Perinatal risk factors for SUDEP: A population-based case-control study
Tomas Andersson Karolinska Institutet conducted the statistical analyses.
Abstract
Sudden unexpected death in epilepsy (SUDEP) is a leading epilepsy-related cause of death. Researchers have highlighted the similarities between SUDEP and sudden infant death syndrome (SIDS), but perinatal risk factors such as those identified for SIDS have not been assessed previously for SUDEP. We conducted a population-based case-control study of 58 SUDEP individuals and 384 living epilepsy controls born after 1982, utilizing the Swedish Medical Birth Register together with other national health registers and individual medical records to examine if prenatal and perinatal factors are associated with SUDEP risk. We observed a 3-fold SUDEP risk increase for infants who were small for gestational age (SGA) (odds ratio [OR] 3.13; 95% confidence interval [CI] 1.05–9.30) and for those with an Apgar score of 0–6 compared to 9–10 at 10 min (OR 3.22; 95% CI 1.05–9.87). After adjusting for a number of known SUDEP risk factors, we observed that the Apgar score between 0 and 6 after 10 min had a 10-fold increased risk for SUDEP OR 10.37 (95% CI 1.49–72.01) and over a 2-fold risk for those born after the 40th gestational week (OR 2.42; 95% CI 1.03–5.65). The potential mechanisms linking low Apgar score, gestational age, and SGA to SUDEP risk remain to be explored.
1 INTRODUCTION
Sudden unexpected death in epilepsy (SUDEP) is a leading epilepsy-related cause of death.1 Several risk factors for SUDEP have been identified but the exact pathophysiology has not been clearly mapped out.2-4 SUDEP has similarities with other sudden unexpected death syndromes, and lessons of relevance for SUDEP could be learned by studying risk factors established for the other syndromes. More specifically, researchers have highlighted the similarities between SUDEP and sudden infant death syndrome (SIDS).5, 6 Like SUDEP, SIDS is essentially a diagnosis of exclusion. In both conditions, death is commonly unwitnessed and the individual is usually found dead in bed or crib in the prone position. Similar final pathways have been proposed, such as impairment of arousal associated with respiratory and cardiac failure.6 The question arise, therefore, if established perinatal risk factors for SIDS, such as prematurity, low birth weight, and maternal smoking during pregnancy7, 8 also apply for SUDEP. Because, to our knowledge, this has not been investigated before, we carried out a population-based case-control study utilizing the Swedish Medical Birth Register together with other national administrative and health registers and individual medical records to examine if prenatal and perinatal factors are associated with SUDEP risk.
2 METHODS
SUDEP is defined as sudden, unexpected, witnessed or unwitnessed, nontraumatic, and nondrowning death of patients with epilepsy with or without evidence of a seizure, excluding documented status epilepticus and in whom postmortem examination does not reveal a structural or toxicological cause for death.9 We classified SUDEP cases as definite, probable according to Annegers’ criteria,10 to facilitate comparison with previous studies.
Our study population was composed of all individuals registered at any time during 1998–2005 in the National Patient Register (NPR)11 with an International Classification of Diseases, Tenth Revision (ICD-10) code for epilepsy (G 40) (n = 78 424), who were alive on June 30, 2006 (n = 60 952). Within the study population, we identified all deaths in 2006–2011 through linkage to the National Cause of Death Register. By review of death certificates and records from caregivers and the police, 255 cases of SUDEP (167 definite, 88 probable) were identified (described in detail previously).12, 13 Five living, same-sexed, epilepsy controls per case were randomly selected from the study population by the National Board of Health and Welfare (n = 1184, after exclusion of 84 individuals deemed not to have epilepsy after review of medical records).
The Swedish Medical Birth Register (MBR) records data on all live births since 1973, but some data of relevance for our analyses, such as maternal smoking, are available only from 1982.14 Therefore, our analytical sample included only SUDEP cases (n = 58) and controls (n = 384) born in 1982 onward.
Information on covariates was collected in an identical manner for cases and controls from NPR (comorbidity), MBR (early life factors) and through careful review of medical records (epilepsy type, seizure frequency, living conditions, and intellectual disability) as in previous studies.12 Intellectual disability was defined as intelligence quotient (IQ) tested below 70, or clearly stated in the medical records that the individual had moderate to severe intellectual impairment and was not able to attend school or work due to intellectual impairment. Mild cognitive impairment or mild learning disabilities did not qualify as intellectual disability. We did not collect data on race/ethnicity.
2.1 Statistics
Statistical Analysis Software (SAS) 9.4 (SAS Institute) was used for all analyses. Characteristics were expressed as means and standard deviation (SD). The associations between SUDEP and clinical characteristics and early life factors were estimated by odds ratios (ORs) with 95% confidence intervals (CIs) calculated by logistic regression and adjusted for age and sex (model 1) and additionally for frequency of tonic-clonic seizures, living conditions, history of nocturnal tonic-clonic seizures, and intellectual disability (model 2). p-Values for mean differences were calculated from the two-sided t distribution. The chi-square distribution was used to calculate p-values for comparison of proportions and incidences.
2.2 Standard protocol approvals, registrations, and patient consent
The study was approved by the ethics committee of Karolinska Institutet, which granted that individual informed consent was not required.
3 RESULTS
Clinical characteristics and baseline data from the MBR for cases and controls are presented in Table 1. Only 3.4% of the cases were free of tonic-clonic seizures in the year leading up to death compared to 73.4% of the controls. Furthermore, the SUDEP cases had substantially more often nocturnal tonic-clonic seizures and less commonly shared a household or a bedroom than the controls (Table 1). More than two thirds of the SUDEP cases had an intellectual disability (70.7%) compared to half (47.7%) of the controls.
| Cases | Controls | p | |
|---|---|---|---|
| n | 58 | 384 | |
| Females, n (%) | 27 (46.6) | 145 (37.8) | |
| Age at epilepsy diagnosis, mean (SD) | 5.7 (5.7) | 5.7 (5.2) | .9826 |
| Epilepsy types | .2888 | ||
| Focal epilepsy, n (%) | 34 (58.6) | 236 (61.5) | |
| Focal and generalized epilepsy, n (%) | 7 (12.1) | 24 (6.3) | |
| Generalized epilepsy, n (%) | 16 (27.6) | 111 (28.9) | |
| Unknown epilepsy type, n (%) | 1 (1.7) | 13 (3.4) | |
| Patients with tonic-clonic seizures last year, n (%) | 56 (96.5) | 102 (26.6) | <.0001 |
| History of nocturnal seizures | 53 (91.4) | 191 (49.7%) | <.0001 |
| Living alone | 21 (36.2) | 47 (12.2) | <.0001 |
| Intellectual disability, n (%) | 41 (70.7) | 183 (47.7) | .0011 |
| Early life factors | |||
| Maternal age in years, mean (SD) | 38.9 (2.5) | 38.5 (3.3) | .2121 |
| Maternal smoking during pregnancy, n (%) | 17 (34.7) | 79 (22.8) | .0682 |
| Gestational age in weeks, mean (SD) | 38.9 (2.5) | 38.5 (3.3) | .2574 |
| Head circumference in cm, mean (SD) | 34.6 (2.4) | 34.4 (2.5) | .6576 |
| Birth weight in grams, mean (SD) | 3 368 (708) | 3 349 (780) | .8515 |
| Small for gestational age, n (%) | 5 (8.9) | 13 (3.4) | .1268 |
| Apgar score ≤6 at 10 min, n (%) | 5 (8.6) | 12 (3.2) | .1295 |
| Major congenital malformation, n (%) | 5 (11.1) | 36 (14.8) | .5199 |
| Parity, first born n (%) | 20 (34.5) | 164 (42.7) | .4951 |
- Note: Information on clinical characteristics was collected through review of medical records (epilepsy type, seizure frequency, living conditions, and intellectual disability) and information on early life factors through the Swedish Medical Birth Register.
- Abbreviations: SUDEP, sudden unexpected death in epilepsy.
We observed a 3-fold SUDEP risk increase for infants who were small for gestational age (OR 3.13; 95% CI 1.05–9.30) and for those with an Apgar score of 0–6 compared to 9–10 at 10 min (OR 3.22; 95% CI 1.05–9.87) (Table 2). No excess risk was seen for an Apgar score between 7 and 8. We did not observe any association between birthweight, head circumference, maternal age, parity, or presence of major congenital malformations and SUDEP risk. Maternal smoking during pregnancy inferred a nonsignificant trend for SUDEP risk (OR 1.79; 95% CI .93–3.43). The same applied to gestational age over 40 weeks (OR 1.83; 95% CI .97–3.45).
| SUDEP n (%) | Controls n (%) | Model 1a OR (95% CI) | Model 2b OR (95% CI) | |
|---|---|---|---|---|
| Maternal smoking during pregnancy | ||||
| No | 32 | 268 | 1 | 1 |
| Yes | 17 | 79 | 1.79 (0.93–3.43) | 1.92 (0.79–4.70) |
| Maternal age (years) | ||||
| ≥35 | 6 | 53 | 1 | 1 |
| 30–34 | 14 | 115 | 1.06 (0.38–2.93) | 1.13 (0.33–3.92) |
| 25–29 | 25 | 136 | 1.57 (0.61–4.07) | 1.65 (0.51–5.31) |
| <25 | 13 | 80 | 1.29 (0.45–3.64) | 2.03 (0.56–7.36) |
| Parity | ||||
| 1 | 20 (34.5) | 164 (42.7) | 1 | 1 |
| 2 | 24 (41.4) | 138 (35.9) | 1.48 (0.78–2.80) | 1.17 (0.52–2.65) |
| ≥3 | 14 (24.1) | 82 (21.4) | 1.34 (0.64–2.80) | 0.94 (0.37–2.41) |
| Gestational week | ||||
| <37c | 7 (12.3) | 55 (14.4) | 1.03 (0.43–2.48) | 1.03 (0.33–3.19) |
| 37–40 | 31 (54.4) | 244 (63.9) | 1 | 1 |
| >40 | 19 (33.3) | 83 (21.7) | 1.83 (0.97–3.45) | 2.42 (1.03–5.65) |
| Small for gestational aged | ||||
| No | 50 (89.3) | 351 (92.9) | 1 | 1 |
| Yes | 5 (8.9) | 13 (3.4) | 3.13 (1.05–9.30) | 3.00 (0.63–14.15) |
| Birth weight (grams) | ||||
| <3000 | 15 (26.3) | 82 (21.6) | 1.28 (0.66–2.48) | 1.47 (0.62–3.48) |
| 3000–3999 | 34 (59.6) | 237 (62.5) | 1 | 1 |
| ≥4000 | 8 (14.0) | 60 (15.8) | 0.90 (0.39–2.07) | 0.90 (0.29–2.75) |
| APGAR score 10 min after birth | ||||
| 9–10 | 51 (87.9) | 347 (92.0) | 1 | 1 |
| 7–8 | 2 (3.4) | 18 (4.8) | 0.84 (0.19–3.80) | 1.24 (0.22–7.05) |
| 0–6 | 5 (8.6) | 12 (3.2) | 3.22 (1.05–9.87) | 10.37 (1.49–72.01) |
| Head circumference (cm) | ||||
| <34 | 14 (27.5) | 73 (21.3) | 1.38 (0.69–2.77) | 1.27 (0.51–3.16) |
| 34–36 | 30 (58.8) | 222 (64.7) | 1 | 1 |
| >36 | 7 (13.7) | 48 (14.0) | 1.12 (0.46–2.74) | 0.77 (0.25–2.37) |
| Congenital malformation | ||||
| No | 40 (88.9) | 208 (85.2) | 1 | 1 |
| Yes | 5 (11.1) | 36 (14.8) | 0.67 (0.24–1.92) | 0.66 (0.17–2.51) |
- a ORs were adjusted for age and sex.
- b Adjusted for age, sex, frequency of tonic-clonic seizures, living conditions, history of nocturnal tonic-clonic seizures, and intellectual disability.
- c All seven cases were born in weeks 32–36, whereas 33 controls were born in weeks 32–36 and 22 during weeks 23–31.
- d Smaller in size than normal for their gestational age, defined as a weight below the 10th percentile for the gestational age.
After additional adjustment for the frequency of tonic-clonic seizures last year, history of nocturnal tonic-clonic seizures, living conditions, and intellectual disability, we observed that Apgar score between 0 and 6 after 10 min inferred a 10-fold increased risk for SUDEP (OR 10.37; 95% CI 1.49–72.01), and being born after 40 weeks of gestation a more than 2-fold risk (OR 2.42; 95% CI 1.03–5.65) (Table 2). Being small for gestational age remained with a 3-fold risk, albeit now not statistically significant (OR 3.00; 95% CI .63–14.15).
4 DISCUSSION
In this population-based case-control study we observed a 10-fold increased risk for SUDEP for individuals with an Apgar score of 0–6 at 10 min after birth after controlling for known SUDEP risk factors.
Apgar score is a common method for evaluating the health of the newborn. Low Apgar score has been associated with increased infant death15 and long-term neurological morbidity in adults.16 A Swedish population-based cohort study found increasing risks of epilepsy, and especially cerebral palsy, with decreasing Apgar scores at 5 and 10 min.17 The causes of cerebral palsy and epilepsy are multifactorial, but prenatal and perinatal events are important risk factors for both conditions. In a similar fashion, several childhood and adult diseases are related to size at birth.18
Prenatal and perinatal events such as intrauterine growth retardation and hypoxia at delivery, which are reflected in a low birth weight and low Apgar score, could make the individual more susceptible to both developing epilepsy and failing arousal mechanisms and therefore succumbing later in SUDEP. Delivery is a stress test for the neonate that is theoretically analogous to a seizure. A low Apgar score could, therefore, be a predictor for postictal vulnerability.
To assess whether the association between low Apgar scores and intrauterine growth restrictions and subsequent SUDEP risk could be mediated through previously established and proposed SUDEP risk factors such as tonic-clonic seizures, nocturnal tonic-clonic seizures, living conditions, and intellectual disability, we conducted analyses adjusting for these covariates (model 2). It is noble that such an adjustment did not attenuate the associations, indicating that low Apgar score and SGA could be independent risk factors, mediated through mechanisms other than poor seizure control, intellectual disability, and living alone. Such mechanisms could include impact on the development and function of brainstem structures involved in autonomic cardiorespiratory control and arousal mechanisms. Dysfunction of the brainstem is thought to play a key role in both SUDEP and SIDS.2, 6, 8
Prenatal exposure to smoking is an established risk factor for SIDS.7, 8 We observed an 80% risk increase for SUDEP, albeit not statistically significant (OR 1.79; 95% CI .93–3.43) for mothers smoking during pregnancy. Although prematurity is a risk factor for SIDS, we observed a 2-fold SUDEP risk in individuals born after week 40.
The strengths of this study include its nationwide design, with cases and controls coming from the same population, and that both SUDEP cases and epilepsy controls were scrutinized and validated. Furthermore, information on the potential risk factors came from extensive medical record reviewing in combination with nationwide registers of high quality, such as the MBR. This gives us a unique and novel way to investigate if pre- and perinatal factors are associated with SUDEP risk and we could also adjust for previously identified SUDEP risk factors to exclude that these serve as mediators on a hypothesized casual chain. Given these strengths, we believe that our results are generalizable to other settings with similar socioeconomic and health care conditions. A major limitation is the sample size, especially regarding SUDEP cases (n = 58), which reduces power and may explain why we could not observe an association with some prenatal factors such as mother’s smoking.
5 CONCLUSION
In this first attempt to address the association between SUDEP and early life factors, we found some perinatal factors to be associated with increased SUDEP risk, suggesting that some components of SUDEP susceptibility may be established very early in life. In the field of SUDEP, there is an active search for biomarkers, not the least those that explain or shed a light on postictal vulnerability and cardiorespiratory dysfunction. In our view, low Apgar score and SGA are potential risk factors or biomarkers that need to be considered with other more established risk factors; and the potential mechanisms linking low Apgar score and SGA to SUDEP risk remain to be explored. These results call for further investigations on larger samples. Furthermore, we should strive to track data on perinatal factors such as Apgar to learn more about these possible SUDEP risk factors in our patients with epilepsy.
ACKNOWLEDGMENTS
The study was supported by funding from Stockholm County Council and Hjarnfonden. The sponsors had no influence on the conduct of the study, analysis, interpretation, writing of the article, or the decision to publish the results
CONFLICT OF INTEREST
Olafur Sveinsson has received grants from Hjarnfonden for this work. Tomas Andersson reports no disclosures. Sofia Carlsson reports no disclosures. Torbjörn Tomson reports speaker’s and advisory board honoraria to his institution from Eisai, Sanofi, Sun Pharmaceutical Industries Ltd, UCB, Angelini, and GW Pharma, and research support from Bial, Eisai, GlaxoSmithKline, Stockholm County Council, Teva, GW Pharma, Angelini, and UCB. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.
