Mortality when considering status epilepticus as the underlying cause of death. AMCOD indicates any mention of cause of death; UCOD, underlying cause of death.
Comparison of trends in principal and nonprincipal status epilepticus (SE)–related discharges.
Trends in status epilepticus (SE) discharges as a function of principal vs nonprincipal diagnosis and intubation status.
Status epilepticus discharges over time as a function of insurance type.
Betjemann JP, Josephson SA, Lowenstein DH, Burke JF. Trends in Status Epilepticus—Related Hospitalizations and MortalityRedefined in US Practice Over Time. JAMA Neurol. 2015;72(6):650-655. doi:10.1001/jamaneurol.2015.0188
Status epilepticus is a common neurologic emergency with significant associated morbidity, mortality, and health care costs, yet limited data exist detailing trends in status epilepticus–related hospitalizations and mortality.
To examine trends in status epilepticus–related hospitalizations and mortality.
Design, Setting, and Participants
We performed 2 retrospective serial cross-sectional studies including 408 304 status epilepticus–related hospital visits using generalizable national data from January 1, 1999, to December 31, 2010, from the Centers for Disease Control and Prevention and the Nationwide Inpatient Sample. Centers for Disease Control and Prevention death certificate data, using International Statistical Classification of Diseases, Tenth Revision, codes, were used to determine nonstandardized and age-standardized rates of status epilepticus as the underlying cause of death in the United States. Data from the Nationwide Inpatient Sample were used to estimate population-standardized status epilepticus–related hospitalization rates using International Statistical Classification of Diseases, Ninth Revision, codes.
Main Outcomes and Measures
Status epilepticus–related hospitalizations were categorized by whether status epilepticus was the principal diagnosis, whether the patient was intubated, and by primary insurance type. Temporal trends were tested using Poisson regression and summarized with quarterly incident rate ratios.
In 2010, status epilepticus was the reported underlying cause of death in 613 deaths (approximately 2 per 1 000 000 persons). Age-standardized status epilepticus–related mortality per 1 000 000 persons increased by only 5.6% (incident rate ratio, 1.004; 95% CI, 1.002-1.006) from 1.79 in 1999 to 1.89 in 2010, while population-standardized hospitalizations for status epilepticus per 100 000 persons increased by 56.4% (incident rate ratio, 1.013; 95% CI, 1.012-1.013) from 8.86 in 1999 to 13.86 in 2010. The largest increase (181.6%; incident rate ratio, 1.030; 95% CI, 1.029-1.030) was seen among intubated patients with nonprincipal diagnoses of status epilepticus. Among varied insurance providers, the largest increase (81.3%) was seen in Medicare patients.
Conclusions and Relevance
A disconnect exists between the relatively stable status epilepticus–related mortality and the marked increase in status epilepticus hospitalizations, likely reflecting an increase in status epilepticus diagnoses through improved diagnostic sensitivity and changes in billing and coding. The definition and general approach to status epilepticus, including resource use, should evolve with these changing epidemiologic trends.
Status epilepticus represents a relatively common neurologic emergency, with an annual incidence ranging from 10 to 41 per 100 000 persons.1- 5 Generalized convulsive status epilepticus is both the most serious and most common type of status epilepticus, representing 45% to 74% of cases.2,5 As many as 30% of patients ultimately diagnosed as having epilepsy will initially present with status epilepticus.2,6 While the overall case fatality associated with status epilepticus approaches 20%, an individual’s mortality risk depends largely on the underlying etiology; acute symptomatic causes (eg, ischemic stroke, intracerebral hemorrhage, acute traumatic brain injury, and hypoxic and anoxic brain injury) are associated with a higher case fatality than are more chronic etiologies (eg, alcohol abuse, preexisting epilepsy, and remote effects of stroke and trauma).4,7- 11
The annual direct inpatient costs associated with status epilepticus approach $4 billion in the United States, making it a costly condition with an urgent need for systems-based approaches that emphasize early recognition and prompt treatment.12 As the duration of status epilepticus increases, it becomes less likely that the seizure will terminate spontaneously,13 the seizure becomes more pharmacoresistant to benzodiazepines,14,15 and the likelihood of neuronal injury increases. Consequently, status epilepticus management has evolved to focus on early seizure termination through the development of newer abortive medications and the use of benzodiazepines in the prehospital setting.16- 21 During the past decade, there has also been an emphasis on extended inpatient electroencephalographic (EEG) monitoring in both the intensive care unit (ICU) and non-ICU settings to identify seizures and status epilepticus that were previously not detected.22- 26
It remains unclear whether these changes have led to changes in the hospitalization rates and mortality of patients with status epilepticus. To determine whether this is the case, we used national data sets to explore trends in status epilepticus–related hospitalizations and mortality over time. We hypothesized that status epilepticus–related hospitalizations would be increasing owing to higher sensitivity to the diagnosis of status epilepticus, and that mortality would be declining owing to early identification of status epilepticus and improved prehospital treatment.
Multicause mortality data from the Centers for Disease Control and Prevention (CDC) from January 1, 1999, to December 31, 2010, were used to determine trends in status epilepticus mortality and data from the Healthcare Cost and Utilization Project Nationwide Inpatient Sample (NIS) were used to determine trends among 408 304 status epilepticus–related hospital visits in the United States from January 1, 1999, to December 31, 2010. As these are limited data sets, they are not regulated by the University of Michigan Institutional Review Board. Informed consent was not obtained, as the original data were captured for other purposes and not primarily for research purposes. The CDC’s National Vital Statistics System collects official death records from local municipalities (eg, city coroner offices) to enable national measurements.27 Their process then takes these death certificates and assigns an underlying cause of death (UCOD) from the list of diagnoses appearing on the death certificate, by applying the World Health Organization definition, “the disease or injury which initiated the train of morbid events leading directly to death, or the circumstances of the accident or violence which produced the fatal injury.”28 These definitions are assigned using a complex set of prioritization rules, which are updated annually by the World Health Organization.29 The Centers for Disease Control and Prevention applies these prioritization rules and compiles multicause mortality files that contain basic demographic information, UCOD, and all listed death certificate diagnoses. If status epilepticus appears in any position on the death certificate, it is considered to be an any mention cause of death for an individual.
Data from the 1999-2010 NIS were used to estimate trends in status epilepticus hospitalizations. The NIS is maintained by the Agency for Healthcare Research and Quality and was developed as part of the Healthcare Cost and Utilization Project.30 The NIS contains data on all hospitalizations in a sample of approximately 20% of all acute-care hospitals in the United States in a given year. The NIS selects hospitals using a stratified sampling design (based on hospital region, location, teaching status, and bed size) to identify a 20% sample of US community hospital visits, and can be used to generate nationally representative estimates.31
Status epilepticus was identified via International Statistical Classification of Diseases, Tenth Revision (ICD-10), codes G41.0-G41.9 in CDC data if listed as the UCOD or if it appeared in any position on the death certificate (any mention cause of death). Status epilepticus was identified in the NIS using International Statistical Classification of Diseases, Ninth Revision (ICD-9), codes 345.2 (petit mal status epilepticus) and 345.3 (grand mal status epilepticus). This approach captures all types of status epilepticus, with prior studies demonstrating excellent positive predictive values for these ICD-9 and ICD-10 codes encompassing the most common forms of status epilepticus.32 Video EEG monitoring was identified in the NIS using ICD-9 procedure code 89.19 for video and radio EEG monitoring.
Nonstandardized mortality rates were calculated by dividing status epilepticus mortality by the US population in each year using data from the US Census.33 The population in standardized age groups was calculated by summing the number of persons in each age group using CDC bridged-race population files.34 Age-standardized mortality rates were then calculated using the standardized population from the year 2000.35 Poisson regression was used to test for temporal trends to predict the number of deaths in 1 quarter, with an offset of the total US population in that year. Trends were summarized with quarterly incident rate ratios (IRRs).
Status epilepticus hospitalizations were categorized by whether an individual patient was intubated (ICD-9 procedure code 96.04 or 96.05) and whether status epilepticus was the principal diagnosis or a secondary diagnosis listed for a visit. As the principal diagnosis is intended to identify the condition that, after study, occasioned the visit to the hospital,36 it is likely that patients with principal status epilepticus are more likely to have preexisting epilepsy or new-onset idiopathic status epilepticus, whereas those with secondary status epilepticus are more likely to represent patients with an acute underlying etiology for status epilepticus.
We hypothesized that increased diagnostic sensitivity over time would lead to a relative increase in patients with acute symptomatic status epilepticus, particularly symptomatic status epilepticus with intubation, reflecting increased use of EEG monitoring in the ICU nationally. The number of status epilepticus hospitalizations overall and the number in each category were estimated using descriptive statistics after applying NIS survey weights. As a post hoc analysis, we also grouped status epilepticus hospitalizations by primary insurance type to explore whether there may be differential effects of coding practices. Finally, we explored overall temporal trends and trends by diagnostic category in the use of video EEG (ICD-9 procedure code 89.19). As this code is most likely insensitive, but reasonably specific, for video EEG monitoring, changes in the code over time likely reflect the relative trends in the use of video EEG over time, but most likely underestimate absolute use.37 For all groups, temporal trends were estimated using Poisson regression.
In 2010, status epilepticus was listed as the underlying cause of death in 613 deaths reported by the CDC, with an overall age-standardized mortality of approximately 2 per 1 000 000 persons. When considered as the UCOD, age-standardized status epilepticus mortality increased by 5.6% from 1.79 in 1999 to 1.89 in 2010 (IRR, 1.004; 95% CI, 1.002-1.006) (Figure 1). Any mention of status epilepticus (any mention cause of death) on the death certificates (IRR, 1.005; 95% CI, 1.004-1.006) was approximately double the UCOD rate during the same study period (Figure 1). Using the NIS data, we found an increase of 56.4% in all population-standardized status epilepticus–related hospitalizations per 100 000 persons from 8.86 in from 1999 to 13.86 in 2010 (IRR, 1.013; 95% CI, 1.012-1.013) (Figure 2). When separating principal status epilepticus from nonprincipal status epilepticus, we found a smaller increase (33.4%) in principal status epilepticus cases (IRR, 1.008; 95% CI, 1.007-1.008) from 5.92 in 1999 to 7.90 in 2010, and a larger relative increase (102.0%) in nonprincipal status epilepticus cases (IRR, 1.021; 95% CI, 1.021-1.022) from 2.95 in 1999 to 5.96 in 2010. Overall video EEG monitoring in the population increased from 1.1% in 1999 to 4.3% in 2010 (IRR, 1.05; P < .01), with the largest relative increase in intubated patients with nonprincipal status epilepticus (IRR, 1.09; P < .01).
Figure 3 depicts principal status epilepticus vs nonprincipal status epilepticus discharges and whether or not the patient was intubated. Each of the 4 categories increased somewhat, but the largest increase (181.8%) per 100 000 persons was seen in intubated patients with nonprincipal status epilepticus (IRR, 1.030; 95% CI, 1.029-1.030) from 1.32 in 1999 to 3.72 in 2010. Similar to the overall status epilepticus–related hospitalization data, these curves show the most dramatic increase after an inflection point around 2005.
In an effort to understand other factors potentially driving the increase in status epilepticus–related hospitalizations, we examined status epilepticus discharges by insurance type (Figure 4). Status epilepticus hospitalizations per 100 000 persons increased for all insurance types, but for Medicare, the rate was relatively constant until 2005, after which there was a rather abrupt 81.1% increase from 2.75 in 2005 to 4.98 in 2010. We observed a similar trend with smaller increases in status epilepticus discharges from Medicaid (36.9%) from 3.09 in 2005 to 4.23 in 2010 and private insurers (13.5%) from 2.82 in 2005 to 3.20 in 2010.
Our study represents a novel attempt to explore possible explanations for the divergent trends between stable status epilepticus–related mortality and a dramatic rise in population-standardized status epilepticus–related hospitalizations. We also found status epilepticus–related mortality to be substantially lower than prior estimates. Given these findings, we have asked whether status epilepticus is already being redefined in clinical practice.
Overall, there is a paucity of recent work addressing epidemiologic trends in status epilepticus. Studies from a single urban public hospital revealed little variation in the etiologies for status epilepticus over a span of 2 decades.38,39 Data from the Rochester Epidemiologic Project demonstrated an increase in the incidence of status epilepticus from 1975 through 1984, largely related to an increased incidence of status epilepticus in the elderly and the recognition of myoclonic status epilepticus following cardiac arrest.3 A study of California hospitals found that the incidence rate of status epilepticus fell by 42% from 1991 to 1998 and that the case fatality rate during that period was lower than previously reported.10 A more recent national survey showed an increase in status epilepticus incidence of 3.5 to 12.5 per 100 000 persons from 1979 to 2010 without a significant change in in-hospital mortality; the authors postulated that the increased incidence might be a reflection of improved diagnostic capabilities and changes in coding practices.40
The simplest explanation for the divergence between mortality and hospitalizations is that status epilepticus diagnoses are increasingly made in clinical circumstances in which they were not previously made. This theory was postulated by prior studies and is supported by our observation that the largest increase in status epilepticus diagnoses occurred in intubated patients with a secondary status epilepticus diagnosis, suggesting a potential role for increased ICU monitoring as an explanation for this trend. In addition, we further investigated the potential role of changes in coding practices and found a marked increase in Medicare status epilepticus visits in the last 5 years of the study period.
Effective October 1, 2007, grand mal status, epileptic, was listed in the major complication and comorbidity list by the Centers for Medicare and Medicaid Services.41 This change created an incentive to list status epilepticus among a patient’s diagnoses to increase the value of their diagnosis related group and may partially explain why the largest increase in status epilepticus–related hospitalizations occurred among the Medicare group. As is often the case, coding and billing modifications made by Medicare are often followed temporally by Medicaid and private insurers, potentially accounting for their delayed increase seen in Figure 4.
The second potential reason for the increase in status epilepticus–related hospitalizations involves a relatively recent appreciation for how common seizures are in hospitalized patients, leading to an increase in the diagnosis of status epilepticus through extended EEG monitoring. We found that the subset of patients who were intubated and had a nonprimary diagnosis of status epilepticus represented the largest increase in status epilepticus cases; presumably, some of these patients are hospitalized for other etiologies and found to be in status epilepticus through extended EEG monitoring. This finding is tentatively supported by the observed increase in video EEG monitoring during the study period. While the codes used to identify video EEG monitoring in this study were likely insensitive, the relative trend suggests that increased EEG monitoring may be contributing to the rise in status epilepticus–related hospitalizations. This finding, along with the further development of neurologic ICUs, may also serve to explain why the increasing trend in status epilepticus–related hospitalizations was seen prior to the billing changes made in 2007.
Status epilepticus–related mortality in our study is significantly lower than previous estimates. Prior studies report an annual status epilepticus incidence of 10 to 41 per 100 000 persons,1- 5 and mortality estimates range from approximately 10% to 20%.5,7,8,10,11,40 Using census data, this estimate translates into roughly 4500 to 18 000 status epilepticus deaths annually in the United States, or approximately 15 to 60 deaths per 1 million persons, similar to the reported status epilepticus mortality rate of 36 to 40 per 1 million persons from 1975 to 1984.3 Both the incidence (10.3-17.1 per 100 000 persons) and case fatality rate (7.6%-9.3%) are lower in countries with centralized health care.1,4 While the lower incidence and case fatality rates in Europe may reflect differences in health care systems, the higher incidence in the United States may in part be related to demographics, with a higher incidence observed among nonwhite minorities.5
Our estimate from CDC death certificate data of 2 status epilepticus–related deaths per 1 million persons translates to a status epilepticus–related mortality of approximately 0.5% to 2%. The most likely explanation for this disconnect is that our study defined mortality as the UCOD while prior work has directly categorized the proportion of patients with status epilepticus who die. While assignment of UCOD relies on many assumptions, we believe that this is a more realistic reflection of true status epilepticus–related mortality and is likely to be increasingly realistic over time. Prior studies of status epilepticus mortality that examined all cases of status epilepticus have acknowledged that the underlying etiology for status epilepticus often drives mortality, but these studies have been limited in their ability to account for the proportion of cases in which status epilepticus is the primary driver of mortality. Given that we observed the most dramatic increases in status epilepticus hospitalizations among intubated patients with nonprincipal status epilepticus, it is likely that the underlying etiology is an increasingly prevalent cause of mortality. Presumably these are patients in the ICU with status epilepticus secondary to an acute underlying etiology (eg, intracerebral hemorrhage, ischemic stroke, trauma) that has a relatively high mortality. This finding suggests a critical shift in how the medical community is redefining clinical status epilepticus, recognizing it more as a sign of severe secondary brain injury and therefore not considering it as the underlying cause of death, and in many instances not even listing it on CDC death certificate data.
An evolving definition of principal vs nonprincipal status epilepticus should focus on more accurately reflecting prognosis. This change in definition may inform decisions affecting resource use, including the utility of diagnostic testing, treatment, and goals of care discussions. This development will be increasingly important given the recent increase in extended EEG monitoring in the ICU, a scarce and costly resource. Numerous publications have documented the relatively high frequency of seizures in the ICU but more data are urgently needed exploring the effect of seizures and status epilepticus on patient outcomes, particularly in nonprincipal status epilepticus cases in which status epilepticus may not be the cause of death.
Our study has some important limitations. The retrospective nature of the study allows for potential selection bias. We compare our data to those from prior status epilepticus studies that differed methodologically in that they often gathered all status epilepticus cases from a single center rather than relying on national databases. Accordingly, prior studies from academic centers may be skewed toward higher mortality given their referral bias for sicker patients compared with our data set, which samples a wide array of hospitals. The definition and diagnosis of status epilepticus in large data sets is by nature imperfect, given that we did not have access to EEG results and clinical data. These weaknesses are mitigated in part by using death certificate data for estimating mortality. While death certificates are the criterion standard for identifying cause of death, it is possible that the complicated UCOD assignment process may underestimate the true proportion of status epilepticus mortality, as has been suggested in the case of diabetes mellitus.42,43 The analysis of increased video EEG monitoring is limited by the lack of an available nationally representative data source that includes data on both hospitalizations and physician-level Current Procedural Terminology codes; however, we instead explored trends in the use of an ICD-9 procedure code for video EEG monitoring. Our data are limited to the United States and their applicability to other countries where models of health care and demographics differ is uncertain.
Status epilepticus is a relatively common neurologic problem, with recent EEG monitoring studies demonstrating a high frequency of seizures in the hospital setting. The incidence of status epilepticus continues to increase, although mortality has remained largely stable. Further study is needed to understand status epilepticus as an epiphenomenon of acute brain injury and its effect on clinical outcomes, which may lead to a revised definition of status epilepticus that more accurately reflects prognosis and affects resource use.
Accepted for Publication: February 10, 2015.
Corresponding Author: John P. Betjemann, MD, Department of Neurology, University of California–San Francisco, San Francisco General Hospital, 1001 Potrero Ave, Bldg 1, Room 101, San Francisco, CA 94110 (firstname.lastname@example.org).
Published Online: April 27, 2015. doi:10.1001/jamaneurol.2015.0188.
Author Contributions: Drs Betjemann and Burke had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: Betjemann, Burke.
Drafting of the manuscript: Betjemann, Burke.
Critical revision of the manuscript for important intellectual content: Josephson, Lowenstein, Burke.
Statistical analysis: Burke.
Administrative, technical, or material support: Betjemann, Josephson, Burke.
Study supervision: Josephson, Lowenstein, Burke.
Conflict of Interest Disclosures: Dr Betjemann reported receiving honoraria for speaking at the Recent Advances in Neurology and the California Neurologic Society meetings and for authoring an article for the Agency for Healthcare Research and Quality. Dr Josephson reported receiving personal compensation as Editor in Chief of Journal Watch Neurology, in an editorial capacity for Continuum Audio, and previously as Associate Editor for Annals of Neurology. Dr Lowenstein reported previously receiving compensation as Associate Editor for Annals of Neurology. Dr Burke reported receiving honoraria from the American Academy of Neurology for writing an article for Continuum and from the University of Rochester for presenting at grand rounds, and funding from Astra Zeneca as an adjudicator in the SOCRATES trial. No other disclosures were reported.