.1186/cc11937. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Funk GC, Lindner G, Druml W, et al. Incidence and prognosis of dysnatremias present on ICU admission. Intensive Care Med. 2010;36:304–311. doi: 10.1007/s00134-009-1692-0. [DOI] [PubMed] [Google Scholar]