Predictors of New-Onset Seizures in Head Trauma
Predictors of New-Onset Seizures in Head Trauma
There were 827 subjects, including children and elderly people, admitted to the emergency room (ER) at Oulu University Hospital during 1999 on account of acute head trauma. To investigate the occurrence and predictors of new-onset seizures among these subjects, we excluded from the cohort those who had had a recorded history of seizures, dementia, TBI, brain infarction or other neurological disease before the index trauma. This was done in order to obtain a cohort of subjects with no previous seizure history or any disease known to predispose the carrier to epilepsy. Subjects with a history of psychiatric disease were included, however, because they are prone to drug and/or alcohol abuse. The subjects of this kind in the present series had psychosis (n=10), depression and/or anxiety disorder (n=9) or personality disorder (n=2), and two of them had also received a diagnosis of alcohol dependence before the index trauma. The exclusions left us with a total of 739 subjects who were followed up until the end of 2009.
Acute head traumas were identified in 1999 from the daily checklists kept in the ER. A nurse checked the list and the ER records daily and picked out any notes that suggested a head trauma, including subjects with multiple injuries, for further assessment (by one of the research team). In addition, the hospital discharge register was checked weekly to identify cases of head trauma and TBI that had not been observed on the basis of the ER checklist. This method maximised the possibilities of identifying head trauma subjects.
The baseline data on all the subjects gathered directly from the hospital charts in 1999 included age at the time of the index trauma, sex, severity of the index head trauma and whether the subject had been under the influence of alcohol at the time of the trauma, together with a history of previous diagnoses. Follow-up data on all the subjects for the period 1999–2009 were then gathered from the National Hospital Discharge (NHD) Register using the record linkage technique, that is, by reference to the subjects' personal social security codes. We did not contact the subjects themselves. The NHD register data included all diagnoses recorded on discharge from any hospital after an inpatient, outpatient or ER visit and also all diagnoses for inpatient visits to health centres (but not for outpatient or ER visits to health centres). Dates of death were obtained from the Statistics Finland Causes-of-Death (SFCD) register. If either the NHD and/or SFCD data included any diagnosis of seizures, we checked the nature of the seizure problem, the possibility of alcohol involvement and other data on the subject in the records of our own hospital as well. A seizure problem was considered to be alcohol-related if it was indicated in the NHD register and/or in our hospital discharge data as being due to acute withdrawal from heavy alcohol drinking. This was possible for those who lived in the catchment area of Oulu University Hospital (89%).
The severity of the head trauma was divided into three categories according to the presence and severity of TBI: no TBI, mild TBI and moderate-to-severe TBI. The last-mentioned category included all subjects who presented with Glasgow Coma Scale (GCS) <13 and also subjects with brain contusion, diffuse axonal injury, traumatic subarachnoid haemorrhage and traumatic intracranial haematomas regardless of their GCS score or need for surgery. Mild TBI included subjects with GCS scores of 13–15 and LOC <30 min or post-traumatic amnesia (PTA) <1 h in the absence of traumatic intracranial findings in a brain CT/MRI. Subjects without a brain CT/MRI were also classified as having TBI if any other criteria for mild TBI were fulfilled. Subjects with a GCS score of 15 and no LOC or no PTA were classified as having head trauma without TBI.
All subjects with alcohol in their blood or breath at the time of sustaining the trauma or were mentioned in the emergency department records as having been found by a nurse or a physician to be under the influence of alcohol were classified as having alcohol-related trauma. The criteria were a smell of alcohol in the breath combined with signs suggesting alcoholic intoxication, such as unsteady gait, slurred speech or aggressive behaviour. Informed consent is not required in our hospital for the testing of breath or blood alcohol, and samples are usually taken at least from all head trauma subjects with impaired consciousness. In the present series of 739 subjects, breath or blood samples were obtained from 189 (25.6%), of whom 186 were alcohol-positives, while an additional 83/739 subjects (11.2%) were judged by professional emergency care providers to be intoxicated.
The data were analysed with IBM SPSS Statistics (V.20). Categorical variables were assessed using conventional statistical methods, including Pearson's χ test and Fisher''s exact test, and continuous variables were compared between groups with t-tests. Cumulative rates of developing a seizure were estimated by the Kaplan-Meier product-limit method, and the curves for the groups were compared using the log-rank test. Cox proportional hazard regression models were used to identify independent predictors of a new-onset seizure during the follow-up period, with age, sex, severity of the index head trauma, presence of alcohol in connection with the index trauma and preceding psychiatric disease included in the final predictive model. The proportionality assumption was checked, and the test for significance was based on changes in log (partial) likelihood. A two-tailed p value of less than 0.05 was considered to be statistically significant.
Material and Methods
There were 827 subjects, including children and elderly people, admitted to the emergency room (ER) at Oulu University Hospital during 1999 on account of acute head trauma. To investigate the occurrence and predictors of new-onset seizures among these subjects, we excluded from the cohort those who had had a recorded history of seizures, dementia, TBI, brain infarction or other neurological disease before the index trauma. This was done in order to obtain a cohort of subjects with no previous seizure history or any disease known to predispose the carrier to epilepsy. Subjects with a history of psychiatric disease were included, however, because they are prone to drug and/or alcohol abuse. The subjects of this kind in the present series had psychosis (n=10), depression and/or anxiety disorder (n=9) or personality disorder (n=2), and two of them had also received a diagnosis of alcohol dependence before the index trauma. The exclusions left us with a total of 739 subjects who were followed up until the end of 2009.
Acute head traumas were identified in 1999 from the daily checklists kept in the ER. A nurse checked the list and the ER records daily and picked out any notes that suggested a head trauma, including subjects with multiple injuries, for further assessment (by one of the research team). In addition, the hospital discharge register was checked weekly to identify cases of head trauma and TBI that had not been observed on the basis of the ER checklist. This method maximised the possibilities of identifying head trauma subjects.
The baseline data on all the subjects gathered directly from the hospital charts in 1999 included age at the time of the index trauma, sex, severity of the index head trauma and whether the subject had been under the influence of alcohol at the time of the trauma, together with a history of previous diagnoses. Follow-up data on all the subjects for the period 1999–2009 were then gathered from the National Hospital Discharge (NHD) Register using the record linkage technique, that is, by reference to the subjects' personal social security codes. We did not contact the subjects themselves. The NHD register data included all diagnoses recorded on discharge from any hospital after an inpatient, outpatient or ER visit and also all diagnoses for inpatient visits to health centres (but not for outpatient or ER visits to health centres). Dates of death were obtained from the Statistics Finland Causes-of-Death (SFCD) register. If either the NHD and/or SFCD data included any diagnosis of seizures, we checked the nature of the seizure problem, the possibility of alcohol involvement and other data on the subject in the records of our own hospital as well. A seizure problem was considered to be alcohol-related if it was indicated in the NHD register and/or in our hospital discharge data as being due to acute withdrawal from heavy alcohol drinking. This was possible for those who lived in the catchment area of Oulu University Hospital (89%).
The severity of the head trauma was divided into three categories according to the presence and severity of TBI: no TBI, mild TBI and moderate-to-severe TBI. The last-mentioned category included all subjects who presented with Glasgow Coma Scale (GCS) <13 and also subjects with brain contusion, diffuse axonal injury, traumatic subarachnoid haemorrhage and traumatic intracranial haematomas regardless of their GCS score or need for surgery. Mild TBI included subjects with GCS scores of 13–15 and LOC <30 min or post-traumatic amnesia (PTA) <1 h in the absence of traumatic intracranial findings in a brain CT/MRI. Subjects without a brain CT/MRI were also classified as having TBI if any other criteria for mild TBI were fulfilled. Subjects with a GCS score of 15 and no LOC or no PTA were classified as having head trauma without TBI.
All subjects with alcohol in their blood or breath at the time of sustaining the trauma or were mentioned in the emergency department records as having been found by a nurse or a physician to be under the influence of alcohol were classified as having alcohol-related trauma. The criteria were a smell of alcohol in the breath combined with signs suggesting alcoholic intoxication, such as unsteady gait, slurred speech or aggressive behaviour. Informed consent is not required in our hospital for the testing of breath or blood alcohol, and samples are usually taken at least from all head trauma subjects with impaired consciousness. In the present series of 739 subjects, breath or blood samples were obtained from 189 (25.6%), of whom 186 were alcohol-positives, while an additional 83/739 subjects (11.2%) were judged by professional emergency care providers to be intoxicated.
The data were analysed with IBM SPSS Statistics (V.20). Categorical variables were assessed using conventional statistical methods, including Pearson's χ test and Fisher''s exact test, and continuous variables were compared between groups with t-tests. Cumulative rates of developing a seizure were estimated by the Kaplan-Meier product-limit method, and the curves for the groups were compared using the log-rank test. Cox proportional hazard regression models were used to identify independent predictors of a new-onset seizure during the follow-up period, with age, sex, severity of the index head trauma, presence of alcohol in connection with the index trauma and preceding psychiatric disease included in the final predictive model. The proportionality assumption was checked, and the test for significance was based on changes in log (partial) likelihood. A two-tailed p value of less than 0.05 was considered to be statistically significant.
Source...