- Can you share the cost of continuous EEG monitoring in the postoperative period?
EEG monitoring is resource intense and is costly as it involves expenses related to equipment, technicians and physician interpretation. At CHOP we have 24-hour in house technicians so that seizures can be identified and treated as soon as possible. The cost of continuous EEG monitoring varies widely between institutions and is also dependent not only on variable hospital costs but also on payer reimbursement.
- Would it be more cost-effective to simply monitor those patients who were determined to be most at-risk, i.e. by univariate or multivariate analyses in your paper?
By univariate analysis we found a higher risk in younger children (age 3 days), children with class IV defects, (single ventricle with aortic arch obstruction), children who had delayed sternal closure, children with longer duration of DHCA (median 47 minutes), ECMO and cardiac arrest. By multivariable analysis children with delayed sternal closure and DHCA were determined to be at higher risk, both these risk factors were strongly associated. However, children with all categories of congenital heart disease and those who did not have DHCA and delayed sternal closure had seizures and the seizure burden was high (62% with status epilepticus). Given that the majority of seizures were subclinical and the long term outcome results of the Boston Circulatory arrest showing that seizure occurrence was the most important predictor of long term outcome we decided to continue to monitor all neonates.
- Does CHOP continue to routinely perform post-operative EEGs in newborns?
Yes for now we continue to monitor all neonates however I am currently in the process of looking at data from December 2013 to December 2015 and hope that we can determine more concrete risk factors so we can monitor only the highest risk patients.
- Is there any evidence to support routine pre-operative imaging such as MRI in the complex neonatal CHD population as a means of risk stratify these patients at higher risk of post-operative seizures; a more directed selection for post-operative EEG?
This is an excellent question. There was no published data that I could identify that associated pre-operative brain injury on MRI with post-operative seizures. This would be a great question to address in institutions that routinely perform pre-operative MRI (at CHOP this is only done as part of a research study) and post-operative EEG monitoring.
- Although you did not report on pre-operative imaging in your study, do you have any data in these patients to suggest an association between MRI findings and seizures?
We do not routinely perform pre-operative MRI imaging on our patients. 6/13 neonates had pre-operative MRIs as part of a research study on pre and post-operative brain injury in neonates with congenital heart disease. Pre-operative brain MRI showed bilateral periventricular leukomalacia and white matter injury with small posterior subdural hemorrhages in 2, an acute infarct involving the posterior frontal and anterior parietal region with brain immaturity in 1, brain immaturity in 1 and a normal MRI in 1.
- Does your data support a decrease in the duration of EEG monitoring, as a way to reduce cost and resource needs?
The median seizure onset was 20 hours (IQR 15, 34) after return to the CICU postoperatively indicating that seizure onset occurred up to 34 hours after initiation of monitoring.
In order to see if we could shorten duration of monitoring and determine EEG characteristics that predict seizure occurrence based on the initial hour on EEG monitoring we performed a regression analysis of EEG variables known in the initial hour of EEG recording including background category, excessive inter-ictal epileptiform discharge presence, and reactivity. The EEG background category was continuous is 30 neonates (19%), appropriately discontinuous in 31 neonates (19%), and excessively discontinuous in 99 neonates (62%). EEG background category did not predict seizure occurrence on univariable analysis (p=0.85).
- Do you feel your data is sufficient to justify a prospective clinical trial? If so, do you believe it is feasible and what would be the specific aim?
The big question is whether neonates can be randomly assigned to treatment of subclinical seizures vs. no treatment and whether treating them makes a difference in terms of long-term neurodevelopmental outcomes.
From the 16-year outcomes of the Boston Circulatory Arrest we learnt that post-operative seizure occurrence was the medical variable most consistently related to worse neuropsychological outcomes, including lower scores on reading and math composites, general memory index, executive function, and visual spatial testing. In that study only clinical seizures were treated with anti-seizure medications. Most seizures were EEG-only seizures and were untreated, which may have contributed to these unfavorable outcomes. It is important to note that these findings were at 16-year outcome and not in the earlier outcome papers.
Although further research is warranted I don’t feel it is feasible, as I don’t feel it is ethical randomize neonates to treatment vs. no treatment of subclinical seizures.
- Operationalization of routine EEG monitoring in many centers would be challenging. Do you have any recommendations regarding prioritization given limited resources?
Based on our data and others as well clinical experience premature infants, DHCA, delayed sternal closure and patients undergoing the Norwood operation seem to be at highest risk. Other risk factors include cardiac arrest and ECMO.
- You recently did a brief survey of centers inquiring to the routine use of post-operative EEG, it was nearly unanimous that centers reserve EEGs for post-arrest and clinical concern for neurologic injury. Do you feel a change in practice is warranted?
EEG monitoring is warranted as our results (predominantly subclinical seizures) and those of the Boston Circulatory arrest (seizure occurrence being the most important predictor of long term outcome) are compelling data. The costs of this resource intense technology have to be weighed with the long-term costs of creating a population where seizures were not identified or treated.
I am currently working on a multicenter study (with 2 other centers) to identify risk factors for post-operative seizures in neonates with congenital heart disease. There is a wide variation in practice from center to center e.g. risk factors identified on our multivariable analysis include use of DHCA which is not performed at many centers. In addition, at other centers there is routine use of delayed sternal closure in most post-operative neonates whereas at our center it is a smaller percentage of patients (16%) at the surgeon’s discretion.
Therefore there is a need for multicenter collaboration to identify risk factors for seizures and to target this resource intense technology to those at highest risk.
- Do you have any insight as to whether treating sub-clinical seizures improves neurological outcome?
There is recent work in term neonates with neonatal hypoxic ischemic encephalopathy from Brigham and Women’s Hospital in Boston. A randomized control trial was conducted where neonates were randomly assigned to either treatment of EEG seizures alone or treatment of clinical seizures. In this population treatment of EEG only seizures resulted in a significant reduction in seizure burden, which correlated with brain injury and performance on neurodevelopmental testing at 18-24 months of age. (Srinivasakumar Pediatrics 2015). There is no data in neonates with congenital heart disease but in our previous evaluation of the neurodevelopmental impact of postoperative electroencephalographic seizures that were detected and treated with anti-seizure medications were associated less severe deficits at 4 years of age (Gaynor JTCVS 2013), including impaired executive function and social interactions, compared to the Boston Circulatory Arrest Study where only clinical seizures were treated.