Intracranial neoplasms are the leading cause of pediatric

Transcription

1 J Neurosurg Pediatrics 13: , 2014 AANS, 2014 The incidence of postoperative hyponatremia and associated neurological sequelae in children with intracranial neoplasms Clinical article Cydni N. Williams, M.D., 1 Jennifer S. Belzer, M.D., 1 Jay Riva-Cambrin, M.D., M.Sc., 2 Angela P. Presson, Ph.D., 3 and Susan L. Bratton, M.D., M.P.H. 1 Departments of 1 Pediatrics, 2 Neurosurgery, and 3 Internal Medicine, Division of Epidemiology, University of Utah, Salt Lake City, Utah Object. Intracranial tumors are common pediatric neoplasms and account for substantial morbidity among children with cancer. Hyponatremia is a known complication of neurosurgical procedures and is associated with higher morbidity among neurosurgical patients. The authors aimed to estimate the incidence of hyponatremia, identify clinical characteristics associated with hyponatremia, and assess the association between hyponatremia and patient outcome among children undergoing surgery for intracranial tumors. Methods. This is a retrospective cohort study of children ranging in age from 0 to 19 years who underwent an initial neurosurgical procedure for an intracranial tumor between January 2001 and February Hyponatremia was defined as serum sodium 130 meq/l during admission. Results. Hyponatremia during admission occurred in 39 (12%) of 319 patients and was associated with young age and obstructive hydrocephalus (relative risk [RR] 2.9 [95% CI ]). Hyponatremic patients were frequently symptomatic; 21% had seizures and 41% had altered mental status. Hyponatremia was associated with complicated care including mechanical ventilation (RR 4.4 [95% CI ]), physical therapy (RR 4 [95% CI ]), supplemental nutrition (RR 5.7 [95% CI ]), and infection (RR 5.7 [95% CI ]). Hyponatremic patients had a 5-fold increased risk of moderate or severe disability on the basis of their Pediatric Cerebral Performance Category score at discharge (RR 5.3 [95% CI ]). Obstructive hydrocephalus (adjusted odds ratio [aor] 3.24 [95% CI ]) and young age (aor 0.92 [95% CI ]) were independently associated with hyponatremia during admission. Hyponatremia was independently associated with moderate or worse disability by Pediatric Cerebral Performance Category score at discharge (aor 6.2 [95% CI ]). Conclusions. Hyponatremia was common, particularly among younger children and those with hydrocephalus. Hyponatremia was frequently symptomatic and was associated with more complicated hospital courses. Hyponatremia was independently associated with worse neurological outcome when adjusted for age and tumor factors. This study serves as an exploratory analysis identifying important risk factors for hyponatremia and associated sequelae. Further research into the causes of hyponatremia and the association with poor outcome is needed to determine if prevention and treatment of hyponatremia can improve outcomes in these children. ( Key Words hyponatremia pediatrics neurosurgery intracranial neoplasm oncology Intracranial neoplasms are the leading cause of pediatric cancer related death, and 60% of survivors are left with disabilities that account for significant morbidity. 21 Approximately 4300 children are diagnosed each year with an intracranial neoplasm, 5 and most will Abbreviations used in this paper: AIC = Akaike Information Criteria; aor = adjusted odds ratio; ICD-9-CM = International Classification of Diseases, Ninth Revision, Clinical Modification; IQR = interquartile range; LOS = length of stay; PCPC = Pediatric Cerebral Performance Category; RR = relative risk; VP = ventriculoperitoneal. J Neurosurg: Pediatrics / Volume 13 / March 2014 undergo one or more neurosurgical procedures for diagnosis and treatment. Hyponatremia is a well-recognized complication that can occur following both acute CNS injury and neurosurgery with a variable incidence reported from 4% to 88%, depending on the process. 3 Hyponatremia in children with neurological disorders is associated with longer hospital stays and poor neurological outcome. 1 However, hyponatremia among children undergoing brain tumor resection has been poorly studied. A variety of causes, including the syndrome of inappropriate antidiuretic hormone secretion, cerebral salt wasting, hypotonic fluid administration, and medication 283

2 C. N. Williams et al. side effects, contribute to increased risk of hyponatremia among these children. 10 Hyponatremia is known to cause seizures, encephalopathy, and cerebral edema, which are potential sources of secondary brain injury among neurosurgical patients. 13,18 We previously analyzed an administrative data set to evaluate the incidence and risk factors for hyponatremia among children undergoing malignant intracranial tumor resections. 20 As this data set was de-identified and based on discharge diagnostic codes, we were unable to assess the severity or timing of hyponatremia and did not have access to pathology or radiological reports. Thus, we conducted the current study to validate and expand our previous research. We aimed to estimate the incidence of hyponatremia during initial surgical care and identify clinical characteristics associated with hyponatremia, and we additionally assessed the association between the development of hyponatremia and patient outcome at discharge. Methods We conducted a retrospective cohort study of children 0 19 years old who were admitted for their first neurosurgical procedure for an intracranial neoplasm to determine the incidence and risk factors for developing in-hospital acute hyponatremia, defined as a serum sodium concentration 130 meq/l. We also evaluated differences in hospital course and patient status at discharge for patients with and without hyponatremia. This study was approved by the University of Utah Institutional Review Board, and we were granted a waiver of need for informed consent. Patient admissions to Primary Children s Medical Center in Salt Lake City, Utah, were identified between January 2001 and February 2012 based on a discharge diagnosis code of ICD-9-CM or (International Classification of Diseases, Ninth Revision, Clinical Modification) corresponding to malignant neoplasm of the brain or benign neoplasm of the brain, and potential study subjects were required to have a concurrent procedure code of ICD-9-CM or corresponding to excision or destruction of lesion or tissue of brain or closed/open biopsy of brain. Once identified through administrative data, an independent review of medical records for the cohort was conducted. Only a patient s first admission for tumor resection or biopsy was included, and subjects were excluded if serum sodium concentration was not measured (n = 14) during admission. Electronic and paper records for the remaining 319 encounters were then reviewed to identify demographic and clinical information. Tumor characteristics were recorded from radiological, surgical, and pathology records. Tumor location was reported among 8 identified anatomical locations: frontal, temporal, parietal, occipital, deep brain, ventricles, cerebellum, and brainstem. Tumor location was then grouped based on anatomical region for analyses as follows: 1) cortical, 2) deep brain and ventricles, 3) cerebellum and brainstem, and 4) overlapping (includes those tumors with involvement of at least 2 of the 3 defined regions). Metastatic disease and obstructive hydrocephalus were determined by radiographic reports. Dates and times of low sodium values were used to review records for hyponatremic seizures and altered mental status from hyponatremia. Seizures or periods of decreased mental status were related to hyponatremia only with clinical documentation and a serum sodium concentration 130 meq/l documented within 1 hour of the symptom. Clinical documentation required a medical provider caring for the patient to state in clinical notes that the symptom was caused by hyponatremia. Supplemental nutrition included any use of enteral feeding tubes or parenteral nutrition. Infection was defined as having any of the following: bacteremia, pneumonia, meningitis, or urinary tract infection. Hyperglycemia was defined as glucose 200 mg/dl. Treatment for hyponatremia was defined as receiving fluid restriction, normal saline fluid bolus, hypertonic saline infusion, or enteral salt supplement. The primary outcome at discharge was the Pediatric Cerebral Performance Category (PCPC) score. 6,7 The PCPC scores from recorded physical examinations at discharge were analyzed among the 6 defined categories and then were grouped for analyses as follows: 1) normal or mild disability; and 2) moderate disability, severe disability, coma or vegetative state, and brain death. Discharge disposition was analyzed in groups as follows: 1) home with no care; and 2) home health nursing, transfer to inpatient rehabilitation, transfer to care facility, or death. Hospital cost data were obtained from Intermountain Health Care records and adjusted for inflation using the United States Department of Labor online calculator found at (accessed December 12, 2012). Collected information was then entered into a database utilizing the OpenClinica software ( clinica.com, developed by Isovera, Inc.). Laboratory data for sodium and glucose values were downloaded for identified encounters from the electronic medical record into a spreadsheet, and laboratory data from admissions occurring prior to the use of electronic records were entered into the spreadsheet by hand. Laboratory data were linked with our OpenClinica database via unique identification numbers. Univariate results were expressed as counts and percentages for categorical variables and medians and interquartile ranges (IQR) for continuous variables. Fisher s exact and Wilcoxon rank sum tests were used for bivariate analyses for categorical and continuous variables, respectively. An Armitage-Trend test was used to relate symptomatic hyponatremia to PCPC score. Relative risks (RRs) with 95% CIs were used to report results for pairs of categorical variables. A multivariate logistic regression model was constructed predicting hyponatremia with age and factors associated with hyponatremia in the bivariate analyses. Forward and backward stepwise model selection was performed using the Akaike Information Criteria (AIC), and results from the final model were reported as adjusted odds ratios (aors) with 95% CI. An additional multivariate logistic regression model was constructed predicting poor neurological outcome (PCPC scores of 3 6) with hyponatremia, age, and tumor factors that may confound the relationship between hyponatremia and 284 J Neurosurg: Pediatrics / Volume 13 / March 2014

3 Postoperative hyponatremia in children with intracranial tumors outcome. Stepwise selection with AIC was again used, and the final model was reported as the aor and 95% CI. Significance was defined as p < 0.05 for exploratory analyses, but validation of results was performed using the Bonferroni adjustment for multiple comparisons (p < ). Statistical analyses were conducted in R (version , All tests were 2-tailed. Results Hyponatremia during admission occurred in 39 (12%) of 319 total patients admitted for their initial surgical procedure. One patient had hyponatremia before and after the procedure, and all other cases occurred only postoperatively. Nadir sodium values ranged from 109 to 130 meq/l. Table 1 compares select demographic information and admission characteristics among patients with and without hyponatremia. Patients with hyponatremia were younger. Hyponatremia was significantly associated with a longer hospital length of stay (LOS) and higher hospital costs. Hyponatremia did not vary by admission year. Infratentorial tumors were most common, accounting for 60% of all tumors. The most common types of tumor were juvenile pilocytic astrocytoma, medulloblastoma, other astrocytoma, and ependymoma (29%, 18%, 14%, and 10% of all tumors, respectively). Table 2 presents tumor and procedure characteristics by hyponatremia groups. Patients with hyponatremia were more likely to have malignant tumors (RR 2.2 [95% CI ]) and to have metastatic disease (RR 2.4 [95% CI ]). Tumor locations were grouped based on 3 anatomical regions, and tumors that overlapped multiple regions had a greater than a 3-fold increased risk of hyponatremia compared with tumors confined to the cerebral cortex (RR 3.3 [95% CI ]). Among tumors overlapping multiple regions, most involved the deep brain or ventricles (42 of 46 overlapping tumors). There was no difference in the risk of hyponatremia between patients undergoing biopsy and patients undergoing tumor resection. Nineteen patients underwent 2 neurosurgical procedures during their first admission. One hundred ninety-six (61%) patients were diagnosed with obstructive hydrocephalus on radiological TABLE 1: Select demographic and clinical features of children with and without hyponatremia* Parameter Na >130 meq/l Na 130 meq/l p Value no. of patients age in yrs median IQR race/ethnicity 1 Caucasian 244 (87) 34 (87) Hispanic 28 (10) 5 (13) other/unknown 8 (3) 0 sex 0.17 male 152 (54) 26 (67) female 128 (46) 13 (33) yr of admission (44) 16 (41) (56) 23 (59) payer 0.46 Medicaid 85 (30) 9 (23) private insurance 182 (65) 28 (72) self-pay 13 (5) 2 (5) source of admission 0.23 emergency department 132 (47) 23 (59) other 147 (52) 16 (41) LOS in days <0.001 median IQR hospital cost <0.001 median $25,000 $52,000 IQR $19,000 36,000 $39, ,000 * Values are presented as the number of patients (%) unless noted otherwise. Remains significant after Bonferroni correction for multiple tests (p < ). Race/ethnicity was not recorded for 2 patients. Admission source could not be determined from records for 1 patient. J Neurosurg: Pediatrics / Volume 13 / March

4 C. N. Williams et al. TABLE 2: Tumor characteristics and surgical management in children with and without hyponatremia No. of Patients (%) Parameter Na >130 meq/l Na 130 meq/l RR (95% CI) no. of patients grouped tumor region cerebral cortex 57 (20) 4 (10) reference deep brain or ventricles 97 (35) 16 (41) 2.2 ( ) cerebellum or brainstem 90 (32) 9 (23) 1.4 ( ) overlapping 36 (13) 10 (26) 3.3 ( ) histology* malignant 141 (50) 28 (72) 2.2 ( ) benign 138 (49) 11 (28) reference metastases any 26 (9) 9 (23) 2.4 ( ) brain 5 (2) 3 (8) spinal cord 11 (4) 4 (10) brain & spinal cord 10 (4) 2 (5) neurosurgical procedure biopsy 36 (13) 3 (8) reference resection 243 (87) 36 (92) 1.7 ( ) multiple procedures 14 (5) 5 (13) 2.3 (1 5.3) obstructive hydrocephalus 164 (59) 32 (82) 2.9 ( ) ventriculostomy any 146 (52) 30 (77) 2.7 ( ) preop 17 (6) 7 (18) 2.7 ( ) VP shunt 22 (8) 15 (38) 4.8 ( ) * One patient had histology that was indeterminate on pathology report. Three patients did not have record of the presence or absence of metastases. Could not determine the type of neurosurgical procedure due to missing operative notes for 1 patient. Nineteen patients had more than 1 resection or biopsy, but no more than 2 procedures per patient. Remains significant after Bonferroni adjustment for multiple tests (p < ). im aging. Only 3 patients developed hydrocephalus after their first procedure that was not diagnosed on initial preoperative imaging. Ninety percent of patients with obstructive hydrocephalus had at least 1 ventriculostomy placed, and 18% required ventriculoperitoneal (VP) shunts during their admission. Patients with either obstructive hydrocephalus on imaging or those treated with ventriculostomy had a greater than 2.5-fold increased risk of hyponatremia compared with patients without these findings. Ventriculoperitoneal shunt placement for persistent hydrocephalus was associated with a nearly 5-fold increased risk of hyponatremia (RR 4.8; 95% CI ) compared with children without shunt placement. Among the 39 hyponatremic patients, 8 (21%) had hyponatremic seizures and 16 (41%) had altered mental status ascribed to hyponatremia; 34 (87%) received at least 1 treatment for hyponatremia. Patients with hyponatremia had more complicated hospital courses than those without hyponatremia (Table 3), requiring more mechanical ventilation after surgery, blood product transfusions, inpatient supplemental nutrition, physical or occupational therapy, and speech therapy. Likewise, they had significantly more infections (RR 5.7 [95% CI ]). Five percent of all patients had postoperative strokes on imaging, which were not associated with hyponatremia. The incidence of seizures with a concurrent Na level > 130 meq/l did not differ between groups. Of patients receiving adjunct tumor therapies during this initial admission, there was a significant association with hyponatremia, but only 2 patients who underwent chemotherapy and no patient who underwent radiation therapy had hyponatremia exclusively after these therapies were given. Hyperglycemia concurrent with hyponatremia occurred in 1 patient, but adjusted sodium values did not correct above the hyponatremia threshold. 14 Patients with hyponatremia had higher PCPC scores with worse neurological outcomes than patients without hyponatremia (Table 4). Moderate or severe disability was significantly more likely among hyponatremic patients (RR 5.3 [95% CI ]). Only 1 patient was discharged in a coma or vegetative state. Two patients died during hospitalization, and 1 met criteria for brain death. Similarly, patients with hyponatremia required significantly more supplemental nutrition and ongoing medical care after 286 J Neurosurg: Pediatrics / Volume 13 / March 2014

5 Postoperative hyponatremia in children with intracranial tumors TABLE 3: Adjunct therapies and complications during admission in children with and without hyponatremia No. of Patients (%) Parameter Na >130 meq/l Na 130 meq/l RR (95% CI) no. of patients infection* 17 (6) 15 (38) 5.7 ( ) mechanical ventilation* 50 (18) 22 (56) 4.4 ( ) blood product transfusion* packed red blood cells 62 (22) 23 (59) 4.0 ( ) other products 20 (7) 11 (28) 3.7 ( ) physical or occupational therapy* 138 (49) 32 (82) 4.0 ( ) speech therapy* 78 (28) 26 (67) 4.1 ( ) inpatient nutrition* oral only 250 (89) 19 (49) reference feeding tube or parenteral 30 (11) 20 (51) 5.7 ( ) seizure w/ Na >130 meq/l 14 (5) 2 (5) 1.0 ( ) adjunct tumor therapy chemotherapy* 7 (3) 12 (31) 7.0 ( ) radiation therapy 3 (1) 3 (8) 4.3 ( ) vasopressin administration 4 (1) 4 (10) 4.4 ( ) stroke 12 (4) 4 (10) 2.2 ( ) serum glucose 200 mg/dl 22 (8) 7 (18) 2.2 ( ) * Remains significant after Bonferroni adjustment for multiple tests (p < ). Infection includes meningitis, bacteremia, urinary tract infection, or pneumonia. hospital discharge than those without hyponatremia. Compared with other hyponatremic patients, those with either altered mental status or seizures caused by hyponatremia showed an association with worse PCPC score (p = and 0.008, respectively). Nadir sodium values did not differ between symptomatic and asymptomatic patients. Given the multiple interrelated biological effects of some factors, we constructed a multivariate logistic regression model predicting hyponatremia with age and factors associated with hyponatremia in the bivariate analyses including tumor location, hydrocephalus, histology, and presence of metastases. After stepwise model selection, preoperative obstructive hydrocephalus was the only independent risk factor among all models and portended a greater than 3-fold odds of hyponatremia in the final model (Table 5). Additionally, we constructed a multivariate logistic regression model to determine risk factors predicting worse neurological outcome assessed by PCPC score. We adjusted the models for age, hyponatremia, and tumor factors including tumor location, hydrocephalus, and malignant histology. Hyponatremia and tumor location remained significant independent risk factors for poor outcome (Table 6) with hyponatremia portending a greater than 6-fold increase in odds of moderate or worse disability by PCPC score in the final model (aor 6.2 [95% CI ]). Discussion We found that 12% of children hospitalized at our institution for initial surgical diagnosis and management of J Neurosurg: Pediatrics / Volume 13 / March 2014 intracranial neoplasms had hyponatremia during admission, and many were symptomatic. Hyponatremia was associated with younger age and obstructive hydrocephalus, particularly in those with persistent obstruction requiring shunt placement. Extensive tumors, malignant histology, and metastatic disease are interrelated biological factors associated with hydrocephalus and risk of hyponatremia. Importantly, hyponatremia was associated with more complicated hospital courses and significantly worse neurological outcomes. When adjusted for age and tumor factors, hyponatremia remained an important independent risk factor for moderate or worse disability assessed by PCPC score at hospital discharge. The incidence of hyponatremia in our patients was 12%, which is consistent with previous reports for surgically treated children with brain tumors. 9,20 Our definition of hyponatremia as 130 meq/l has been used in previous reports 2 4,9,10 and provides a more stringent and clinically relevant definition for analyses than the laboratory defined level of 135 meq/l. Additionally, the distribution of tumor location and tumor type are similar to national reports regarding pediatric brain tumor epidemiology. 5 Our patient population was 87% Caucasian, which differs from the general population. However, brain tumors are significantly more common in Caucasians, accounting for 85% of children nationally, 5 and it seems unlikely that race or ethnicity would contribute to the development hyponatremia. We found that hyponatremia was associated with young age, which is likely multifactorial; considerations include varying histology and location of tumors between 287

6 C. N. Williams et al. TABLE 4: Neurological outcomes in children with and without hyponatremia No. of Patients (%) Parameter Na >130 meq/l Na 130 meq/l Relative Risk (95% CI) no. of patients PCPC score at discharge* 1, normal 131 (47) 6 (15) reference 2, mild disability 94 (34) 8 (21) 1.8 ( ) 3, moderate disability 48 (17) 17 (44) 6.0 ( ) 4, severe disability 6 (2) 7 (18) 12.3 ( ) 5, coma or vegetative 0 1 (3) 22.8 ( ) 6, brain death 1 (0.4) 0 0 grouped PCPC scores* normal or mild disability 225 (80) 14 (36) reference moderate, severe, coma, brain death 55 (20) 25 (64) 5.3 ( ) discharge disposition* home w/o care 218 (78) 19 (49) reference other 62 (22) 20 (51) 3.0 ( ) discharge supplemental nutrition* 19 (7) 18 (46) 6.5 ( ) any neurological deficits at discharge* 149 (53) 33 (85) 4.1 ( ) * Remains significant after Bonferroni adjustment for multiple tests (p < ). Other discharge disposition includes transfer to care facility, inpatient rehabilitation, home health nursing care, or death. ages, susceptibility to sodium loss from CSF drainage, or relative immaturity of neural and renal tissues. Previous studies found that seizures after brain tumor resection, 9 hyponatremia after third ventriculostomy, 11 and cerebral salt wasting 8 are all more commonly reported in younger children. Our study supports other findings that hyponatremia is associated with changes in intracranial pressure. 1 We showed an association with hyponatremia and hydrocephalus, and this is consistent with our previous database study showing a greater than 2-fold risk for hyponatremia in children with obstructive hydrocephalus. 20 The exact cause of the association between hydrocephalus and hyponatremia is not known, and we cannot ascertain with our data whether hydrocephalus causes hyponatremia directly or is another marker for severe disease. There was a strong association between hyponatremia and children requiring ventriculostomy or VP shunt placement in this study, and case reports have also suggested CSF drainage as a potential cause of hyponatremia. 19 Obstructive hydrocephalus with intracranial tumors has been linked to long-term adverse outcomes and disability among patients harboring brain tumors. 12,15,16 No studies have evaluated a link between long-term outcome and hyponatremia alone or in association with obstructive hydrocephalus. TABLE 5: Factors associated with the odds of hyponatremia: final multivariate analysis following stepwise regression Factor aor (95% CI) age in yrs 0.92 ( ) metastatic disease 2.20 ( ) preop obstructive hydrocephalus 3.24 ( ) The risk of hyponatremia was not associated with extent of surgical intervention, but hyponatremia was more common among children with tumors that overlapped multiple anatomical regions, likely reflecting greater tumor size. This differs from our previous study in which we found that hyponatremia was more common among children with tumors in the deep brain or ventricles. 20 Ascertainment of tumor location in our initial report was based on ICD-9-CM diagnosis codes, while the current study was able to assign tumor location using multiple sources, including imaging and operative reports. Interestingly, 91% of the tumors classified as overlapping regions in our current analysis involved the deep brain or ventricles. Furthermore, 25% of all tumors overlapped more than one of the 8 identified anatomical locations in this study, which is much higher than that in our previous report (13%). The ascertainment of anatomical involvement is likely more accurate in the current analysis as TABLE 6: Factors associated with the odds of poor outcome: final multivariate analysis following stepwise regression* Factor aor (95% CI) hyponatremia 130 meq/l 6.19 ( ) age in yrs 0.96 ( ) tumor region cortical reference cerebellum or brainstem 2.67 ( ) deep brain or ventricles 2.87 ( ) overlapping 4.09 ( ) * Poor outcome is defined as a PCPC score of 3 6 at hospital discharge compared with scores of 1 and J Neurosurg: Pediatrics / Volume 13 / March 2014

7 Postoperative hyponatremia in children with intracranial tumors it does not rely on discharge coding. Additionally, the prior report had a larger study size and was limited to malignant tumors, which may also account for some differences. Hyponatremia can cause or exacerbate seizures, encephalopathy, and cerebral edema, which are all potential sources of secondary brain injury. 13 Our patients had a 21% incidence of hyponatremic seizure and a 41% incidence of altered mental status ascribed to low sodium based on real-time provider documentation, which are surprisingly high. These values are also likely an underestimate of symptomatic hyponatremia as we required concurrent hyponatremia and provider documentation implicating hyponatremia as the cause in our definitions. Additionally, early symptoms of hyponatremia such as nausea, vomiting, and confusion are nonspecific, overlapping with complaints of neurosurgical patients, and the cause may not be easily distinguished. Similar to our previous database study, hyponatremia was associated with more complicated hospital courses. 20 This study showed that hyponatremia was associated with a longer LOS, higher costs, infection, postoperative mechanical ventilation, supplemental nutrition, and need for specialized therapies. Additionally, this study showed an association with hyponatremia and worse neurological outcomes as evidenced by higher PCPC scores at discharge. The presence of symptomatic hyponatremia was also significantly associated with worse PCPC scores, supporting the association of hyponatremia with higher morbidity and the need for further research. Despite these associations, our study cannot ascertain whether hyponatremia reflects more complicated disease or accounts in part for worse brain injury. Hyponatremia has been linked to complications and worse outcomes in a variety of other populations. Higher mortality associated with hyponatremia has been published among adult surgical intensive care patients, 17 among children awaiting liver transplant, 4 and among hospitalized cancer patients. 2 Al-Zahraa Omar et al. 1 showed that hyponatremia was associated with longer hospital stay and worse neurological outcomes among children with various neurological diseases. Moritz and Ayus 13 reported an increased risk of encephalopathy from hyponatremia among children with neurological diagnoses. Our retrospective study cannot show a causal link between hyponatremia and worse outcome, but given the potential for injury by hyponatremia and the strong association with worse outcome noted in our study, further research in this area is needed. Despite the independent association between poor outcome and hyponatremia noted in our study, the relationship could contain residual confounding from age and tumor factors and likewise may be altered by the severity and duration of hyponatremia. In-depth analysis of these factors should be studied in the future to better characterize the link between hyponatremia and poor outcome. Despite having a relatively large sample size and agreement with previous reports, there are several study limitations to consider. Due to retrospective data collection, accurate ascertainment of some information cannot be ensured, and many of our variables were dependent on J Neurosurg: Pediatrics / Volume 13 / March 2014 complete charting, such as the identification of seizure and altered mental status. We also did not attempt to identify causes of hyponatremia in this study. However, we did note that only 1 patient had hyponatremia prior to their procedure, only 2 patients had hyponatremia exclusively after chemotherapy, no patients had hyponatremia exclusively after radiation therapy, and concurrent hyperglycemia did not affect sodium measurements. Identification of cause was beyond the scope of this initial exploratory analysis but deserves further research in the future to determine risk factors and associations that may differ between the various pathophysiological causes of hyponatremia. We are also unable to ensure the accuracy of our 12% estimate for incidence of hyponatremia or potential associations with complications as sicker patients or those with lower sodium values may be more likely to have repeated sodium levels checked than those with less complicated courses. However, 12% can be considered a minimum estimate. Additionally, given the increased vigilance for hyponatremia in recent years, patients admitted later in our study period may have been more likely to have their sodium level checked or treated as care practices evolved, and the effects of interventions were not assessed by this study. However, given the potential for increased complications and worse outcome with hyponatremia among these patients, it reinforces the need to carefully monitor sodium and urine output in the postoperative period. Conclusions Hyponatremia occurs in approximately 12% of children undergoing initial neurosurgical procedures for intracranial neoplasms. Hyponatremia is commonly symptomatic, with a 41% incidence of altered mental status and a 21% incidence of seizures among hyponatremic patients. Young children and those with obstructive hydrocephalus should be closely monitored. Hyponatremia was associated with more complicated hospital courses and was independently associated with a 6-fold increase in odds of moderate or worse disability assessed by PCPC score at discharge when adjusted for age and tumor factors. Further research is needed to delineate the potential causes of hyponatremia and further assess the link between hyponatremia and worsened neurological outcomes. Prospective studies to develop targeted monitoring and intervention strategies to decrease hyponatremia are needed to determine if this decreases complications and improves outcomes. Disclosure Through the employment of author A.P., this investigation was supported by the University of Utah Study Design and Biostatistics Center, with funding in part from the National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health, through Grant No. 8UL1TR (formerly UL1RR025764). Author contributions to the study and manuscript preparation include the following. Conception and design: Williams, Bratton. Acquisition of data: Williams, Belzer. Analysis and interpretation of data: all authors. Drafting the article: all authors. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of 289

8 C. N. Williams et al. all authors: Williams. Statistical analysis: Williams, Riva-Cambrin, Presson, Bratton. Administrative/technical/material support: Belzer, Presson. Study supervision: Bratton. References 1. Al-Zahraa Omar F, Al Bunyan M: Severe hyponatremia as poor prognostic factor in childhood neurologic diseases. J Neurol Sci 151: , Berghmans T, Paesmans M, Body JJ: A prospective study on hyponatraemia in medical cancer patients: epidemiology, aetiology and differential diagnosis. Support Care Cancer 8: , Bussmann C, Bast T, Rating D: Hyponatraemia in children with acute CNS disease: SIADH or cerebral salt wasting? Childs Nerv Syst 17:58 63, Carey RG, Bucuvalas JC, Balistreri WF, Nick TG, Ryckman FR, Yazigi N: Hyponatremia increases mortality in pediatric patients listed for liver transplantation. Pediatr Transplant 14: , Dolecek TA, Propp JM, Stroup NE, Kruchko C: CBTRUS statistical report: primary brain and central nervous system tumors diagnosed in the United States in Neuro Oncol 14 Suppl 5:v1 v49, 2012 (Erratum in Neuro Oncol 15: , 2013) 6. Fiser DH: Assessing the outcome of pediatric intensive care. J Pediatr 121:68 74, Fiser DH, Long N, Roberson PK, Hefley G, Zolten K, Brodie- Fowler M: Relationship of pediatric overall performance category and pediatric cerebral performance category scores at pediatric intensive care unit discharge with outcome measures collected at hospital discharge and 1- and 6-month follow-up assessments. Crit Care Med 28: , Hardesty DA, Kilbaugh TJ, Storm PB: Cerebral salt wasting syndrome in post-operative pediatric brain tumor patients. Neurocrit Care 17: , Hardesty DA, Sanborn MR, Parker WE, Storm PB: Perioperative seizure incidence and risk factors in 223 pediatric brain tumor patients without prior seizures. Clinical article. J Neurosurg Pediatr 7: , Jiménez R, Casado-Flores J, Nieto M, García-Teresa MA: Cerebral salt wasting syndrome in children with acute central nervous system injury. Pediatr Neurol 35: , Lang SS, Bauman JA, Aversano MW, Sanborn MR, Vossough A, Heuer GG, et al: Hyponatremia following endoscopic third ventriculostomy: a report of 5 cases and analysis of risk factors. Clinical article. J Neurosurg Pediatr 10:39 43, Macedoni-Luksic M, Jereb B, Todorovski L: Long-term sequelae in children treated for brain tumors: impairments, disability, and handicap. Pediatr Hematol Oncol 20:89 101, Moritz ML, Ayus JC: New aspects in the pathogenesis, prevention, and treatment of hyponatremic encephalopathy in children. Pediatr Nephrol 25: , Reddy P, Mooradian AD: Diagnosis and management of hyponatraemia in hospitalised patients. Int J Clin Pract 63: , Reimers TS, Ehrenfels S, Mortensen EL, Schmiegelow M, Sønderkaer S, Carstensen H, et al: Cognitive deficits in longterm survivors of childhood brain tumors: identification of predictive factors. Med Pediatr Oncol 40:26 34, Riva-Cambrin J, Detsky AS, Lamberti-Pasculli M, Sargent MA, Armstrong D, Moineddin R, et al: Predicting postresection hydrocephalus in pediatric patients with posterior fossa tumors. Clinical article. J Neurosurg Pediatr 3: , Sakr Y, Rother S, Ferreira AM, Ewald C, Dünisch P, Riedemmann N, et al: Fluctuations in serum sodium level are associated with an increased risk of death in surgical ICU patients. Crit Care Med 41: , Smith DM, McKenna K, Thompson CJ: Hyponatraemia. Clin Endocrinol (Oxf) 52: , Tobias JD: Cerebrospinal fluid losses through ventricular catheters leading to hyponatremia in two children. South Med J 84: , Williams C, Simon TD, Riva-Cambrin J, Bratton SL: Hyponatremia with intracranial malignant tumor resection in children. Clinical article. J Neurosurg Pediatr 9: , Wilne S, Koller K, Collier J, Kennedy C, Grundy R, Walker D: The diagnosis of brain tumours in children: a guideline to assist healthcare professionals in the assessment of children who may have a brain tumour. Arch Dis Child 95: , 2010 Manuscript submitted July 15, Accepted December 9, The abstract of this work was presented in poster form at the Pediatric Academic Society s annual meeting, Washington DC, May Please include this information when citing this paper: published online January 10, 2014; DOI: / PEDS Address correspondence to: Cydni N. Williams, M.D., Pediatric Critical Care, 295 Chipeta Way, Salt Lake City, UT Cydni.Williams@hsc.utah.edu. 290 J Neurosurg: Pediatrics / Volume 13 / March 2014