- Email: [email protected]
Men with nonapnea sleep disorders have a high risk of developing subsequent epilepsy: A nationwide population-based cohort study
Preventive Medicine 91 (2016) 211–216
Contents lists available at ScienceDirect
Preventive Medicine journal homepage: www.elsevier.com/locate/ypmed
Men with nonapnea sleep disorders have a high risk of developing subsequent epilepsy: A nationwide population-based cohort study Tomor Harnod, MD, PhD a,b, Yu-Chiao Wang, MSc c,d, Chia-Hung Kao, MD e,f,g,⁎ a
Department of Neurosurgery, Hualien Tzu Chi General Hospital, Buddhist Tzu Chi Medical Foundation, Hualien, Taiwan College of Medicine, Tzu Chi University, Hualien, Taiwan Management Office for Health Data, China Medical University Hospital, Taichung, Taiwan d College of Medicine, China Medical University, Taichung, Taiwan e Graduate Institute of Clinical Medical Science and School of Medicine, College of Medicine, China Medical University, Taichung, Taiwan f Department of Nuclear Medicine and PET Center, China Medical University Hospital, Taichung, Taiwan g Department of Bioinformatics and Medical Engineering, Asia University, Taichung, Taiwan b c
a r t i c l e
i n f o
Article history: Received 7 January 2016 Received in revised form 6 August 2016 Accepted 24 August 2016 Available online 25 August 2016 Keywords: Cohort study Epilepsy Insomnia National Health Insurance Sleep disorder
a b s t r a c t Objective. This nationwide population-based cohort study evaluated the effects of nonapnea sleep disorders (NSDs) on the development of epilepsy. Methods. We identified 63,865 patients aged ≥20 years, diagnosed with NSDs (ICD-9-CM: 307.4 or 780.5), and without coding for apnea-related sleep disorders (ICD-9-CM: 780.51, 780.53, or 780.57) during 2000– 2003 as the NSD cohort. In addition, we enrolled a comparison cohort of 127,728 patients. We calculated the adjusted hazard ratio (aHR) for developing epilepsy (ICD-9-CM: 345) after adjustment for age, sex, comorbidities, and drug use. A Kaplan–Meier analysis was used to measure the cumulative incidence of epilepsy between the 2 groups until the end of 2011. Results. The cumulative incidence of epilepsy was significantly higher in the NSD cohort than in the comparison cohort. The aHR for developing epilepsy in the NSD cohort was 1.52 (95% CI = 1.37–1.69). The risk of developing epilepsy was higher among males (aHR = 1.41) than among females. The age-stratified effects of NSDs on developing epilepsy were the highest among patients aged ≥65 years. When comorbidities and NSDs coexisted, the risk of epilepsy was specifically increased in patients having an NSD and stroke (aHR: 8.61, 95% CI: 7.43–9.98) in addition to brain tumors (aHR: 7.66, 95% CI: 5.06–11.6). Conclusion. This study indicated that patients with NSDs have a higher risk of developing epilepsy and that the risk is much higher among men and older patients. These findings suggest that NSDs constitute a predisposing, possibly independent factor for developing subsequent epilepsy in adulthood. © 2016 Elsevier Inc. All rights reserved.
1. Introduction Both epilepsy and sleep disorders (SDs) are common neurological disorders, and either one of them can affect the quality of daily life in affected patients. Epilepsy has a prevalence of 0.5%–1% in the general population worldwide (Hauser et al., 1991; Benbadis and Allen, 2000), and epilepsy exerts a more serious effect on patients' lives than SDs do, such as the high risks of premature death from heart disease and stroke (Neligan et al., 2011; Aurlien et al., 2012). A 2.5-fold mortality risk, compared with the general Taiwanese population, was recently reported for Southern Taiwanese patients with epilepsy (Chang et al., 2012).
⁎ Corresponding author at: Graduate Institute of Clinical Medical Science and School of Medicine, College of Medicine, China Medical University, No. 2, Yuh-Der Road, Taichung 404, Taiwan. E-mail address: [email protected] (C.-H. Kao).
http://dx.doi.org/10.1016/j.ypmed.2016.08.034 0091-7435/© 2016 Elsevier Inc. All rights reserved.
The causes of epilepsy indicate that the disease either has an idiopathic origin or appears as symptomatic epilepsy secondarily following a brain insult. Numerous causes lead to symptomatic epilepsy in adulthood, with the 3 major causes being head injury, stroke, and brain tumors (Annegers et al., 1996). According to current knowledge, a seizure develops from an imbalanced expression of the excitatory and inhibitory ion channels in the central nervous system (CNS). The electrophysiological instability in the cortex presents as a spike in an electroencephalogram, and the hypersynchronization of spikes may develop into an epileptic seizure. Sleep deprivation is a well-known factor for increasing seizure susceptibility and often provokes epileptic seizures (Bostock et al., 2015; Lucey et al., 2015; Schmitt, 2015). Sleep is vital for living. SDs result in profound consequences for physiology, behavior, and daily abilities during waking hours (Ohayon, 2009). Symptoms of dyssomnia or insomnia may be temporary or chronic, depending on the type of SD. Currently, nonapnea SDs (NSDs)
212
T. Harnod et al. / Preventive Medicine 91 (2016) 211–216
without evidence of hypoxic damage to the CNS are considered less threatening to human health. In this study, we determined whether NSDs are independent and predisposing factors that lead to adulthood epilepsy and subsequently affect the life quality and life-span of affected patients by enrolling an NSD cohort from Taiwan's nationwide population-based database.
2. Methods and materials 2.1. Data source The Taiwan Bureau of National Health Insurance developed the National Health Insurance (NHI) program, which has been in effect since March 1995. This program offers full medical service to more than 99% of Taiwan residents (approximately 23 million). The claims data on the Taiwan NHI program constitute a large computerized database called the National Health Insurance Research Database (NHIRD). We used the Longitudinal Health Insurance Database (LHID 2000), a subset of the NHIRD, in this population-based cohort study. The LHID 2000 contains claims data on 1 million randomly selected beneficiaries from the 23 million insured individuals who registered during 1996–2011. Demographic data (such as those on birth and sex) and diagnostic data (such as the date and diagnoses of outpatient visits, and the date of admission and discharge with diagnoses) of all insured individuals are included in this database. In our study, International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes were recorded by a general physician, a psychiatrist, or a neurologist and were used to define diseases and medical procedures. The NHIRD is highly representative of the general population of Taiwan because the NHI is a single-payer and universal program. All insurance claims are scrutinized by medical reimbursement specialists and subjected to peer review according to standard diagnostic criteria. In addition, physicians and hospitals are penalized for incorrect diagnoses and codes. Therefore, the diagnoses of sleep disorders, epilepsy, and comorbidities in this study are highly reliable (Cheng et al., 2011; Harnod et al., 2015). Furthermore, the NHIRD data are anonymized for protecting personal privacy. This study was approved by the Institutional Review Board of China Medical University.
2.3. Data analysis The SAS 9.4 statistical package (SAS Institute Inc., NC, USA) was used for the statistical analyses, and 2-tailed P b 0.05 was considered significant. The baseline distribution of demographic variables and potential risk factors (history of comorbidities and drug use) in the NSD and comparison cohorts were analyzed using a chi-square test for categorical variables and a Student t-test for continuous variables. The incidence rates of epilepsy development in both cohorts were calculated after stratifying each variable. The incidence rate ratio (IRR) and 95% confidence intervals (95% CIs) of epilepsy were presented using the Poisson regression model. The crude hazard ratios (HRs) and 95% CIs of epilepsy in the NSD and comparison cohorts were calculated using the Cox proportional hazard regression model. After adjustment for sex, age, each comorbidity (including hypertension, hyperlipidemia, diabetes, stroke, anxiety, depression, head injury, and brain tumors), and drug use (no/ yes), the association between NSDs and epilepsy was evaluated using a multivariate Cox proportional hazard model to estimate the adjusted HR (aHR) and 95% CI. We also used the stepwise Cox proportional hazard model to estimate the risk of epilepsy associated with NSDs. The parameters of sex, age, each comorbidity, and drug use (no/yes) were selected stepwise. In addition, we conducted a Kaplan–Meier analysis on the cumulative incidence of epilepsy in the 2 cohorts by using R software (R Foundation for Statistical Computing, Vienna, Austria) and estimated the difference between the 2 cumulative incidence curves by using a logrank test.
3. Results The baseline characteristics of demographic factors in the NSD and comparison cohorts are presented in Table 1. We enrolled 63,865 patients in the NSD cohort and 127,728 patients in the comparison cohort. The distributions of sex and age were similar in the 2 cohorts, with female being the predominant sex (63.7%). Moreover, young to middle-
Table 1 Demographics, comorbidities, and history of drug use in the NSD and comparison cohorts.
2.2. Selection of study patients Our study enrolled 63,865 patients aged ≥20 years and newly diagnosed with NSDs (ICD-9-CM: 307.4, 780.5, excluding780.51, 780.53, and 780.57) during 2000–2003. The index date was defined as when a patient received a primary diagnosis of an NSD for the first time. Through frequency matching with age (in 5-year intervals), sex, and the index year, 2 patients without SD were randomly selected from the LHID 2000 for each patient with an NSD to form the comparison cohort (n = 127,728). Patients who developed epilepsy (ICD-9-CM: 345) before the index date and those aged b20 years were excluded. The follow-up period was measured in both cohorts from the index date until the development of epilepsy, withdrawal from the LHID database, or December 31, 2011, which ever occurred first. Drug exposure histories in the subjects of this study were measured before the end date, and their types of exposure were categorized as zolpidem, benzodiazepine (BZD), or both. In addition, we adjusted for comorbidities. The comorbidities were defined as diseases diagnosed before the study end date, including hypertension (ICD-9-CM: 401–405), hyperlipidemia (272), diabetes (250), anxiety (300.0, 300.2, 300.3, 308.3, and 309.81), depression (296.2, 296.3, 296.82, 300.4, 309.0, 309.1, 3092.8, and 311), head injury (850–854, and 959.01), stroke (430–438), and brain tumors (225, 191, 192, 194.3, and 194.4).
Sex Female Male Age, year 20–45 45–65 ≥65 Mean (standard deviation)a Comorbidity Hypertension Hyperlipidemia Diabetes Stroke Anxiety Depression Head injury Brain tumors Drug use Non Zolpidem BZD Both Mean follow-up years (standard deviation)a
Comparison cohort (N = 127,728)
NSD cohort (N = 63,865)
n
%
n
%
81,390 46,338
63.7 36.3
40,695 23,170
63.7 36.3
P value
0.99
0.99 53,742 42.1 47,354 37.1 26,632 20.9 49.8 (16.2)
26,871 42.1 23,677 37.1 13,317 20.9 49.8 (16.2)
0.45
47,515 30,805 23,812 17,273 10,749 4394 7945 523
30,926 22,565 15,156 13,346 23,916 13,716 5978 423
48.4 35.3 23.7 20.9 37.4 21.5 9.36 0.66
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
2079 3.26 406 0.64 21,292 33.3 40,088 62.8 9.17 (2.25)
b0.0001
37.2 24.1 18.6 13.5 8.42 3.44 6.22 0.41
31,458 24.6 1005 0.79 77,617 60.8 17,648 13.8 9.00 (2.44)
NSD: nonapnea sleep disorder; BZD: benzodiazepine; chi-square test. a Student t-test.
b0.0001
T. Harnod et al. / Preventive Medicine 91 (2016) 211–216
aged patients were predominant (79.2% with 20–65 years), with the mean age being 49.8 years (standard deviation: 16.2 years) in both cohorts. More NSD patients had comorbidities and drug use than did those in the comparison cohort, and the NSD cohort had a significantly longer mean follow-up period than that of the comparison cohort (9.17 vs 9.00 years) (chi-square P b 0.0001, Table 1). During the follow-up period, the NSD cohort revealed a higher cumulative incidence of developing epilepsy than that of the comparison cohort (log-rank test P b 0.0001, Fig. 1). The overall IRR of epilepsy was 1.92-fold in the NSD cohort compared with that in the comparison cohort (1.47 per 1000 person-years vs 0.76 per 1000 person-years, Table 2). After adjustment for age, sex, each comorbidity, and drug use, the NSD cohort had a 1.52-fold risk (95% CI: 1.37–1.69, P b 0.001) of developing epilepsy (aHR: 1.51 and 1.53, 95% CI: 1.30–1.74 and 1.32–1.78 for female and male patients, respectively). The male patients in the NSD cohort had a higher incidence rate of epilepsy than that of the females (1.17 vs 2.02 per 1000 person-years for female and male patients, respectively). The incidence rate of epilepsy increased with age, but the NSD patients aged 20–45 years clearly had a higher risk of epilepsy development than controls aged 20–45 years (aHR: 1.99, 95% CI: 1.57–2.51). The comorbidity-stratified effects of NSDs on epilepsy were the highest in patients without comorbidities (aHR: 2.66 and 1.51, 95% CI: 1.95–3.63 and 1.37–1.68 for those without comorbidities and those with one or more comorbidities, respectively). A significantly higher association between NSDs and epilepsy development was observed among the NSD patients with drug use compared with the controls with drug use (aHR: 1.54, 95% CI: 1.39–1.72; Table 2). Table 3 presents the individual effects of different factors on the risk of epilepsy development. The significantly increased risk factors for epilepsy were male sex, old age, hypertension, diabetes, stroke, depression, head injury, and brain tumors. After adjustment for age, sex, each comorbidity, and drug use, a low risk of epilepsy was revealed in subjects with hyperlipidemia (aHR: 0.64, 95% CI: 0.57–0.71) and anxiety (aHR: 0.84, 95% CI: 0.74–0.95). Moreover, drug use did not significantly influence epilepsy development. Regarding sex, male patients were at a higher risk of epilepsy (aHR: 1.41, 95% CI: 1.28–1.55) than female patients. Compared with that in the young adults aged 20–45 years, the aHR of epilepsy in the elderly patients aged ≥65 years significantly increased by 1.52-fold (95% CI: 1.30–1.78) (Table 3). The 3 comorbidities related to epilepsy etiology — brain tumors, stroke, and head injury —
Fig. 1. Cumulative incidence of epilepsy in the comparison cohort (solid line) and nonapnea sleep disorder (NSD) cohort (dashed line).
213
strongly increased the risk of epilepsy development (aHR: 5.14, 4.70, and 2.13; 95% CI: 4.05–6.52, 4.17–5.30, and 1.89–2.39, respectively). Table 4 shows the stepwise Cox proportional hazard model analysis for estimating 11 parameters possibly associated with epilepsy development. The risk of epilepsy was 1.55-fold (95% CI: 1.39–1.71) in the NSD patients compared with the controls. Moreover, sex, age, and the comorbidities of brain tumors, stroke, head injury, hypertension, diabetes, and depression were positive predictors of epilepsy (Table 4). After adjustment for covariates in this model, the 3 comorbidities of brain tumors (aHR: 5.14), stroke (aHR: 4.63), and head injury (aHR: 2.14) were determined to interact with NSDs in contributing substantially to epilepsy development. Table 5 presents the interaction effects between NSDs and stroke (interaction P b 0.0001), head injury (interaction P = 0.6993), and brain tumors (interaction P b 0.0001) on epilepsy development. When comorbidities and NSDs coexisted, the risk of epilepsy increased markedly in patients with an NSD and stroke (aHR: 8.61, 95% CI: 7.43–9.98) or in patients with an NSD and brain tumors (aHR: 7.66, 95% CI: 5.06– 11.6) (Table 5). 4. Discussion Sleep is a highly organized and dynamic behavior regulated by the hypothalamus and maintains the internal homeostatic and circadian rhythm (Roth and Roehrs, 2000; Saper et al., 2005). Several internal or environmental factors can reduce or disrupt sleep efficiency, and SDs with symptoms of dyssomnia and insomnia have been commonly reported as comorbidities for numerous neurological and psychiatric diseases including epilepsy (Schmitt, 2015; Roth and Roehrs, 2000). However, few studies have demonstrated the causality between SDs and epilepsy, particularly regarding the predisposition of NSDs for epilepsy. As shown in Fig. 1, the risk of developing epilepsy increased in the NSD cohort as the follow-up period increased. NSD patients had a higher risk of developing subsequent epilepsy, and male patients of old age were particularly at risk. Occasionally, nocturnal seizures from frontal lobe epilepsy can be misdiagnosed as an SD with parasomnia or insomnia (Schmitt, 2015). Because we enrolled patients with NSDs during 2000–2003 and followed their cumulative incidence of epilepsy for 12 years, a patient with suspected frontal lobe epilepsy was extremely likely to have received electroencephalography for diagnosing epilepsy. Thus, a prolonged delay in diagnosing epilepsy was unlikely in this study, although the reverse causality could not be completely ruled out theoretically. Furthermore, our follow-up period was measured in both cohorts from the index date until the development of epilepsy, withdrawal from the follow-up in LHID database, or December 31, 2011, which ever occurred first. Because of the obviously fewer comorbidities and lower epilepsy occurrence in the comparison cohort, the earlier withdrawal before the end of 2011 in the comparison cohort might have been noted more often, probably resulting in the shorter mean follow-up time. We unexpectedly observed that the risk of epilepsy was higher in the NSD cohort than in the comparison cohort, regardless of whether there were comorbidities. The NSD patients with comorbidities had a 1.51-fold risk of epilepsy compared with the controls with comorbidities, and NSD patients without comorbidities exhibited a 2.66-fold risk of epilepsy compared with the controls without comorbidities. Further interpretation let us realize that the aHRs of 1.51 and 2.66 cannot be compared with each other, due to that their comparison groups were not the same. Table 2 shows that subjects with comorbidities had a higher incidence rate of epilepsy than that of subjects without comorbidities in both cohorts. However, the inconsistent aHRs in the cohorts with and without comorbidities demonstrate that these comorbidities were not major confounders or mediators driving the relationship between NSDs and epilepsy development in this study. Moreover, again, it was unexpected to observe that the crude HRs for hyperlipidemia and anxiety to develop epilepsy were 1.26 and 1.47 respectively, and
214
T. Harnod et al. / Preventive Medicine 91 (2016) 211–216
Table 2 Incidence rates and adjusted hazard ratios of epilepsy stratified by sex, age, comorbidities (yes/no), and drug use (yes/no) between the NSD and comparison cohorts. Variables
Overalla Sexb Female Male Age, yearc 20–45 45–65 ≥65 Comorbidityd No Yes Drug usee No Yes
Comparison cohort
NSD cohort
IRR (95% CI)
Adjusted HR (95% CI)
1.47
1.92 (1.87–1.98)⁎
1.52 (1.37–1.69)⁎
382,326 203,458
1.17 2.02
1.90 (1.82–1.97)⁎ 1.96 (1.87–2.06)⁎
1.51 (1.30–1.74)⁎ 1.53 (1.32–1.78)⁎
239 267 353
255,279 223,771 106,733
0.94 1.19 3.31
3.13 (2.98–3.29)⁎ 1.73 (1.65–1.82)⁎ 1.60 (1.50–1.70)⁎
1.99 (1.57–2.51)⁎ 1.29 (1.07–1.55)⁎⁎ 1.50 (1.28–1.75)⁎
0.21 1.29
71 788
135,846 449,938
0.52 1.75
2.52 (2.38–2.66)⁎ 1.36 (1.31–1.41)⁎
2.66 (1.95–3.63)⁎ 1.51 (1.37–1.68)⁎
0.44 0.87
6 853
19,059 566,725
0.31 1.51
0.72 (0.59–0.86)⁎ 1.73 (1.68–1.79)⁎
1.13 (0.49–2.59) 1.54 (1.39–1.72)⁎
Event
PY
Rate
Event
PY
Rate
877
1,149,892
0.76
859
585,783
461 416
745,721 404,172
0.62 1.03
448 411
150 307 420
501,590 445,313 202,989
0.30 0.69 2.07
116 761
558,294 591,598
124 753
282,203 867,689
NSD: nonapnea sleep disorder; PY, person-years; rate, incidence rate (per 1000 person-years); IRR, incidence rate ratio. ⁎ P b 0.001. ⁎⁎ P b 0.01. a Adjusted HR: adjusted for sex, continuous variable of age, each comorbidity, and drug use (no/yes). b Adjusted HR: adjusted for continuous variable of age, each comorbidity, and drug use (no/yes). c Adjusted HR: adjusted for sex, each comorbidity, and drug use (no/yes). d Adjusted HR: adjusted for sex, continuous variable of age, and drug use (no/yes). e Adjusted HR: adjusted for sex, continuous variable of age, and each comorbidity.
Table 3 Adjusted hazard ratios of epilepsy in subjects with different risk factors. Variables
Event
Crude HR (95% CI)
Adjusted HR (95% CI)a
Comparison cohort NSD cohort Sex Female Male Age groups, years 20–45 45–65 ≥65 Comorbidity Hypertension No Yes Hyperlipidemia No Yes Diabetes No Yes Stroke No Yes Anxiety No Yes Depression No Yes Head injury No Yes Brain tumors No Yes Drug use No Yes
877 859
1.00 1.92 (1.75–2.11)⁎
1.00 1.51 (1.36–1.67)⁎
909 827
1.00 1.69 (1.53–1.85)⁎
1.00 1.41 (1.28–1.55)⁎
389 574 773
1.00 1.67 (1.47–1.90)⁎ 4.82 (4.27–5.45)⁎
1.00 1.06 (0.91–1.22) 1.52 (1.30–1.78)⁎
554 1182
1.00 3.17 (2.87–3.51)⁎
1.00 1.34 (1.18–1.53)⁎
115 585
1.00 1.26 (1.14–1.39)⁎
1.00 0.64 (0.57–0.71)⁎
1121 615
1.00 2.23 (2.02–2.46)⁎
1.00 1.22 (1.09–1.36)⁎
748 988
1.00 7.72 (7.02–8.49)⁎
1.00 4.70 (4.17–5.30)⁎
1294 442
1.00 1.47 (1.32–1.64)⁎
1.00 0.84 (0.74–0.95)⁎⁎
1435 301
1.00 1.96 (1.73–2.21)⁎
1.00 1.22 (1.06–1.40)⁎⁎
1358 378
1.00 3.65 (3.26–4.09)⁎
1.00 2.13 (1.89–2.39)⁎
1664 72
1.00 9.19 (7.26–11.6)⁎
1.00 5.14 (4.05–6.52)⁎
130 1606
1.00 2.59 (2.17–3.10)⁎
1.00 1.19 (0.99–1.44)
NSD: nonapnea sleep disorder. a Adjusted HR: adjusted for sex, age groups (including 20–45, 45–65, and ≥65 years), each comorbidity (including hypertension, hyperlipidemia, diabetes, stroke, anxiety, depression, head injury, and brain tumors), and drug use (no/yes). ⁎ P b 0.001. ⁎⁎ P b 0.01.
the aHRs for hyperlipidemia and anxiety were lower than 1.00 after adjustment for age, sex, comorbidities, and drug use. Regarding our understanding, patients with hyperlipidemia and anxiety may more frequently use a medical care system or take BZD in Taiwan, probably reducing the epilepsy risk. However, the controversial trend in these 2 comorbid disorders is probably linked to more complicated interaction with others to affect epilepsy development; hence, future studies are necessary to analyze these details further. Overall, long-term sleep hygiene may be more necessary than considered previously for stabilizing the excitatory–inhibitory CNS balance (Cobabe et al., 2015). We suggest that the presence of NSDs constitute a predisposing, possibly independent factor for the development of future epilepsy, but it is likely weaker than brain tumors, stroke, and head injury. Table 4 lists 11 variables having a significant association with epilepsy. However, there were 3 comorbidities predisposing for symptomatic epilepsy — brain tumors, stroke, and head injury — with NSDs being found to contribute substantially to this model. Table 5 presents the aHRs of epilepsy with the joint effects of NSDs and these 3 comorbidities. The statistically significant P value for interaction would reveal how 2 independent variables interact with each other if the effect from one variable differs on the level of the other. There might be a specific interaction when the magnitude of a variable effect is much greater than the effect of another variable. For example, concerning the relationship between NSDs and brain tumors in this study, brain tumors and NSDs respectively contributed nearly 12.5- and 2-fold risks of epilepsy. However, NSD patients with brain tumors had only a 7.66-fold Table 4 Parameters for epilepsy estimated using the stepwise Cox proportional hazard model. Variable
Adjusted HR (95% CI)
Estimate Standard error P value
Brain tumors Stroke Head injury NSD Sex Age Hypertension Hyperlipidemia Diabetes Anxiety Depression
5.14 4.63 2.14 1.55 1.41 1.01 1.28 0.63 1.21 0.84 1.23
1.64 1.53 0.76 0.44 0.34 0.01 0.24 −0.46 0.19 −0.17 0.21
(4.05–6.52) (4.11–5.21) (1.91–2.41) (1.39–1.71) (1.28–1.55) (1.00–1.02) (1.12–1.46) (0.57–0.71) (1.08–1.35) (0.75–0.95) (1.07–1.41)
0.121 0.061 0.059 0.053 0.049 0.002 0.067 0.056 0.056 0.062 0.069
NSD: nonapnea sleep disorder; presented c-statistic was 0.785.
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 0.0003 b0.0001 0.0008 0.0056 0.0028
T. Harnod et al. / Preventive Medicine 91 (2016) 211–216 Table 5 Adjusted hazard ratios of epilepsy associated with the interaction of NSDs with stroke, head injury, and brain tumors. Variable
N
Event
Adjusted HRd (95% CI)
P value b0.0001a
NSD No No Yes Yes
Stroke No Yes No Yes
110,455 17,273 50,519 13,346
357 520 391 468
1.00 7.72 (6.66–8.94)⁎ 2.41 (2.09–2.78)⁎ 8.61 (7.43–9.98)⁎
NSD No No Yes Yes
Head injury No Yes No Yes
119,783 7945 57,887 5978
717 160 641 218
1.00 2.68 (2.26–3.19)⁎ 1.80 (1.62–2.01)⁎ 5.06 (4.35–5.89)⁎
NSD No No Yes Yes
Brain tumors No Yes No Yes
127,205 523 63,442 423
828 49 836 23
1.00 12.5 (9.40–16.7)⁎ 1.97 (1.79–2.17)⁎ 7.66 (5.06–11.6)⁎
0.6993b
b0.0001c
NSD: nonapnea sleep disorder. ⁎ P b 0.001. a P value for interaction between NSDs and stroke. b P value for interaction between NSDs and head injury. c P value for interaction between NSDs and brain tumors. d Adjusted HR: adjusted for continuous variables of age and sex.
risk of epilepsy development. Such an interaction is due to our measured event being epilepsy occurrence, and the patients suffering from brain tumors and NSDs might have had premature mortality before epilepsy had the chance to develop. From this data, we could determine why the interaction P value was crucial for describing the relationship between these 2 variables. However, NSDs and head injury respectively contributed nearly 3- and 2-fold risks of epilepsy, and the aHR of epilepsy was 5.06 in NSD patients with head injury. Nevertheless, there was no significant P value for the interaction (0.6993) between these 2 variables. Therefore, physicians should be particularly aware of the specifically higher risk of developing epilepsy when treating aged patients with NSDs and either stroke or brain tumors. The hypothalamus not only shuts off the arousal system and stabilizes the switch from wakefulness to sleep (Saper et al., 2005), but is also involved in certain episodic brain disorders because it integrates endocrine regulation, autonomic control, and various body rhythms (Overeem et al., 2002). Some recent studies have reported that shorter sleep duration, lower sleep efficiency, and insomnia are associated with changes in hypothalamic–pituitary–adrenal axis through dysregulation of the axis (Zhang et al., 2014; Castro-Diehl et al., 2015). Moreover, SDrelated chronic stress was associated with increased seizure susceptibility through alterations in excitatory–inhibitory transmission, suppressed neurogenesis, and hippocampal damage (Maguire and Salpekar, 2013). Concerning epileptogenesis, regardless of whether the causes are idiopathic, cryptogenic, or symptomatic, sex-related differences of epilepsy development in the brain remain unclear. Epileptogenesis is a process through which a healthy brain becomes predisposed to epileptic seizures, and excessive cortical excitability is the current widely accepted pathogenesis. Chauvel et al. reported that one of the excitatory transmissions, the glutamate–NMDA receptor, was more easily suppressed in female rats than in male rats (Chauvel et al., 2012). Moreover, some transcranial magnetic stimulation studies on human patients have demonstrated that the decrease of cortical excitability through aging may be more evident in women, whereas excitability is more preserved in aging men (Pitcher et al., 2003; Smith et al., 2011). These differences of glutamate transmission and aging changes of cortex excitability between the sexes might render older men with NSDs more vulnerable to subsequent epilepsy development. The strength of this study was its nationwide population-based design and representativeness. However, this study had some limitations. First, we could not contact the patients directly to confirm their type of
215
NSD or the type of epilepsy for determining whether focal or generalized epilepsy was common in some types of NSD patients, because the NHIRD contains anonymized personal information and identification numbers. Similarly, data on etiology and frequency of epilepsy, presence or absence of aura, family history, and actual use of antiepileptic drugs were unavailable. It would have been meaningful to differentiate idiopathic epilepsy from secondary epilepsy among our study subjects, although we had adjusted the 3 major etiologies for symptomatic epilepsy in this study. Moreover, further analysis between the subtypes of NSDs and their risks of developing epilepsy might have offered more data to strengthen the results. Second, although our study design included adequate controls for numerous confounding factors, bias could be present because of the presence of unmeasured or unknown confounders. However, research has confirmed the high accuracy and validity of diagnoses recorded using ICD-9-CM codes in various NHIRD studies (Cheng et al., 2011; Harnod et al., 2015; Yeh et al., 2013), suggesting that reliable evidence has been obtained in similar study designs and that the results of the present study are valuable for understanding the causality between NSDs and epilepsy.
5. Conclusion This population-based cohort study revealed a higher risk of epilepsy development in patients with NSDs and that this risk is particularly high for men and older patients. These findings suggest that NSDs constitute a predisposing, possibly independent factor for developing subsequent epilepsy in adulthood. Additional large, unbiased, and population-based studies are necessary before confirmatory conclusions can be drawn.
Author contributions Conception/design: Tomor Harnod, Chia-Hung Kao Provision of study material and patients: Chia-Hung Kao Collection and assembly of data: All authors Data analysis and interpretation: All authors Manuscript writing: All authors Final approval of manuscript: All authors.
Conflict of interest All the authors declare no conflict of interest.
Transparency document The Transparency document associated with this article can be found, in online version.
Acknowledgments This study is supported in part by the Taiwan Ministry of Health and Welfare Clinical Trial and Research Center of Excellence (MOHW105-TDU-B-212-133019), China Medical University (CMU) Hospital, Academia Sinica Taiwan Biobank Stroke Biosignature Project (BM10501010037), NRPB Stroke Clinical Trial Consortium (MOST 104-2325-B-039-005), Tseng-Lien Lin Foundation, Taichung, Taiwan, Taiwan Brain Disease Foundation, Taipei, Taiwan, and Katsuzo and Kiyo Aoshima Memorial Funds, Japan. Support was also received from CMU under the Aim for Top University Plan of the Ministry of Education, Taiwan. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. No additional external funding was received for this study.
216
T. Harnod et al. / Preventive Medicine 91 (2016) 211–216
References Annegers, J.F., Rocca, W.A., Hauser, W.A., 1996. Causes of epilepsy: contributions of the Rochester epidemiology project. Mayo Clin. Proc. 71, 570–575. Aurlien, D., Larsen, J.P., Gjerstad, L., et al., 2012. Comorbid and underlying diseases - major determinants of excess mortality in epilepsy. Seizure 21, 573–577. Benbadis, S.R., Allen, H.W., 2000. An estimate of the prevalence of psychogenic non-epileptic seizures. Seizure 9, 280–281. Bostock, E.C., Kirkby, K.C., Garry, M.I., et al., 2015. Comparison of precipitating factors for mania and partial seizures: indicative of shared pathophysiology? J. Affect. Disord. 183, 57–67. Castro-Diehl, C., Diez Roux, A.V., Redline, S., et al., 2015. Association of sleep duration and quality with alterations in the hypothalamic-pituitary adrenocortical axis: the MultiEthnic Study of Atherosclerosis (MESA). J. Clin. Endocrinol. Metab. 100, 3149–3158. Chang, Y.H., Ho, W.C., Tsai, J.J., et al., 2012. Risk of mortality among patients with epilepsy in southern Taiwan. Seizure 21, 254–259. Chauvel, V., Vamos, E., Pardutz, A., et al., 2012. Effect of systemic kynurenine on cortical spreading depression and its modulation by sex hormones in rat. Exp. Neurol. 236, 207–214. Cheng, C.L., Kao, Y.H., Lin, S.J., et al., 2011. Validation of the National Health Insurance Research Database with ischemic stroke cases in Taiwan. Pharmacoepidemiol. Drug Saf. 20, 236–242. Cobabe, M.M., Sessler, D.I., Nowacki, A.S., et al., 2015. Impact of sleep duration on seizure frequency in adults with epilepsy: a sleep diary study. Epilepsy Behav. 43, 143–148. Harnod, T., Wang, Y.C., Kao, C.H., 2015. Association between benzodiazepine use and epilepsy occurrence: a nationwide population-based case–control study. Medicine 94, e1571.
Hauser, W.A., Annegers, J.F., Kurland, L.T., 1991. Prevalence of epilepsy in Rochester, Minnesota: 1940–1980. Epilepsia 32, 429–445. Lucey, B.P., Leahy, A., Rosas, R., et al., 2015. A new model to study sleep deprivation-induced seizure. Sleep 38, 777–785. Maguire, J., Salpekar, J.A., 2013. Stress, seizures, and hypothalamic–pituitary–adrenal axis targets for the treatment of epilepsy. Epilepsy Behav. 26, 352–362. Neligan, A., Bell, G.S., Johnson, A.L., et al., 2011. The long-term risk of premature mortality in people with epilepsy. Brain 134, 388–395. Ohayon, M.M., 2009. Observation of the natural evolution of insomnia in the American general population cohort. Sleep Med. Clin. 4, 87–92. Overeem, S., van Vliet, J.A., Lammers, G.J., et al., 2002. The hypothalamus in episodic brain disorders. Lancet Neurol. 1, 437–444. Pitcher, J.B., Ogston, K.M., Miles, T.S., 2003. Age and sex differences in human motor cortex input-output characteristics. J. Physiol. 546, 605–613. Roth, T., Roehrs, T., 2000. Sleep organization and regulation. Neurology 54, S2–S7. Saper, C.B., Scammell, T.E., Lu, J., 2005. Hypothalamic regulation of sleep and circadian rhythms. Nature 437, 1257–1263. Schmitt, B., 2015. Sleep and epilepsy syndromes. Neuropediatrics 46, 171–180. Smith, A.E., Sale, M.V., Higgins, R.D., et al., 2011. Male human motor cortex stimulus-response characteristics are not altered by aging. J. Appl. Physiol. 110, 206–212. Yeh, C.C., Wang, H.H., Chou, Y.C., et al., 2013. High risk of gastrointestinal hemorrhage in patients with epilepsy: a nationwide cohort study. Mayo Clin. Proc. 88, 1091–1098. Zhang, J., Lam, S.P., Li, S.X., et al., 2014. A community-based study on the association between insomnia and hypothalamic-pituitary-adrenal axis: sex and pubertal influences. J. Clin. Endocrinol. Metab. 99, 2277–2287.
Contents lists available at ScienceDirect
Preventive Medicine journal homepage: www.elsevier.com/locate/ypmed
Men with nonapnea sleep disorders have a high risk of developing subsequent epilepsy: A nationwide population-based cohort study Tomor Harnod, MD, PhD a,b, Yu-Chiao Wang, MSc c,d, Chia-Hung Kao, MD e,f,g,⁎ a
Department of Neurosurgery, Hualien Tzu Chi General Hospital, Buddhist Tzu Chi Medical Foundation, Hualien, Taiwan College of Medicine, Tzu Chi University, Hualien, Taiwan Management Office for Health Data, China Medical University Hospital, Taichung, Taiwan d College of Medicine, China Medical University, Taichung, Taiwan e Graduate Institute of Clinical Medical Science and School of Medicine, College of Medicine, China Medical University, Taichung, Taiwan f Department of Nuclear Medicine and PET Center, China Medical University Hospital, Taichung, Taiwan g Department of Bioinformatics and Medical Engineering, Asia University, Taichung, Taiwan b c
a r t i c l e
i n f o
Article history: Received 7 January 2016 Received in revised form 6 August 2016 Accepted 24 August 2016 Available online 25 August 2016 Keywords: Cohort study Epilepsy Insomnia National Health Insurance Sleep disorder
a b s t r a c t Objective. This nationwide population-based cohort study evaluated the effects of nonapnea sleep disorders (NSDs) on the development of epilepsy. Methods. We identified 63,865 patients aged ≥20 years, diagnosed with NSDs (ICD-9-CM: 307.4 or 780.5), and without coding for apnea-related sleep disorders (ICD-9-CM: 780.51, 780.53, or 780.57) during 2000– 2003 as the NSD cohort. In addition, we enrolled a comparison cohort of 127,728 patients. We calculated the adjusted hazard ratio (aHR) for developing epilepsy (ICD-9-CM: 345) after adjustment for age, sex, comorbidities, and drug use. A Kaplan–Meier analysis was used to measure the cumulative incidence of epilepsy between the 2 groups until the end of 2011. Results. The cumulative incidence of epilepsy was significantly higher in the NSD cohort than in the comparison cohort. The aHR for developing epilepsy in the NSD cohort was 1.52 (95% CI = 1.37–1.69). The risk of developing epilepsy was higher among males (aHR = 1.41) than among females. The age-stratified effects of NSDs on developing epilepsy were the highest among patients aged ≥65 years. When comorbidities and NSDs coexisted, the risk of epilepsy was specifically increased in patients having an NSD and stroke (aHR: 8.61, 95% CI: 7.43–9.98) in addition to brain tumors (aHR: 7.66, 95% CI: 5.06–11.6). Conclusion. This study indicated that patients with NSDs have a higher risk of developing epilepsy and that the risk is much higher among men and older patients. These findings suggest that NSDs constitute a predisposing, possibly independent factor for developing subsequent epilepsy in adulthood. © 2016 Elsevier Inc. All rights reserved.
1. Introduction Both epilepsy and sleep disorders (SDs) are common neurological disorders, and either one of them can affect the quality of daily life in affected patients. Epilepsy has a prevalence of 0.5%–1% in the general population worldwide (Hauser et al., 1991; Benbadis and Allen, 2000), and epilepsy exerts a more serious effect on patients' lives than SDs do, such as the high risks of premature death from heart disease and stroke (Neligan et al., 2011; Aurlien et al., 2012). A 2.5-fold mortality risk, compared with the general Taiwanese population, was recently reported for Southern Taiwanese patients with epilepsy (Chang et al., 2012).
⁎ Corresponding author at: Graduate Institute of Clinical Medical Science and School of Medicine, College of Medicine, China Medical University, No. 2, Yuh-Der Road, Taichung 404, Taiwan. E-mail address: [email protected] (C.-H. Kao).
http://dx.doi.org/10.1016/j.ypmed.2016.08.034 0091-7435/© 2016 Elsevier Inc. All rights reserved.
The causes of epilepsy indicate that the disease either has an idiopathic origin or appears as symptomatic epilepsy secondarily following a brain insult. Numerous causes lead to symptomatic epilepsy in adulthood, with the 3 major causes being head injury, stroke, and brain tumors (Annegers et al., 1996). According to current knowledge, a seizure develops from an imbalanced expression of the excitatory and inhibitory ion channels in the central nervous system (CNS). The electrophysiological instability in the cortex presents as a spike in an electroencephalogram, and the hypersynchronization of spikes may develop into an epileptic seizure. Sleep deprivation is a well-known factor for increasing seizure susceptibility and often provokes epileptic seizures (Bostock et al., 2015; Lucey et al., 2015; Schmitt, 2015). Sleep is vital for living. SDs result in profound consequences for physiology, behavior, and daily abilities during waking hours (Ohayon, 2009). Symptoms of dyssomnia or insomnia may be temporary or chronic, depending on the type of SD. Currently, nonapnea SDs (NSDs)
212
T. Harnod et al. / Preventive Medicine 91 (2016) 211–216
without evidence of hypoxic damage to the CNS are considered less threatening to human health. In this study, we determined whether NSDs are independent and predisposing factors that lead to adulthood epilepsy and subsequently affect the life quality and life-span of affected patients by enrolling an NSD cohort from Taiwan's nationwide population-based database.
2. Methods and materials 2.1. Data source The Taiwan Bureau of National Health Insurance developed the National Health Insurance (NHI) program, which has been in effect since March 1995. This program offers full medical service to more than 99% of Taiwan residents (approximately 23 million). The claims data on the Taiwan NHI program constitute a large computerized database called the National Health Insurance Research Database (NHIRD). We used the Longitudinal Health Insurance Database (LHID 2000), a subset of the NHIRD, in this population-based cohort study. The LHID 2000 contains claims data on 1 million randomly selected beneficiaries from the 23 million insured individuals who registered during 1996–2011. Demographic data (such as those on birth and sex) and diagnostic data (such as the date and diagnoses of outpatient visits, and the date of admission and discharge with diagnoses) of all insured individuals are included in this database. In our study, International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes were recorded by a general physician, a psychiatrist, or a neurologist and were used to define diseases and medical procedures. The NHIRD is highly representative of the general population of Taiwan because the NHI is a single-payer and universal program. All insurance claims are scrutinized by medical reimbursement specialists and subjected to peer review according to standard diagnostic criteria. In addition, physicians and hospitals are penalized for incorrect diagnoses and codes. Therefore, the diagnoses of sleep disorders, epilepsy, and comorbidities in this study are highly reliable (Cheng et al., 2011; Harnod et al., 2015). Furthermore, the NHIRD data are anonymized for protecting personal privacy. This study was approved by the Institutional Review Board of China Medical University.
2.3. Data analysis The SAS 9.4 statistical package (SAS Institute Inc., NC, USA) was used for the statistical analyses, and 2-tailed P b 0.05 was considered significant. The baseline distribution of demographic variables and potential risk factors (history of comorbidities and drug use) in the NSD and comparison cohorts were analyzed using a chi-square test for categorical variables and a Student t-test for continuous variables. The incidence rates of epilepsy development in both cohorts were calculated after stratifying each variable. The incidence rate ratio (IRR) and 95% confidence intervals (95% CIs) of epilepsy were presented using the Poisson regression model. The crude hazard ratios (HRs) and 95% CIs of epilepsy in the NSD and comparison cohorts were calculated using the Cox proportional hazard regression model. After adjustment for sex, age, each comorbidity (including hypertension, hyperlipidemia, diabetes, stroke, anxiety, depression, head injury, and brain tumors), and drug use (no/ yes), the association between NSDs and epilepsy was evaluated using a multivariate Cox proportional hazard model to estimate the adjusted HR (aHR) and 95% CI. We also used the stepwise Cox proportional hazard model to estimate the risk of epilepsy associated with NSDs. The parameters of sex, age, each comorbidity, and drug use (no/yes) were selected stepwise. In addition, we conducted a Kaplan–Meier analysis on the cumulative incidence of epilepsy in the 2 cohorts by using R software (R Foundation for Statistical Computing, Vienna, Austria) and estimated the difference between the 2 cumulative incidence curves by using a logrank test.
3. Results The baseline characteristics of demographic factors in the NSD and comparison cohorts are presented in Table 1. We enrolled 63,865 patients in the NSD cohort and 127,728 patients in the comparison cohort. The distributions of sex and age were similar in the 2 cohorts, with female being the predominant sex (63.7%). Moreover, young to middle-
Table 1 Demographics, comorbidities, and history of drug use in the NSD and comparison cohorts.
2.2. Selection of study patients Our study enrolled 63,865 patients aged ≥20 years and newly diagnosed with NSDs (ICD-9-CM: 307.4, 780.5, excluding780.51, 780.53, and 780.57) during 2000–2003. The index date was defined as when a patient received a primary diagnosis of an NSD for the first time. Through frequency matching with age (in 5-year intervals), sex, and the index year, 2 patients without SD were randomly selected from the LHID 2000 for each patient with an NSD to form the comparison cohort (n = 127,728). Patients who developed epilepsy (ICD-9-CM: 345) before the index date and those aged b20 years were excluded. The follow-up period was measured in both cohorts from the index date until the development of epilepsy, withdrawal from the LHID database, or December 31, 2011, which ever occurred first. Drug exposure histories in the subjects of this study were measured before the end date, and their types of exposure were categorized as zolpidem, benzodiazepine (BZD), or both. In addition, we adjusted for comorbidities. The comorbidities were defined as diseases diagnosed before the study end date, including hypertension (ICD-9-CM: 401–405), hyperlipidemia (272), diabetes (250), anxiety (300.0, 300.2, 300.3, 308.3, and 309.81), depression (296.2, 296.3, 296.82, 300.4, 309.0, 309.1, 3092.8, and 311), head injury (850–854, and 959.01), stroke (430–438), and brain tumors (225, 191, 192, 194.3, and 194.4).
Sex Female Male Age, year 20–45 45–65 ≥65 Mean (standard deviation)a Comorbidity Hypertension Hyperlipidemia Diabetes Stroke Anxiety Depression Head injury Brain tumors Drug use Non Zolpidem BZD Both Mean follow-up years (standard deviation)a
Comparison cohort (N = 127,728)
NSD cohort (N = 63,865)
n
%
n
%
81,390 46,338
63.7 36.3
40,695 23,170
63.7 36.3
P value
0.99
0.99 53,742 42.1 47,354 37.1 26,632 20.9 49.8 (16.2)
26,871 42.1 23,677 37.1 13,317 20.9 49.8 (16.2)
0.45
47,515 30,805 23,812 17,273 10,749 4394 7945 523
30,926 22,565 15,156 13,346 23,916 13,716 5978 423
48.4 35.3 23.7 20.9 37.4 21.5 9.36 0.66
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001
2079 3.26 406 0.64 21,292 33.3 40,088 62.8 9.17 (2.25)
b0.0001
37.2 24.1 18.6 13.5 8.42 3.44 6.22 0.41
31,458 24.6 1005 0.79 77,617 60.8 17,648 13.8 9.00 (2.44)
NSD: nonapnea sleep disorder; BZD: benzodiazepine; chi-square test. a Student t-test.
b0.0001
T. Harnod et al. / Preventive Medicine 91 (2016) 211–216
aged patients were predominant (79.2% with 20–65 years), with the mean age being 49.8 years (standard deviation: 16.2 years) in both cohorts. More NSD patients had comorbidities and drug use than did those in the comparison cohort, and the NSD cohort had a significantly longer mean follow-up period than that of the comparison cohort (9.17 vs 9.00 years) (chi-square P b 0.0001, Table 1). During the follow-up period, the NSD cohort revealed a higher cumulative incidence of developing epilepsy than that of the comparison cohort (log-rank test P b 0.0001, Fig. 1). The overall IRR of epilepsy was 1.92-fold in the NSD cohort compared with that in the comparison cohort (1.47 per 1000 person-years vs 0.76 per 1000 person-years, Table 2). After adjustment for age, sex, each comorbidity, and drug use, the NSD cohort had a 1.52-fold risk (95% CI: 1.37–1.69, P b 0.001) of developing epilepsy (aHR: 1.51 and 1.53, 95% CI: 1.30–1.74 and 1.32–1.78 for female and male patients, respectively). The male patients in the NSD cohort had a higher incidence rate of epilepsy than that of the females (1.17 vs 2.02 per 1000 person-years for female and male patients, respectively). The incidence rate of epilepsy increased with age, but the NSD patients aged 20–45 years clearly had a higher risk of epilepsy development than controls aged 20–45 years (aHR: 1.99, 95% CI: 1.57–2.51). The comorbidity-stratified effects of NSDs on epilepsy were the highest in patients without comorbidities (aHR: 2.66 and 1.51, 95% CI: 1.95–3.63 and 1.37–1.68 for those without comorbidities and those with one or more comorbidities, respectively). A significantly higher association between NSDs and epilepsy development was observed among the NSD patients with drug use compared with the controls with drug use (aHR: 1.54, 95% CI: 1.39–1.72; Table 2). Table 3 presents the individual effects of different factors on the risk of epilepsy development. The significantly increased risk factors for epilepsy were male sex, old age, hypertension, diabetes, stroke, depression, head injury, and brain tumors. After adjustment for age, sex, each comorbidity, and drug use, a low risk of epilepsy was revealed in subjects with hyperlipidemia (aHR: 0.64, 95% CI: 0.57–0.71) and anxiety (aHR: 0.84, 95% CI: 0.74–0.95). Moreover, drug use did not significantly influence epilepsy development. Regarding sex, male patients were at a higher risk of epilepsy (aHR: 1.41, 95% CI: 1.28–1.55) than female patients. Compared with that in the young adults aged 20–45 years, the aHR of epilepsy in the elderly patients aged ≥65 years significantly increased by 1.52-fold (95% CI: 1.30–1.78) (Table 3). The 3 comorbidities related to epilepsy etiology — brain tumors, stroke, and head injury —
Fig. 1. Cumulative incidence of epilepsy in the comparison cohort (solid line) and nonapnea sleep disorder (NSD) cohort (dashed line).
213
strongly increased the risk of epilepsy development (aHR: 5.14, 4.70, and 2.13; 95% CI: 4.05–6.52, 4.17–5.30, and 1.89–2.39, respectively). Table 4 shows the stepwise Cox proportional hazard model analysis for estimating 11 parameters possibly associated with epilepsy development. The risk of epilepsy was 1.55-fold (95% CI: 1.39–1.71) in the NSD patients compared with the controls. Moreover, sex, age, and the comorbidities of brain tumors, stroke, head injury, hypertension, diabetes, and depression were positive predictors of epilepsy (Table 4). After adjustment for covariates in this model, the 3 comorbidities of brain tumors (aHR: 5.14), stroke (aHR: 4.63), and head injury (aHR: 2.14) were determined to interact with NSDs in contributing substantially to epilepsy development. Table 5 presents the interaction effects between NSDs and stroke (interaction P b 0.0001), head injury (interaction P = 0.6993), and brain tumors (interaction P b 0.0001) on epilepsy development. When comorbidities and NSDs coexisted, the risk of epilepsy increased markedly in patients with an NSD and stroke (aHR: 8.61, 95% CI: 7.43–9.98) or in patients with an NSD and brain tumors (aHR: 7.66, 95% CI: 5.06– 11.6) (Table 5). 4. Discussion Sleep is a highly organized and dynamic behavior regulated by the hypothalamus and maintains the internal homeostatic and circadian rhythm (Roth and Roehrs, 2000; Saper et al., 2005). Several internal or environmental factors can reduce or disrupt sleep efficiency, and SDs with symptoms of dyssomnia and insomnia have been commonly reported as comorbidities for numerous neurological and psychiatric diseases including epilepsy (Schmitt, 2015; Roth and Roehrs, 2000). However, few studies have demonstrated the causality between SDs and epilepsy, particularly regarding the predisposition of NSDs for epilepsy. As shown in Fig. 1, the risk of developing epilepsy increased in the NSD cohort as the follow-up period increased. NSD patients had a higher risk of developing subsequent epilepsy, and male patients of old age were particularly at risk. Occasionally, nocturnal seizures from frontal lobe epilepsy can be misdiagnosed as an SD with parasomnia or insomnia (Schmitt, 2015). Because we enrolled patients with NSDs during 2000–2003 and followed their cumulative incidence of epilepsy for 12 years, a patient with suspected frontal lobe epilepsy was extremely likely to have received electroencephalography for diagnosing epilepsy. Thus, a prolonged delay in diagnosing epilepsy was unlikely in this study, although the reverse causality could not be completely ruled out theoretically. Furthermore, our follow-up period was measured in both cohorts from the index date until the development of epilepsy, withdrawal from the follow-up in LHID database, or December 31, 2011, which ever occurred first. Because of the obviously fewer comorbidities and lower epilepsy occurrence in the comparison cohort, the earlier withdrawal before the end of 2011 in the comparison cohort might have been noted more often, probably resulting in the shorter mean follow-up time. We unexpectedly observed that the risk of epilepsy was higher in the NSD cohort than in the comparison cohort, regardless of whether there were comorbidities. The NSD patients with comorbidities had a 1.51-fold risk of epilepsy compared with the controls with comorbidities, and NSD patients without comorbidities exhibited a 2.66-fold risk of epilepsy compared with the controls without comorbidities. Further interpretation let us realize that the aHRs of 1.51 and 2.66 cannot be compared with each other, due to that their comparison groups were not the same. Table 2 shows that subjects with comorbidities had a higher incidence rate of epilepsy than that of subjects without comorbidities in both cohorts. However, the inconsistent aHRs in the cohorts with and without comorbidities demonstrate that these comorbidities were not major confounders or mediators driving the relationship between NSDs and epilepsy development in this study. Moreover, again, it was unexpected to observe that the crude HRs for hyperlipidemia and anxiety to develop epilepsy were 1.26 and 1.47 respectively, and
214
T. Harnod et al. / Preventive Medicine 91 (2016) 211–216
Table 2 Incidence rates and adjusted hazard ratios of epilepsy stratified by sex, age, comorbidities (yes/no), and drug use (yes/no) between the NSD and comparison cohorts. Variables
Overalla Sexb Female Male Age, yearc 20–45 45–65 ≥65 Comorbidityd No Yes Drug usee No Yes
Comparison cohort
NSD cohort
IRR (95% CI)
Adjusted HR (95% CI)
1.47
1.92 (1.87–1.98)⁎
1.52 (1.37–1.69)⁎
382,326 203,458
1.17 2.02
1.90 (1.82–1.97)⁎ 1.96 (1.87–2.06)⁎
1.51 (1.30–1.74)⁎ 1.53 (1.32–1.78)⁎
239 267 353
255,279 223,771 106,733
0.94 1.19 3.31
3.13 (2.98–3.29)⁎ 1.73 (1.65–1.82)⁎ 1.60 (1.50–1.70)⁎
1.99 (1.57–2.51)⁎ 1.29 (1.07–1.55)⁎⁎ 1.50 (1.28–1.75)⁎
0.21 1.29
71 788
135,846 449,938
0.52 1.75
2.52 (2.38–2.66)⁎ 1.36 (1.31–1.41)⁎
2.66 (1.95–3.63)⁎ 1.51 (1.37–1.68)⁎
0.44 0.87
6 853
19,059 566,725
0.31 1.51
0.72 (0.59–0.86)⁎ 1.73 (1.68–1.79)⁎
1.13 (0.49–2.59) 1.54 (1.39–1.72)⁎
Event
PY
Rate
Event
PY
Rate
877
1,149,892
0.76
859
585,783
461 416
745,721 404,172
0.62 1.03
448 411
150 307 420
501,590 445,313 202,989
0.30 0.69 2.07
116 761
558,294 591,598
124 753
282,203 867,689
NSD: nonapnea sleep disorder; PY, person-years; rate, incidence rate (per 1000 person-years); IRR, incidence rate ratio. ⁎ P b 0.001. ⁎⁎ P b 0.01. a Adjusted HR: adjusted for sex, continuous variable of age, each comorbidity, and drug use (no/yes). b Adjusted HR: adjusted for continuous variable of age, each comorbidity, and drug use (no/yes). c Adjusted HR: adjusted for sex, each comorbidity, and drug use (no/yes). d Adjusted HR: adjusted for sex, continuous variable of age, and drug use (no/yes). e Adjusted HR: adjusted for sex, continuous variable of age, and each comorbidity.
Table 3 Adjusted hazard ratios of epilepsy in subjects with different risk factors. Variables
Event
Crude HR (95% CI)
Adjusted HR (95% CI)a
Comparison cohort NSD cohort Sex Female Male Age groups, years 20–45 45–65 ≥65 Comorbidity Hypertension No Yes Hyperlipidemia No Yes Diabetes No Yes Stroke No Yes Anxiety No Yes Depression No Yes Head injury No Yes Brain tumors No Yes Drug use No Yes
877 859
1.00 1.92 (1.75–2.11)⁎
1.00 1.51 (1.36–1.67)⁎
909 827
1.00 1.69 (1.53–1.85)⁎
1.00 1.41 (1.28–1.55)⁎
389 574 773
1.00 1.67 (1.47–1.90)⁎ 4.82 (4.27–5.45)⁎
1.00 1.06 (0.91–1.22) 1.52 (1.30–1.78)⁎
554 1182
1.00 3.17 (2.87–3.51)⁎
1.00 1.34 (1.18–1.53)⁎
115 585
1.00 1.26 (1.14–1.39)⁎
1.00 0.64 (0.57–0.71)⁎
1121 615
1.00 2.23 (2.02–2.46)⁎
1.00 1.22 (1.09–1.36)⁎
748 988
1.00 7.72 (7.02–8.49)⁎
1.00 4.70 (4.17–5.30)⁎
1294 442
1.00 1.47 (1.32–1.64)⁎
1.00 0.84 (0.74–0.95)⁎⁎
1435 301
1.00 1.96 (1.73–2.21)⁎
1.00 1.22 (1.06–1.40)⁎⁎
1358 378
1.00 3.65 (3.26–4.09)⁎
1.00 2.13 (1.89–2.39)⁎
1664 72
1.00 9.19 (7.26–11.6)⁎
1.00 5.14 (4.05–6.52)⁎
130 1606
1.00 2.59 (2.17–3.10)⁎
1.00 1.19 (0.99–1.44)
NSD: nonapnea sleep disorder. a Adjusted HR: adjusted for sex, age groups (including 20–45, 45–65, and ≥65 years), each comorbidity (including hypertension, hyperlipidemia, diabetes, stroke, anxiety, depression, head injury, and brain tumors), and drug use (no/yes). ⁎ P b 0.001. ⁎⁎ P b 0.01.
the aHRs for hyperlipidemia and anxiety were lower than 1.00 after adjustment for age, sex, comorbidities, and drug use. Regarding our understanding, patients with hyperlipidemia and anxiety may more frequently use a medical care system or take BZD in Taiwan, probably reducing the epilepsy risk. However, the controversial trend in these 2 comorbid disorders is probably linked to more complicated interaction with others to affect epilepsy development; hence, future studies are necessary to analyze these details further. Overall, long-term sleep hygiene may be more necessary than considered previously for stabilizing the excitatory–inhibitory CNS balance (Cobabe et al., 2015). We suggest that the presence of NSDs constitute a predisposing, possibly independent factor for the development of future epilepsy, but it is likely weaker than brain tumors, stroke, and head injury. Table 4 lists 11 variables having a significant association with epilepsy. However, there were 3 comorbidities predisposing for symptomatic epilepsy — brain tumors, stroke, and head injury — with NSDs being found to contribute substantially to this model. Table 5 presents the aHRs of epilepsy with the joint effects of NSDs and these 3 comorbidities. The statistically significant P value for interaction would reveal how 2 independent variables interact with each other if the effect from one variable differs on the level of the other. There might be a specific interaction when the magnitude of a variable effect is much greater than the effect of another variable. For example, concerning the relationship between NSDs and brain tumors in this study, brain tumors and NSDs respectively contributed nearly 12.5- and 2-fold risks of epilepsy. However, NSD patients with brain tumors had only a 7.66-fold Table 4 Parameters for epilepsy estimated using the stepwise Cox proportional hazard model. Variable
Adjusted HR (95% CI)
Estimate Standard error P value
Brain tumors Stroke Head injury NSD Sex Age Hypertension Hyperlipidemia Diabetes Anxiety Depression
5.14 4.63 2.14 1.55 1.41 1.01 1.28 0.63 1.21 0.84 1.23
1.64 1.53 0.76 0.44 0.34 0.01 0.24 −0.46 0.19 −0.17 0.21
(4.05–6.52) (4.11–5.21) (1.91–2.41) (1.39–1.71) (1.28–1.55) (1.00–1.02) (1.12–1.46) (0.57–0.71) (1.08–1.35) (0.75–0.95) (1.07–1.41)
0.121 0.061 0.059 0.053 0.049 0.002 0.067 0.056 0.056 0.062 0.069
NSD: nonapnea sleep disorder; presented c-statistic was 0.785.
b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 0.0003 b0.0001 0.0008 0.0056 0.0028
T. Harnod et al. / Preventive Medicine 91 (2016) 211–216 Table 5 Adjusted hazard ratios of epilepsy associated with the interaction of NSDs with stroke, head injury, and brain tumors. Variable
N
Event
Adjusted HRd (95% CI)
P value b0.0001a
NSD No No Yes Yes
Stroke No Yes No Yes
110,455 17,273 50,519 13,346
357 520 391 468
1.00 7.72 (6.66–8.94)⁎ 2.41 (2.09–2.78)⁎ 8.61 (7.43–9.98)⁎
NSD No No Yes Yes
Head injury No Yes No Yes
119,783 7945 57,887 5978
717 160 641 218
1.00 2.68 (2.26–3.19)⁎ 1.80 (1.62–2.01)⁎ 5.06 (4.35–5.89)⁎
NSD No No Yes Yes
Brain tumors No Yes No Yes
127,205 523 63,442 423
828 49 836 23
1.00 12.5 (9.40–16.7)⁎ 1.97 (1.79–2.17)⁎ 7.66 (5.06–11.6)⁎
0.6993b
b0.0001c
NSD: nonapnea sleep disorder. ⁎ P b 0.001. a P value for interaction between NSDs and stroke. b P value for interaction between NSDs and head injury. c P value for interaction between NSDs and brain tumors. d Adjusted HR: adjusted for continuous variables of age and sex.
risk of epilepsy development. Such an interaction is due to our measured event being epilepsy occurrence, and the patients suffering from brain tumors and NSDs might have had premature mortality before epilepsy had the chance to develop. From this data, we could determine why the interaction P value was crucial for describing the relationship between these 2 variables. However, NSDs and head injury respectively contributed nearly 3- and 2-fold risks of epilepsy, and the aHR of epilepsy was 5.06 in NSD patients with head injury. Nevertheless, there was no significant P value for the interaction (0.6993) between these 2 variables. Therefore, physicians should be particularly aware of the specifically higher risk of developing epilepsy when treating aged patients with NSDs and either stroke or brain tumors. The hypothalamus not only shuts off the arousal system and stabilizes the switch from wakefulness to sleep (Saper et al., 2005), but is also involved in certain episodic brain disorders because it integrates endocrine regulation, autonomic control, and various body rhythms (Overeem et al., 2002). Some recent studies have reported that shorter sleep duration, lower sleep efficiency, and insomnia are associated with changes in hypothalamic–pituitary–adrenal axis through dysregulation of the axis (Zhang et al., 2014; Castro-Diehl et al., 2015). Moreover, SDrelated chronic stress was associated with increased seizure susceptibility through alterations in excitatory–inhibitory transmission, suppressed neurogenesis, and hippocampal damage (Maguire and Salpekar, 2013). Concerning epileptogenesis, regardless of whether the causes are idiopathic, cryptogenic, or symptomatic, sex-related differences of epilepsy development in the brain remain unclear. Epileptogenesis is a process through which a healthy brain becomes predisposed to epileptic seizures, and excessive cortical excitability is the current widely accepted pathogenesis. Chauvel et al. reported that one of the excitatory transmissions, the glutamate–NMDA receptor, was more easily suppressed in female rats than in male rats (Chauvel et al., 2012). Moreover, some transcranial magnetic stimulation studies on human patients have demonstrated that the decrease of cortical excitability through aging may be more evident in women, whereas excitability is more preserved in aging men (Pitcher et al., 2003; Smith et al., 2011). These differences of glutamate transmission and aging changes of cortex excitability between the sexes might render older men with NSDs more vulnerable to subsequent epilepsy development. The strength of this study was its nationwide population-based design and representativeness. However, this study had some limitations. First, we could not contact the patients directly to confirm their type of
215
NSD or the type of epilepsy for determining whether focal or generalized epilepsy was common in some types of NSD patients, because the NHIRD contains anonymized personal information and identification numbers. Similarly, data on etiology and frequency of epilepsy, presence or absence of aura, family history, and actual use of antiepileptic drugs were unavailable. It would have been meaningful to differentiate idiopathic epilepsy from secondary epilepsy among our study subjects, although we had adjusted the 3 major etiologies for symptomatic epilepsy in this study. Moreover, further analysis between the subtypes of NSDs and their risks of developing epilepsy might have offered more data to strengthen the results. Second, although our study design included adequate controls for numerous confounding factors, bias could be present because of the presence of unmeasured or unknown confounders. However, research has confirmed the high accuracy and validity of diagnoses recorded using ICD-9-CM codes in various NHIRD studies (Cheng et al., 2011; Harnod et al., 2015; Yeh et al., 2013), suggesting that reliable evidence has been obtained in similar study designs and that the results of the present study are valuable for understanding the causality between NSDs and epilepsy.
5. Conclusion This population-based cohort study revealed a higher risk of epilepsy development in patients with NSDs and that this risk is particularly high for men and older patients. These findings suggest that NSDs constitute a predisposing, possibly independent factor for developing subsequent epilepsy in adulthood. Additional large, unbiased, and population-based studies are necessary before confirmatory conclusions can be drawn.
Author contributions Conception/design: Tomor Harnod, Chia-Hung Kao Provision of study material and patients: Chia-Hung Kao Collection and assembly of data: All authors Data analysis and interpretation: All authors Manuscript writing: All authors Final approval of manuscript: All authors.
Conflict of interest All the authors declare no conflict of interest.
Transparency document The Transparency document associated with this article can be found, in online version.
Acknowledgments This study is supported in part by the Taiwan Ministry of Health and Welfare Clinical Trial and Research Center of Excellence (MOHW105-TDU-B-212-133019), China Medical University (CMU) Hospital, Academia Sinica Taiwan Biobank Stroke Biosignature Project (BM10501010037), NRPB Stroke Clinical Trial Consortium (MOST 104-2325-B-039-005), Tseng-Lien Lin Foundation, Taichung, Taiwan, Taiwan Brain Disease Foundation, Taipei, Taiwan, and Katsuzo and Kiyo Aoshima Memorial Funds, Japan. Support was also received from CMU under the Aim for Top University Plan of the Ministry of Education, Taiwan. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. No additional external funding was received for this study.
216
T. Harnod et al. / Preventive Medicine 91 (2016) 211–216
References Annegers, J.F., Rocca, W.A., Hauser, W.A., 1996. Causes of epilepsy: contributions of the Rochester epidemiology project. Mayo Clin. Proc. 71, 570–575. Aurlien, D., Larsen, J.P., Gjerstad, L., et al., 2012. Comorbid and underlying diseases - major determinants of excess mortality in epilepsy. Seizure 21, 573–577. Benbadis, S.R., Allen, H.W., 2000. An estimate of the prevalence of psychogenic non-epileptic seizures. Seizure 9, 280–281. Bostock, E.C., Kirkby, K.C., Garry, M.I., et al., 2015. Comparison of precipitating factors for mania and partial seizures: indicative of shared pathophysiology? J. Affect. Disord. 183, 57–67. Castro-Diehl, C., Diez Roux, A.V., Redline, S., et al., 2015. Association of sleep duration and quality with alterations in the hypothalamic-pituitary adrenocortical axis: the MultiEthnic Study of Atherosclerosis (MESA). J. Clin. Endocrinol. Metab. 100, 3149–3158. Chang, Y.H., Ho, W.C., Tsai, J.J., et al., 2012. Risk of mortality among patients with epilepsy in southern Taiwan. Seizure 21, 254–259. Chauvel, V., Vamos, E., Pardutz, A., et al., 2012. Effect of systemic kynurenine on cortical spreading depression and its modulation by sex hormones in rat. Exp. Neurol. 236, 207–214. Cheng, C.L., Kao, Y.H., Lin, S.J., et al., 2011. Validation of the National Health Insurance Research Database with ischemic stroke cases in Taiwan. Pharmacoepidemiol. Drug Saf. 20, 236–242. Cobabe, M.M., Sessler, D.I., Nowacki, A.S., et al., 2015. Impact of sleep duration on seizure frequency in adults with epilepsy: a sleep diary study. Epilepsy Behav. 43, 143–148. Harnod, T., Wang, Y.C., Kao, C.H., 2015. Association between benzodiazepine use and epilepsy occurrence: a nationwide population-based case–control study. Medicine 94, e1571.
Hauser, W.A., Annegers, J.F., Kurland, L.T., 1991. Prevalence of epilepsy in Rochester, Minnesota: 1940–1980. Epilepsia 32, 429–445. Lucey, B.P., Leahy, A., Rosas, R., et al., 2015. A new model to study sleep deprivation-induced seizure. Sleep 38, 777–785. Maguire, J., Salpekar, J.A., 2013. Stress, seizures, and hypothalamic–pituitary–adrenal axis targets for the treatment of epilepsy. Epilepsy Behav. 26, 352–362. Neligan, A., Bell, G.S., Johnson, A.L., et al., 2011. The long-term risk of premature mortality in people with epilepsy. Brain 134, 388–395. Ohayon, M.M., 2009. Observation of the natural evolution of insomnia in the American general population cohort. Sleep Med. Clin. 4, 87–92. Overeem, S., van Vliet, J.A., Lammers, G.J., et al., 2002. The hypothalamus in episodic brain disorders. Lancet Neurol. 1, 437–444. Pitcher, J.B., Ogston, K.M., Miles, T.S., 2003. Age and sex differences in human motor cortex input-output characteristics. J. Physiol. 546, 605–613. Roth, T., Roehrs, T., 2000. Sleep organization and regulation. Neurology 54, S2–S7. Saper, C.B., Scammell, T.E., Lu, J., 2005. Hypothalamic regulation of sleep and circadian rhythms. Nature 437, 1257–1263. Schmitt, B., 2015. Sleep and epilepsy syndromes. Neuropediatrics 46, 171–180. Smith, A.E., Sale, M.V., Higgins, R.D., et al., 2011. Male human motor cortex stimulus-response characteristics are not altered by aging. J. Appl. Physiol. 110, 206–212. Yeh, C.C., Wang, H.H., Chou, Y.C., et al., 2013. High risk of gastrointestinal hemorrhage in patients with epilepsy: a nationwide cohort study. Mayo Clin. Proc. 88, 1091–1098. Zhang, J., Lam, S.P., Li, S.X., et al., 2014. A community-based study on the association between insomnia and hypothalamic-pituitary-adrenal axis: sex and pubertal influences. J. Clin. Endocrinol. Metab. 99, 2277–2287.