Abstract
Objective:
To identify factors associated with in-hospital mortality in neonates and children undergoing continuous electroencephalogram monitoring (cEEG) in the intensive care unit (ICU).
Methods:
We performed a retrospective observational study in patients from birth to 21 years who underwent clinically indicated cEEG in the ICU from 2011 to 2013. The main outcome measure was in-hospital mortality.
Results:
Six-hundred and twenty-five patients (54.2% males) met eligibility criteria, of whom 211 were neonates (55% males, 24.8% premature) and 414 were pediatric patients (53.9% males). Electrographic seizures occurred in 176 (28.2%) patients and status epilepticus (SE) occurred in 20 (11.4%) patients. The time from ICU admission to cEEG initiation was 16.7 (5.1–94.4) hours. Eighty-nine (14.2%) patients [30 (14.2%) neonates, and 59 (14.3%) pediatric patients] died in the hospital. In neonates –after controlling for gender and prematurity— independent factors associated with mortality were prematurity [OR=2.63 (95% CI: 1.06–6.5), p=0.037], presence of SE [OR=8.82 (95% CI: 1.74–44.57), p=0.008], and time from ICU admission to initiation of cEEG [OR=1.002 (95% CI: 1.001–1.004) per hour, p=0.008]. In pediatric patients –after controlling for gender and age— independent factors associated with mortality were the absence of seizures [OR=4.3 (95% CI: 1.40.08–11.90.67, p=0.007], the presence of SE [OR= 7.76 (95% CI: 1.47–40.91), p=0.016], and the time from ICU admission to initiation of cEEG [OR= 1.001 (95% CI: 1.0002–1.001) per hour, p=0.005].
Significance:
Both presence of electrographic SE and time from ICU admission to cEEG initiation were independent factors associated with mortality in neonates and pediatric patients with cEEG in the ICU.
Keywords: Epilepsy, Critical care, Mortality, Outcome, Time
INTRODUCTION
Electrographic seizures are common in the pediatric intensive care unit (ICU) affecting 10–50% of patients undergoing clinically indicated continuous electroencephalogram (cEEG) monitoring1; 2. Around one third of these electrographic seizures are exclusively subclinical and would go undetected without the use of EEG1. The incidence of seizures is higher during the neonatal period than in any other time of life3. In addition, the great majority of neonatal seizures are either clinically subtle or clinically undetectable3; 4. The EEG background provides insight into global neurologic function in patients in whom the physical exam is limited due to sedation or because of the intrinsic limitations of neonatal exam. The use of cEEG monitoring in the ICU is growing exponentially at a pace of approximately 30% per year both in adults5 and children6 and current guidelines emphasize the importance of cEEG as the gold standard to diagnose paroxysmal events in the neonatal ICU7.
Electrographic seizures and electrographic SE are independently associated with outcome in neonates8–11, children1; 2; 9; 10; 12; 13, and adults14; 15. This association is not necessarily causal; the extent to which electrographic seizures produce secondary brain injury --as compared with simply representing a biomarker of more severe acute brain injury— remains unknown16. The association between seizure duration and outcome may be confounded by several factors. Potential confounders such as age and underlying etiology are usually controlled for; however the time from the onset of the acute condition to cEEG monitoring is rarely controlled for or even evaluated as a potential confounder or potential outcome predictor. A series of 49 adults evaluating delay in cEEG as a potential predictor of outcome15 showed that for every hour delay in diagnosis of electrographic seizures, the odds ratio for mortality increased by 0.131 or 13%15. The time to detection and treatment of seizure may be important. A recent series shows that early seizure detection and treatment is associated with a higher likelihood of seizure termination17. Surprisingly, there are no other series evaluating time to cEEG as a potential predictor of outcome.
This study aims to address this gap in knowledge by describing the factors associated with in-hospital mortality in neonates and children undergoing cEEG in the ICU, including delay to initiation of cEEG monitoring.
PATIENTS AND METHODS
Standard protocol approvals, registrations, and patient consents.
This study was approved by the Institutional Review Board at Boston Children’s Hospital.
Study design.
This was a retrospective descriptive study conducted at Boston Children’s Hospital, a tertiary care pediatric hospital. The purpose of this study was to evaluate the factors associated with short-term mortality both in neonates and children who underwent a clinically indicated cEEG in the ICU.
Patients.
There are several intensive care settings in our center: 1) the neonatal ICU, which admits patients with age at admission from birth to under 6 months of age, 2) the cardiac ICU, which admits patients with age at admission from birth to adults, 3) the medical and surgical ICU, which admits patients with age at admission from birth to adults and 4) the medical ICU, which admits patients with age at admission from under one month of age to adults. We did not include patients from the semi-intensive care program units. At our institution a neuroICU consult team covering all pediatric and neonatal ICUs and an epilepsy team including epileptologists and EEG technologists are immediately available at all times. All ICU patients with a primary or secondary neurological condition are evaluated by the neuroICU team. The neuroICU team based on clinical judgement suggests the need for a cEEG to the epilepsy team, the cEEG gets started as soon as possible, and the initial EEG tracing is read within 1 hour by the epilepsy team. Subsequent EEG reads occur every approximately 8 hours or more frequently if deemed necessary by the findings on the initial or subsequent EEG readings. The neuroICU team sometimes evaluates the patient before deciding on the need for an EEG, but if the clinical history suggests that the patient will need a cEEG –like in the case of moderate to severe neonatal hypoxic-ischemic encephalopathy— the cEEG is started before evaluating the patient. At our institution, seizures (both electro-clinical and electrographic-only) detected first on cEEG are actively treated unless in patients with poor prognosis in whom redirection of care to comfort-only measures is the goal of care.
Inclusion criteria were: 1) age from birth –including premature babies— to 21 years, 2) cEEG monitoring in one of the ICU settings (neonatal ICU, cardiac ICU, medical and surgical ICU, and the medical ICU) at Boston Children’s Hospital, 3) ICU admission between January 1st, 2011 to December 31st, 2013. Exclusion criteria were: 1) cEEG in a non-ICU setting, or 2) cEEG in the ICU during a scheduled ICU admission or epilepsy surgery. cEEG initiated in other setting but continued in the ICU were included in the analysis. Critically ill newborns present with different etiologies, evolutions, and outcomes than older critical pediatric patients. Data were analyzed separately for newborns and pediatric patients. Newborns were defined as patients with an age of less than 28 days of life normalized to 40 weeks of gestational age, regardless of gestational age at birth and regardless of specific ICU of admission. We considered prematurity as birth before 37 weeks of gestational age.
EEG data.
EEGs were performed using the international standard 10–20 system of electrode placement. We considered an EEG to be a cEEG when it was recorded uninterruptedly for at least 3 hours, following the threshold initially proposed by the Pediatric Critical Care EEG group6. When there were several ICU admissions with cEEG for the same patient during the study period, only the first one was considered. Electrographic seizures were defined as abnormal, paroxysmal EEG events that differed from the background activity, lasted longer than 10 seconds (unless if associated with clinical signs, in which case they were considered seizures regardless of duration), and evolved in frequency, morphology, and spatial distribution18. For the purposes of this study, we considered electrographic seizures as any seizure detected on EEG, which could be electro-clinical (with a clinical correlate) or electrographic-only (exclusively subclinical)18. Electrographic status epilepticus was defined as a continuous electrographic seizure lasting at least 30 minutes or recurrent electrographic seizures totaling more than 50% of any one-hour epoch18. All EEGs were evaluated clinically and the information extracted from the EEG reports.
Variables.
The main outcome measure was in-hospital mortality. We collected information on factors associated with this outcome including: 1) age, 2) gender, 3) gestational age (only for newborns), 4) etiology, 5) presence of electrographic seizures or electrographic SE during the cEEG monitoring, 6) time from ICU admission to start of cEEG monitoring, and 7) time from ICU admission to first detected seizure. Etiologies were classified as follows: 1) structural if in the presence of clinically relevant MRI abnormality including hypoxic injury and ischemic or hemorrhagic infarction, 2) non-structural if there was no focal abnormality on MRI but there was suspicion of genetic, metabolic, and infectious etiologies, and 3) non-CNS indicating critical illness secondary to a system outside the central nervous system (e.g. cardiac). Study data were collected and managed using REDCap electronic data capture tools hosted at Boston Children’s Hospital. REDCap (Research Electronic Data Capture) is a secure, web-based application designed to support data capture for research studies19.
Statistical analysis.
Demographic and clinical characteristics were summarized with descriptive statistics. On univariate analysis, we evaluated the association between the primary outcome –mortality, a dichotomous outcome— and different factors associated with outcome, with simple logistic regression for continuous variables and with Fisher’s exact test for categorical variables. In multivariable analysis, a complete case analysis logistic regression was performed: the outcome was in-hospital mortalityand the predictors were age, gender, prematurity (only for newborns), presence of electrographic seizures, presence of electrographic SE, and time from ICU admission to cEEG initiation. We considered prematurity as birth before 37 weeks of gestational age. Statistical significance was set at a conventional two-sided alpha value of 0.05. All statistical analyses were performed with STATA 12 (Stata Corp., College Station, TX).
RESULTS
Demographic and clinical features.
Six-hundred and twenty-five patients (54.2% males) met eligibility criteria. Median (p25-p75) age was 0.75 (0.03–7.58) years. There were 211 newborns (55.0% males) who had a postnatal age of 3.95 (0.9–13.1) days and 414 pediatric patients (53.9% males) who had an age of 4.17 (0.8–11.3) years. The time from ICU admission to cEEG initiation was 16.7 (5.1–94.4) hours, 15.1 (4.9–104.8) hours in newborns, and 17.8 (5.1–91.4) hours in pediatric patients (Figure 1). Eighty-nine (14.2%) patients died in-hospital, of which 30 (14.2%) were newborns, and 59 (14.3%) were pediatric patients. Details on the main demographic and clinical features can be found in Table 1.
Figure 1. Time from ICU admission to cEEG monitoring in hours.
Data for newborns appear in green and data for pediatric patients appear in dark red. For clarity the x axis is arbitrarily truncated at 72 hours.
cEEG: Continuous electroencephalogram. ICU: Intensive care unit.
Table 1.
Demographic and clinical features.
| Variable | Outcome measures | Neonatal group (age in days) (N=211) | Pediatric group (age in years) (N=414) | Total (age in years) (N=625) |
|---|---|---|---|---|
| Age (N=625) | Median (p25-p75) | 3.95 (0.91–13.08) | 4.17 (0.75–11.33) | 0.75 (0.03–7.58) |
| Prematurity defined as birth before 37 weeks of gestational age (N=210) | 52 (24.8%) | NA | NA | |
| Gender (N=625) | Male | 116 (55.0%) | 223 (53.9%) | 339 (54.2%) |
| Female | 95 (45.0%) | 191 (46.1%) | 286 (45.8%) | |
| Prior diagnosis of epilepsy (N=416) | 0 (0%) | 147 (35.5%) | 147 (35.3%) | |
| Etiology (N=625) | Symptomatic-structural | 123 (58.3%) | 213 (51.5%) | 336 (53.8%) |
| Symptomatic-non structural | 24 (11.4%) | 117 (28.3%) | 141 (22.6%) | |
| Unknown | 4 (1.9%) | 6 (1.5%) | 10 (1.6%) | |
| Non CNS | 60 (28.4%) | 78 (18.8%) | 138 (22.1%) | |
| Presence of acute trigger for seizures (N=625) | 205 (97.2%) | 396 (95.7%) | 601 (96.2%) | |
| Electrographic seizures (N=625) | 74 (35.1%) | 102 (24.6%) | 176 (28.2%) | |
| Electrographic status epilepticus among patients with electrographic seizures (N=176) | 8 (10.8%) | 12 (11.8%) | 20 (11.4%) | |
| Time from ICU admission to cEEG in hours (N=625) | Median (p25-p75) | 15.1 (4.9–104.8) | 17.8 (5.1–91.4) | 16.7 (5.1–94.4) |
| Time from ICU admission to first detected seizure (N=176) | Median (p25-p75) | 12.9 (6.2–28.4) | 25.5 (9.7–99.9) | 18.7 (7.9–63.5) |
| Time from start of cEEG to first detected seizure in minutes (N=169) | Median (p25-p75) | 98 (25–240) | 91 (27–645) | 97 (26–399) |
| Discharge disposition (N=625) | Alive | 181 (85.8%) | 355 (85.8%) | 536 (85.8%) |
| Dead | 30 (14.2%) | 59 (14.3%) | 89 (14.2%) | |
cEEG: Continuous electroencephalogram. CNS: Central nervous system. ICU: Intensive care unit. NA: Not applicable. P25-p75: 25th percentile to 75th percentile. SD: Standard deviation.
Univariate analysis.
On univariate analysis, the presence of electrographic SE, and the time elapsed from ICU admission to cEEG initiation were associated with increased mortality. Other factors associated with mortality were prematurity (only in newborns) and lack of seizures (only in pediatric patients). Age, gender, etiology, presence of an acute trigger (when present), and time from ICU admission to first detected seizure were not associated with increased mortality (Table 2).
Table 2.
Univariate analysis of factors associated with mortality.
| Newborns (age in days) | Pediatric (age in years) | Global (age in years) | ||||
|---|---|---|---|---|---|---|
| Relevant outcome | p-value | Relevant outcome | p-value | Relevant outcome | p-value | |
| Age | OR=1.02 | 0.088 | OR=0.97 | 0.275 | OR=0.98 | 0.357 |
| Prematurity | Mortality in prematures=25.0% | 0.02* | NA | NA | NA | NA |
| Mortality in term newborns=10.8% | ||||||
| Gender | Mortality in males=16.4% | 0.429 | Mortality in males=14.8% | 0.779 | Mortality in males=15.3% | 0.423 |
| Mortality in females=11.6% | Mortality in females=13.6% | Mortality in females=12.9% | ||||
| Etiology | Mortality in symptomatic-structural=14.6% | 0.956 | Mortality in symptomatic-structural=17.4% | 0.132 | Mortality in symptomatic-structural=16.4% | 0.197 |
| Mortality in symptomatic-non structural=16.7% | Mortality in symptomatic-non structural=8.6% | Mortality in symptomatic-non structural=9.9% | ||||
| Mortality in unknown etiology=0% | Mortality in unknown etiology=0% | Mortality in unknown etiology=0% | ||||
| Mortality in non-CNS etiology=13.3% | Mortality in non-CNS etiology=15.4% | Mortality in non-CNS etiology=14.5% | ||||
| Acute trigger | Mortality in patients with an acute trigger=14.6% | 0.597 | Mortality in patients with an acute trigger=14.7% | 0.490 | Mortality in patients with an acute trigger=14.6% | 0.231 |
| Mortality in patients without an acute trigger=0% | Mortality in patients without an acute trigger=5.6% | Mortality in patients without an acute trigger=4.2% | ||||
| Electrographic seizures | Mortality in electrographic seizures=16.2% | 0.542 | Mortality in electrographic seizures=6.9% | 0.014* | Mortality in electrographic seizures=10.8% | 0.129 |
| Mortality in no electrographic seizures=13.1% | Mortality in no electrographic seizures=16.7% | Mortality in no electrographic seizures=15.6% | ||||
| Electrographic SE | Mortality in electrographic SE=50.0% | 0.02* | Mortality in electrographic SE=25.0% | 0.034* | Mortality in electrographic SE=35.0% | 0.002* |
| Mortality in no electrographic SE=12.1% | Mortality in no electrographic SE=4.4% | Mortality in no electrographic SE=7.7% | ||||
| Time in hours from ICU admission to cEEG | OR=1.0022 | 0.003* | OR=1.0008 | 0.002* | OR=1.0009 | <0.0005* |
| (95% CI: 1.0008 – 1.0037) | (95% CI: 1.0003 – 1.0013) | (95% CI: 1.0004 – 1.0015) | ||||
| Time in hours from ICU admission to detection of the first seizure | OR=1.0033 | 0.393 | OR=1.0007 | 0.328 | OR=1.00044 | 0.514 |
| (95% CI: 0.9955 – 1.0109) | (95% CI: 0.9993 – 1.0021) | (95% CI: 0.9992 – 1.0018) | ||||
cEEG: Continuous electroencephalogram. CI: Confidence interval. CNS: Central nervous system. OR: Odds ratio. SE: Status epilepticus.
*: p<0.05.
Multivariate analysis.
In newborns –after controlling for gender and prematurity— factors associated with mortality were prematurity [OR=2.63 (95% CI: 1.06–6.5), p=0.037], presence of SE [OR=8.82 (95% CI: 1.74–44.57), p=0.008], and time from ICU admission to initiation of cEEG [OR=1.002 (95% CI: 1.001–1.004) or 0.2% (95% CI: 0.1%−0.4%) per hour, p=0.008], but gender, and presence of seizures were not associated with mortality (Table 3). In pediatric patients –after controlling for gender and age— factors associated with mortality were absence of seizures [OR=4.3 (95% CI: 1.5–12.4), p=0.007], presence of SE [OR= 7.76 (95% CI: 1.47–40.91), p=0.016], and time from ICU admission to initiation of cEEG [OR= 1.001 (95% CI: 1.0002–1.001) or 0.1% (95% CI: 0.02%−0.1%) per hour, p=0.005]. Age and gender were not associated with mortality (Table 4). When adding etiology as a categorical variable with 4 categories (symptomatic structural, symptomatic non-structural, unknown, and non-CNS) or as presence or absence of an acute trigger to the logistic regression model, etiology remained non-significantly associated with mortality, and the results did not vary substantially (Tables e-1, e-2, e-3, and e-4).
Table 3.
Logistic regression for newborns with death as the outcome.
| Variables included in the regression | Odds ratio | 95% CI | p-value |
|---|---|---|---|
| Intercept | 0.054 | 0.022–0.132 | NA |
| Male (0:female, 1:male) | 1.611 | 0.668–3.889 | 0.289 |
| Premature (0:term, 1: premature <37 weeks) | 2.626 | 1.061–6.502 | 0.037 |
| Electrographic seizures (0:no seizures, 1:seizures) | 1.351 | 0.506–3.612 | 0.548 |
| Electrographic SE (0:no SE, 1: SE) | 8.815 | 1.743–44.565 | 0.008 |
| Time from ICU admission to start of continuous EEG in hours | 1.002 | 1.001–1.004 | 0.008 |
CI: Confidence interval. NA: Not applicable. OR: Odds ratio. SE: Status epilepticus.
Table 4.
Logistic regression for pediatric patients with death as the outcome.
| Variables included in the regression | Odds ratio | 95% CI | p-value |
|---|---|---|---|
| Intercept | 0.185 | 0.107–0.321 | NA |
| Age in years | 0.986 | 0.939–1.036 | 0.588 |
| Male (0:female, 1:male) | 0.998 | 0.562–1.772 | 0.994 |
| No electrographic seizures (0: seizures, 1: no seizures) | 4.310 | 1.497–12.346 | 0.007 |
| Electrographic SE (0:no electrographic SE, 1:electrographic SE) | 7.755 | 1.470–40.914 | 0.016 |
| Time from ICU admission to start of continuous EEG in hours | 1.001 | 1.0002–1.001 | 0.005 |
CI: Confidence interval. NA: Not applicable. OR: Odds ratio. SE: Status epilepticus.
cEEG and seizures that started before ICU admission.
Fifty patients (16 newborns and 34 pediatric patients) started on cEEG before ICU admission and 13 patients (6 newborns and 7 pediatric patients) started on cEEG before ICU admission. Considering only the 575 patients (195 newborns and 380 pediatric patients) who started cEEG recordings after ICU admission, the results on the prediction value of time to cEEG did not change (Tables e-5 and e-6).
DISCUSSION
In this single-center study we looked at the impact of several clinical and EEG variables on hospital mortality. Our main findings are that, in critically ill neonates and children on clinically indicated cEEG, the time elapsed from ICU admission to cEEG initiation and the presence of electrographic SE are independently associated with increased in-hospital mortality.
Factors associated with outcome for critical patients in the ICU help identify those patients at higher risk for cognitive sequelae and mortality20–22. While factors such as age and –in most cases etiology— are not amenable to intervention, other factors such as time to recognition and treatment are potentially modifiable and may impact other factors such as electrographic seizure duration20–22. There is an association between electrographic seizure burden on cEEG and worse outcomes in terms of morbidity and mortality for neonates8–11, pediatric patients1; 2; 9; 10; 12; 13, and adults14; 15. However, this association is not necessarily causal. Electrographic seizure burden can independently contribute to worse outcomes. However, it is also possible that electrographic seizure duration simply represents a biomarker of a more severe underlying etiology which may not improve with treatment4. Further, while the above-mentioned studies corrected for potential confounders in the association between electrographic seizure burden and outcome such as age, gender, or etiology1; 2; 8–14, they did not take into account the time from presentation to the hospital until cEEG initiation1; 2; 8–14. The time from onset of the critical condition to EEG evaluation might influence prognosis in two ways. First, it might confound the relationship between electrographic seizures and outcome as electrographic seizures may be accruing prior to EEG initiation and represent an additional unrecognized seizure burden. Second, time to diagnosis of an EEG abnormality might be in itself an outcome predictor because delayed diagnoses might imply delayed treatment. The literature has focused preferentially on seizure burden with little attention to time to diagnosis and intervention. In a series of 49 critically ill adults with cEEG monitoring, the only factors independently associated with mortality were seizure duration (OR=1.131 per hour) and delay to diagnosis (OR=1.039 per hour)15. The mortality was 36% when patients were evaluated by EEG and diagnosed with non-convulsive seizures within 30 minutes of onset, 39% when this occurred 1–24 hours after onset, and 75% in those with diagnosis later than 24 hours15. In a study on the impact of a standardized EEG monitoring pathway in critically ill children, 41 patients admitted before pathway implementation were compared with 21 similar patients after the implementation of the pathway17. The median (p25-p75) interval from seizure onset to administration of antiepileptic drugs was shorter in patients treated after the implementation of the pathway [64 (50–101) versus 139 (71–189) minutes]17. Further, the interval from seizure onset to antiepileptic drug administration was also shorter in the patients on the pathway [31 (20–49) versus 71 (33–131) minutes]17. Importantly, seizure termination occurred more frequently following initial antiepileptic drug administration in patients on the EEG monitoring pathway (67% versus 27%)17. Surprisingly, there are no other series which have evaluated the relationship between the time elapsed from onset of the acute condition to cEEG initiation.
Our series addresses this gap in knowledge by evaluating the factors associated with in-hospital mortality, including the time from onset of the critical condition to cEEG initiation. Both in newborns and pediatric patients, seizure burden (represented by the presence of electrographic SE), and time from ICU admission to cEEG monitoring initiation were independent factors associated with mortality. Prematurity was an additional independent factor associated with mortality in newborns with an OR of 2.63. However, the presence of electrographic SE was a stronger factor associated with mortality with an OR of 8.82. In newborns, for each hour delay from ICU admission to cEEG initiation, the mortality OR increased by 0.002. Although this increase might not seem particularly relevant, it means that for each day of delay since ICU admission to cEEG initiation, the mortality OR increased by 0.055 or 5.5%. In older pediatric patients, the lack of seizures was a factor associated with mortality with an OR of 4.3. Although this may appear counterintuitive, it might reflect the inability of severely damaged brains to generate seizures. In older pediatric patients, the presence of electrographic SE increased the mortality OR by 7.76 and for each hour delay since ICU admission to cEEG initiation, the mortality OR increased by 0.001 or by 0.019 or 1.9% per day. Our data indicate that time from ICU admission to initiation of cEEG monitoring is an independent factor associated with short-term mortality both in newborns and older pediatric patients.
The association between time to cEEG and in-hospital mortality provides an additional potentially modifiable risk factor for critical children in the ICU and also gives insight into the association between electrographic SE and mortality. A growing body of literature suggests that electrographic seizures are not only a biomarker of more severe brain injury, but they probably independently contribute to physiological changes, secondary brain injury, and worse outcomes18. Our series shows that the time to cEEG influences short-term mortality, further supporting the hypothesis that seizure burden is an independent factor in outcome2; 18. In addition, our results suggest that a timely identification of electrographic seizures and electrographic SE might not only modify management and potentially decrease electrographic seizure burden23, but also might be associated with better objective outcomes.
Strengths and weaknesses.
The time from ICU admission to cEEG initiation is a surrogate variable. The interval from the onset of the critical illness to the time of first detected electrographic seizure, while perhaps unmeasurable if not undocumentable, may be the ideal variable. The time to first detected electrographic seizure could only be evaluated in patients who had electrographic seizures, who represent approximately 20–30% of our study population. In this subpopulation with electrographic seizures we noted 19 deaths, and the relatively low proportion of death within our cohort likely accounts for lack of a significant association between time to first seizure and in-hospital mortality. However, our data represent an attempt to evaluate the importance of timely cEEG initiation in the ICU, and to tentatively establish this identifiable measure as a potential quality indicator. Our patient population is not necessarily representative of all critically ill children in the ICU or of all children with cEEG in an ICU. Our retrospective analysis may suffer from confounding by indication: the decision of whether and when to start cEEG was not random, but based on the clinical judgment of the medical team. It can be argued that cEEG was started later in patients who were more severe at the beginning and could not initially tolerate cEEG or required more immediate life-saving procedures; under that assumption delay of cEEG could potentially be a surrogate marker for disease severity. Only prospective studies that randomize patients to receiving or not receiving cEEG in the ICU may help resolve this dilemma. The best next approach to a randomized prospective study is to perform a multivariable regression controlling for potential confounders (or surrogate markers of severity). In our multivariable regression model we controlled for prematurity (only for newborns), presence of electrographic seizures, and presence of electrographic SE. As etiology can be a potential confounder of outcome, we performed a multivariable analysis with etiology among the potential confounders (Table e-1, e-2, e-3, and e-4): time from ICU admission to start of cEEG remained highly significant in both pediatric and neonatal patients. Even if results do not have the same level of evidence as a prospective randomized study, our study suggests an effect of time to cEEG after controlling for different markers of severity. We have no data on what causes delays in the time to start cEEG. Future prospective studies of our research group plan to assess the presumed causes of delay such as different availability of technologists at different hours during the day, different degree of clinical suspicion for seizures, or need for more urgent procedures before cEEG or severe patient instability. Our center is a large academic referral center, potentially limiting the generalizability of results. A population large enough to evaluate factors associated with mortality can only be recruited in large centers or in multicenter studies, but there is no reason to believe that factors associated with mortality in our series would be different in smaller centers. A related challenge to generalizability is that patients who undergo cEEG are not a random sample of all patients in the ICU, but probably represent a subpopulation more prone to electrographic seizures. Due to logistical and ethical issues, few data are available regarding the yield of cEEG on patients in the ICU with no clinical suspicion of seizures. In a series of 69 adults with non-traumatic subarachnoid hemorrhage the first 17 patients underwent cEEG for clinical suspicion of electrographic seizures but the following 52% underwent cEEG as part of a protocol regardless of clinical suspicion24. In patients with clinical suspicion electrographic seizures occurred in 17.7%, while in the patients on cEEG per protocol, electrographic seizures occurred in 9.6%; further, among the 35 patients monitored per protocol and without a clinical suspicion for seizures, electrographic seizures occurred in only 8.6%24. These results suggest that the incidence of seizures is higher in patients in which cEEG is placed for a clinical suspicion of electrographic seizures. Our sample is not representative of all patients in the ICU and our results are generalizable to patients who have a clinically indicated cEEG in the ICU. It is reasonable to assume that the frequency of seizures is lower in patients in the ICU who do not undergo cEEG, but only prospective randomized studies will provide a more reliable estimate for these patients.
We divided our population in newborns and pediatric patients. Although it would have been interesting to stratify newborns by degrees of prematurity the numbers did not allow such detail. In our series, etiology was not associated with in-hospital mortality in the univariate or multivariate analyses. This lack of association remained for etiology in a four-item classification: symptomatic structural, symptomatic non-structural, unknown, and non-CNS etiology as well as for a two-item classification: presence or absence of an acute trigger. Although this might seem counterintuitive, several series of neonatal, pediatric, and adult patients with cEEG in the ICU did not find an independent association of etiology with outcome11; 13; 15. One of the major factors that determines the presence of electrographic seizures is the a-priori severity of illness or the degree of clinical suspicion for seizures. Unfortunately, we had no way to determine the severity of illness or the degree of clinical suspicion in the retrospective data analysis.
Our series evaluated an easy to quantify and crude outcome such as in-hospital mortality. There is limited literature on more subtle outcomes, such as cognitive function. A series of 137 children with normal neurodevelopment prior to an acute neurologic condition with admission to the ICU suggests that seizure burden worsens neurobehavioral outcome25. Our results set the stage to include time to cEEG initiation as a factor associated with mortality and a potential predictor of other outcomes, e.g. cognitive or neurological functioning in future studies. Further, our findings suggest that early monitoring may improve outcomes. Current literature not only shows good cost-effective profile for cEEG in the ICU, but also potential for improved outcomes at lower costs26; 27. Major strengths of our series include a large patient population as well as finding of an independent association between timely utilization of cEEG and better outcome.
Future directions.
Much emphasis has been placed on the need to treat electro-clinical and electrographic-only seizures to decrease seizure burden2; 10. However, little attention has been paid to how the timing of treatment affects outcome15. In pediatric convulsive SE, marked delays in treatment administration are common28. To date, timing is not studied as a determinant factor of outcome in most studies of seizures and SE in the ICU1; 2; 8–14. Our results set the stage for future multicenter prospective studies in which time to seizure detection and time to treatment are evaluated in detail as one of the potential predictors of mortality in patients with cEEG in the ICU.
Conclusion.
In critically ill neonatal and pediatric patients both seizure electrographic SE and time from ICU admission to initiation of cEEG monitoring are factors associated with short-term mortality.
Supplementary Material
KEY POINTS.
Critically-ill neonates and children are frequently monitored with continuous electroencephalogram (cEEG).
We retrospectively studied factors associated with in-hospital mortality in 625 patients, 211 neonates, and 414 pediatric patients.
Independent factors associated with mortality in neonates were prematurity, presence of status epilepticus, and time to initiation of cEEG.
Independent factors in pediatric patients were the absence of seizures, the presence of status epilepticus, and time to initiation of cEEG.
ACKNOWLEDGEMENTS
Iván Sánchez Fernández is funded by a grant for the study of Epileptic Encephalopathies from “Fundación Alfonso Martín Escudero” and by the HHV6 Foundation.
Tobias Loddenkemper serves on the Laboratory Accreditation Board for Long Term (Epilepsy and Intensive Care Unit) Monitoring, on the Council (and as 2nd Vice President) of the American Clinical Neurophysiology Society, on the American Board of Clinical Neurophysiology, as an Associate Editor for Seizure, and as an Associate Editor for Wyllie’s Treatment of Epilepsy 6th edition. He is part of pending patent applications to detect and predict seizures and to diagnose epilepsy. He receives research support from the Epilepsy Research Fund, the American Epilepsy Society, the Epilepsy Foundation of America, the Epilepsy Therapy Project, PCORI, the Pediatric Epilepsy Research Foundation, CURE, HHV-6 Foundation, and received research grants from Lundbeck, Eisai, Upsher-Smith, Acorda, and Pfizer. He serves as a consultant for Zogenix, Upsher Smith and Lundbeck. He performs video electroencephalogram long-term and ICU monitoring, electroencephalograms, and other electrophysiological studies at Boston Children’s Hospital and affiliated hospitals and bills for these procedures and he evaluates pediatric neurology patients and bills for clinical care. He has received speaker honorariums from national societies including the AAN, AES and ACNS, and for grand rounds at various academic centers. His wife, Dr. Karen Stannard, is a pediatric neurologist and she performs video electroencephalogram long-term and ICU monitoring, electroencephalograms, and other electrophysiological studies and bills for these procedures and she evaluates pediatric neurology patients and bills for clinical care.
Footnotes
The authors report no potential conflicts of interest.
ETHICAL PUBLICATION STATEMENT
We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.
REFERENCES
- 1.Abend NS, Arndt DH, Carpenter JL, et al. Electrographic seizures in pediatric ICU patients: cohort study of risk factors and mortality. Neurology 2013;81:383–391. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Payne ET, Zhao XY, Frndova H, et al. Seizure burden is independently associated with short term outcome in critically ill children. Brain 2014;137:1429–1438. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.McCoy B, Hahn CD. Continuous EEG monitoring in the neonatal intensive care unit. J Clin Neurophysiol 2013;30:106–114. [DOI] [PubMed] [Google Scholar]
- 4.Abend NS, Wusthoff CJ, Goldberg EM, et al. Electrographic seizures and status epilepticus in critically ill children and neonates with encephalopathy. Lancet Neurol 2013;12:1170–1179. [DOI] [PubMed] [Google Scholar]
- 5.Ney JP, van der Goes DN, Nuwer MR, et al. Continuous and routine EEG in intensive care: utilization and outcomes, United States 2005–2009. Neurology 2013;81:2002–2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Sanchez SM, Carpenter J, Chapman KE, et al. Pediatric ICU EEG monitoring: current resources and practice in the United States and Canada. J Clin Neurophysiol 2013;30:156–160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Shellhaas RA, Chang T, Tsuchida T, et al. The American Clinical Neurophysiology Society’s Guideline on Continuous Electroencephalography Monitoring in Neonates. J Clin Neurophysiol 2011;28:611–617. [DOI] [PubMed] [Google Scholar]
- 8.Glass HC, Glidden D, Jeremy RJ, et al. Clinical Neonatal Seizures are Independently Associated with Outcome in Infants at Risk for Hypoxic-Ischemic Brain Injury. J Pediatr 2009;155:318–323. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Lambrechtsen FA, Buchhalter JR. Aborted and refractory status epilepticus in children: a comparative analysis. Epilepsia 2008;49:615–625. [DOI] [PubMed] [Google Scholar]
- 10.McBride MC, Laroia N, Guillet R. Electrographic seizures in neonates correlate with poor neurodevelopmental outcome. Neurology 2000;55:506–513. [DOI] [PubMed] [Google Scholar]
- 11.Pisani F, Cerminara C, Fusco C, et al. Neonatal status epilepticus vs recurrent neonatal seizures: clinical findings and outcome. Neurology 2007;69:2177–2185. [DOI] [PubMed] [Google Scholar]
- 12.Gwer S, Idro R, Fegan G, et al. Continuous EEG monitoring in Kenyan children with non-traumatic coma. Arch Dis Child 2012;97:343–349. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Topjian AA, Gutierrez-Colina AM, Sanchez SM, et al. Electrographic status epilepticus is associated with mortality and worse short-term outcome in critically ill children. Crit Care Med 2013;41:215–223. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Oddo M, Carrera E, Claassen J, et al. Continuous electroencephalography in the medical intensive care unit. Crit Care Med 2009;37:2051–2056. [DOI] [PubMed] [Google Scholar]
- 15.Young GB, Jordan KG, Doig GS. An assessment of nonconvulsive seizures in the intensive care unit using continuous EEG monitoring: an investigation of variables associated with mortality. Neurology 1996;47:83–89. [DOI] [PubMed] [Google Scholar]
- 16.Abend NS. Electrographic status epilepticus in children with critical illness: Epidemiology and outcome. Epilepsy Behav 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Williams RP, Banwell B, Berg RA, et al. Impact of an ICU EEG monitoring pathway on timeliness of therapeutic intervention and electrographic seizure termination. Epilepsia 2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Abend NS, Chapman KE, Gallentine WB, et al. Electroencephalographic monitoring in the pediatric intensive care unit. Curr Neurol Neurosci Rep 2013;13:330. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.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] [PMC free article] [PubMed] [Google Scholar]
- 20.Legriel S, Mourvillier B, Bele N, et al. Outcomes in 140 critically ill patients with status epilepticus. Intensive Care Med 2008;34:476–480. [DOI] [PubMed] [Google Scholar]
- 21.Power KN, Gramstad A, Gilhus NE, et al. Adult nonconvulsive status epilepticus in a clinical setting: Semiology, aetiology, treatment and outcome. Seizure 2015;24:102–106. [DOI] [PubMed] [Google Scholar]
- 22.Raspall-Chaure M, Chin RF, Neville BG, et al. Outcome of paediatric convulsive status epilepticus: a systematic review. Lancet Neurol 2006;5:769–779. [DOI] [PubMed] [Google Scholar]
- 23.van Rooij LG, Toet MC, van Huffelen AC, et al. Effect of treatment of subclinical neonatal seizures detected with aEEG: randomized, controlled trial. Pediatrics 2010;125:e358–366. [DOI] [PubMed] [Google Scholar]
- 24.O’Connor KL, Westover MB, Phillips MT, et al. High risk for seizures following subarachnoid hemorrhage regardless of referral bias. Neurocrit Care 2014;21:476–482. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Abend NS, Wagenman KL, Blake TP, et al. Electrographic status epilepticus and neurobehavioral outcomes in critically ill children. Epilepsy Behav 2015;49:238–244. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Abend NS, Topjian AA, Williams S. How much does it cost to identify a critically ill child experiencing electrographic seizures? J Clin Neurophysiol 2015;32:257–264. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Abend NS, Topjian AA, Williams S. Could EEG Monitoring in Critically Ill Children Be a Cost-effective Neuroprotective Strategy? J Clin Neurophysiol 2015;32:486–494. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Sánchez Fernández I, Abend NS, Agadi S, et al. Time from convulsive status epilepticus onset to anticonvulsant administration in children. Neurology 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.