If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Corresponding author at: Child Neurology Unit, Presidio Ospedaliero Provinciale Santa Maria Nuova, AUSL-IRCCS di Reggio Emilia, Viale Risorgimento 80, 42123, Reggio Emilia, Italy.
Neonatal seizures in preterm infants are associated with unfavourable outcomes.
•
We review rates of death, epilepsy, cerebral palsy, developmental delay and normal outcome from studies in the 2000’s.
•
Specific independent risk factors have been seldom investigated.
Abstract
Introduction
Neonatal seizures (NS) are associated with increased mortality and risk of cerebral palsy, epilepsy and intellectual disability. We performed a systematic review with the primary objective to delineate the rate of these outcomes following NS in preterm infants from studies published in the 2000’s and the secondary objective to identify risk factors.
Methods
Inclusion criteria: original articles published between 1/1/2000 and 12/31/2018, written in English, evaluating newborns ≤37 weeks of gestational age and suffering from NS, in which at least one of these was evaluated: epilepsy, cerebral palsy, intellectual disability/developmental delay, normal outcome, death.
Results
Twenty-two papers were selected and all were observational, with a retrospective design in 15. Three were population-based and twenty-one have a comparison. It has been found a 22–80 % of mortality, 11.3–38.9 % of epilepsy, 12–84.6 % of cerebral palsy, and 20–42.7 % of intellectual disability/developmental delay rate. An increased risk for all outcomes considered was reported.
Risk factors for specific outcomes were provided by a minority of studies. However, inclusion criteria, definition of NS and measured outcomes, follow-up lengths differed considerably between studies.
Discussion
Results of the selected studies are only partially comparable or generalizable because of differences in study design. They have a risk for potential biases, although they provide (if analyzed) readily available prognostic factors, easy to apply in clinical practice. Prospective, population-based studies with EEG-defined NS are warranted in order to produce evidence-based guidance for management of preterm newborns with seizures.
However, the rate of normal outcome has decreased, mainly linked to a rise in the percentage of preterm babies with (minor) motor disabilities. The rate of epilepsy, as recently reviewed [
] has substantially remained unchanged in the last 60 years.
The risk of unfavorable outcome is higher in those preterm babies suffering from neonatal seizures (NS). In cohorts of preterm newborns, seizures are among the best risk indicators for long-term neurologic morbidity [
]. Seizures on or after day 2 of life predict neurologic outcome at 2 years with an adjusted odds ratio (OR) of 5.8 (1.9–17.8) and at 5 years of age with an adjusted OR of 3.0 (1.1–8.6)4. In a further study, newborns with NS have an OR of 18 for adverse outcome (95 % CI 3–100), with a 14-times (95 % CI 2–86) OR increase per decreasing week of gestational age [
A number of clinically-relevant differences exists between full-term and preterm infants with seizures. First of all, seizures in the preterm newborn are more frequent [
], representing a diagnostic challenge, with a risk of underdiagnosis. Furthermore, various clinical studies document a later age at seizures onset in the preterm newborn, throughout the whole first week of life and in some cases even beyond, especially in infants born before the 29th wGA [
]. This probably accounts for the lower seizure rate recently documented with continuous EEG in the first 72 h of life (5 % of preterm newborns < 32 wGA) [
], underscoring a possible indication for longer monitoring. This difference in the period of time at increased risk probably reflects the immaturity of the preterm brain to initiate and sustain electrographic discharges [
], which also depend on gestational age: intraventricular hemorrhage and its complications are the leading cause in very and extremely preterm newborns [
The main risk factors for seizures in the preterm newborn include: gestational age (with a 9 % increase in seizures for each week decrease in gestational age among low birth weight infants) [
We aimed to evaluate the mortality risk and neurologic prognosis (epilepsy, cerebral palsy, intellectual disability/developmental delay, normal outcome) in infants born preterm and suffering from NS. As a secondary aim, we sought to identify risk factors for the above mentioned unfavourable outcomes.
We used the following inclusion criteria: original articles, written in English, published between 01/01/2000 and 07/31/2018, addressing at least one of the following outcomes: death, epilepsy, cerebral palsy, intellectual disability, normal neurologic outcome. We included both data on death at discharge and death at the last follow-up.
Exclusion criteria: reviews, case reports, articles published on animals, written in languages other than English and not reporting on the rate of occurrence of the various outcomes were excluded.
We performed the last PubMed search on the 9th March 2019 using the following terms: (outcome) AND (preterm) AND (neonatal) AND (seizures) in order to identify papers with the primary objective to investigate the outcome of NS in preterm infants (defined as neonates born with a gestational age ≤37 weeks). Additional relevant articles were retrieved by manually searching through the references list of selected articles. The search string is displayed in Appendix 1.
We reviewed search results by title and abstract and excluded all non-pertinent results. For the remaining studies, we used a form featuring information on the study design (prospective versus retrospective), the study population, the method of seizure ascertainment (clinical versus EEG), the presence of a comparison group, the follow-up length and the prognostic factors for death or specific neurologic outcomes as listed above, and study strengths or limitations/risk for biases.
Study strengths targeted: prospective studies, EEG diagnosis of NS, continuous EEG seizure monitoring, control/comparison groups, rigorous documentation and reporting of results. Limitations included: small sample sizes, retrospective studies, uncontrolled studies, clinical definition of NS, limited EEG monitoring, lack of control/comparison groups, and missing data. We checked for risk of selection bias, confounding bias and design bias.
Statistically significant risk factors for each of the unfavorable outcomes, when available, were also collected.
2.1 Statistical analysis
We performed a descriptive analysis for the main outcomes (i.e. normal outcome, mortality, epilepsy, cerebral palsy and developmental delay). Data are summarized as median of percentages and interquartile range (IQR). The odds ratio (OR) with 95 % confidence interval (95 % CC) was calculated to assess the effect of NS (preterm with NS vs preterm without NS) and gestational age (preterm with NS vs full-term with NS) on the main outcomes in controlled studies. Statistical heterogeneity was investigated with the Chi2 test, with a significance level of P < .05, and with an I2 test, significant for I2 >70 %. A random effect model was used when heterogeneity was significantly evident; otherwise, a fixed effect model was adopted. RevMan version 5.3.12 was used for data analysis.
3. Results
Our initial search yielded 156 outputs. Of these, 142 were excluded based upon review of title and abstract or by full review of the full-text. The remaining 14 articles were evaluated in detail. By analyzing references in relevant articles, we found eight additional papers fulfilling inclusion criteria (Appendix 2).
Of the 22 articles, seven are prospective (7/22, 31.8 %) and 15 (68.1 %) retrospective, three are population-based (13.6 %), three provide data from a registry or a database (13.6 %), while the remaining 16 are hospital-based (72.7 %).
The outcomes of mortality, normal outcome, epilepsy, cerebral palsy, and intellectual disability/developmental delay are reported in separate tables as percentages and as median and interquartile ranges (IQR) of percentages in the following sections.
3.1 Normal outcome
Six articles report on the rates of normal neurologic outcome in preterm survivors of NS. Two are prospective [
The rate of preterm newborns with seizures surviving without disabilities is highly variable (0–22.3 %; median: 15.8 %; IQR: 53.2 %), which is lower than in preterm infants without seizures (26.4–80 %; median: 53.2 %; IQR: 26.8 %; OR: 0.26; 95 % CI: 0.19-0.37; heterogeneity assessment: I2: 62 %, p: 0.10, Chi2: 2.67) (Fig. 1a, b). Furthermore, in a cohort of preterm infants <29wGA, none of the patients experiencing seizures had a favourable outcome [
]. The follow-up length ranges from short-term (discharge or death) and 10 years (Table 1).
Fig. 1a. Forest plot of the odds ratio for normal outcome in preterm babies with NS versus preterm babies without NS. b. Forest plot of the odds ratio for normal outcome in preterm versus full-term babies with NS.
], mortality in preterm newborns with NS (22–46.2 %; median: 39.2 %; IQR:10.7 %) is higher than in preterm newborns without (5–15.4 %; median: 10.2 %; IQR: 5.2 %) [
] have a comparison group of preterm newborns without seizures: two of them consider all gestational ages below 38 wGA and find 33 and 41.6 % of mortality in patients with NS versus 5 and 20.5 % in those without [
]. The only retrospective, population-based study documented a 34.9 % mortality in preterm infants compared to 12.2 % in full-term infants (both groups with NS) [
Overall, the mortality in preterm with EEG definition of NS (33.3 %–45.4 %; median: 36.0 %; IQR: 8.7 %) was higher than in preterm without NS (5-2.5 %; median: 15.3 %; IQR: 11.7 %).
After data aggregation from the two groups of studies (clinical and EEG definition of NS), we found a significantly increased risk of mortality in preterm with NS compared to preterm without (OR: 4.07; 95 % CI: 3.50–4.73; heterogeneity assessment: I2: 64 %, p: 0.006, Chi2:19.66) (Fig. 2a, b).
Fig. 2a. Forest plot of the odds ratio for mortality in preterm babies with NS versus preterm babies without NS. b. Forest plot of the odds ratio for mortality in preterm versus full-term babies with NS.
], while two studies (14.3 %), from the same research group, provided these risk factors for death: a BW < 1000 g, need for cardiopulmonary resuscitation, and a moderately/severely abnormal cerebral ultrasound scan [
]. One additional study from the same research group, analyzed mortality risk factors in a mixed cohort of preterm and full-term newborns and found the 5-minutes Apgar score and etiology to be predictive in the multivariate analysis [
One final study (only performing a correlation analysis and not distinguishing for gestational ages) identified a correlation between death and positive pressure resuscitation, mechanical ventilation, perinatal asphyxia, infection, gestational age (PT versus FT) and low birth weight (<2500 g) [
Increased number of febrile seizures in children born very preterm: relation of neonatal, febrile and epileptic seizures and neurological dysfunction to seizure outcome at 16 years of age.
Increased number of febrile seizures in children born very preterm: relation of neonatal, febrile and epileptic seizures and neurological dysfunction to seizure outcome at 16 years of age.
Increased number of febrile seizures in children born very preterm: relation of neonatal, febrile and epileptic seizures and neurological dysfunction to seizure outcome at 16 years of age.
], while in the remaining two studies, the rate of epilepsy is lower in ex-preterm than in ex-full-term newborns with NS (27.3–46 %; median: 38.4 %; IQR: 8.6 % versus 15.2–37.9 %; median: 32.2 %. IQR: 9.9 %) [
]. Both studies comparing preterm infants with and without NS document a higher rate of epilepsy in the former (11.3–25 %; median 18.2 %; IQR: 6.9 % versus 3.1–4.6 %; median: 3.9 %; IQR: 0.9 %) [
Increased number of febrile seizures in children born very preterm: relation of neonatal, febrile and epileptic seizures and neurological dysfunction to seizure outcome at 16 years of age.
Data aggregation revealed a significantly increased risk of epilepsy in ex-preterm with NS as compared to ex-preterm without NS (OR: 4.47; 95 % CI: 3.28–6.08; heterogeneity assessment: I2: 33 %, p: 0.23, Chi2: 2.98). On the other hand, no difference was found between ex-preterm and ex-full-term with NS (OR: 1.25; 95 % CI: 0.75–2.06; heterogeneity assessment: I2: 46 %, p: 0.07, Chi2: 12.93) (Fig. 3a, b).
Fig. 3a. Forest plot of the odds ratio for epilepsy in preterm babies with NS versus preterm babies without NS. b Forest plot of the odds ratio for epilepsy in preterm versus full-term babies with NS.
Increased number of febrile seizures in children born very preterm: relation of neonatal, febrile and epileptic seizures and neurological dysfunction to seizure outcome at 16 years of age.
] (1/12, 8.3 %) reported SE in preterm newborns ≤29wGA as a risk factor for epilepsy. Another study (not included in our analysis) from the same population found severely abnormal cerebral ultrasound scan finding [
] as a further risk factor. An additional study, not reporting the raw incidence of epilepsy in preterm newborns separately (and therefore not included in Table 3) reports female gender and presence of neonatal status epilepticus as independent predictors of epilepsy in the mixed cohort of preterm and full-term newborns [
]. CP was defined as a qualitative motor disorder due to non-progressive interference with brain development, occurring before central nervous system growth is complete in two papers [
]. One article distinguished between moderate CP (defined as no ambulation or ambulation only with assistive devices, with ability to sit, independently or with support), and severe CP (inability to ambulate or sit with support) [
]. They all document a higher rate of CP in preterm infants, irrespective of an EEG or clinical definition of NS (33–84.6 %; median: 41.2 %; IQR: 25.8 %, in the first group, versus 25.3–43 %; median: 32.7 %; IQR: 8.85 %, in the second group). Similarly, the two studies comparing preterm with and without NS [
] showed a higher incidence in the first group (12.0–22.7 %; median: 17.4 %; IQR: 5.3 versus 3.8–3.9 %; median: 3.85 %; IQR: 0.05 %). Overall, there is a significantly higher risk of cerebral palsy in preterm with history of NS in comparison with preterm without (OR: 5.15; 95 % CI: 2.47–10.74; heterogeneity assessment: I2: 82 %, p: 0.02, Chi2: 5.71) (Fig. 4a, b).
Fig. 4a. Forest plot of the odds ratio for cerebral palsy in preterm babies with NS versus preterm babies without NS. b. Forest plot of the odds ratio for cerebral palsy in preterm versus full-term babies with NS.
], which, nevertheless, did not distinguish between preterm and full-term infants but analyzed the whole cohort, finding an independent role of perinatal asphyxia, severely abnormal background EEG, and need for resuscitation at birth (Table 4).
3.5 Intellectual disability/developmental delay
Seven studies evaluate intellectual disability or developmental delay in infants surviving NS [
] for cases ascertained up to 2007, while the Bayley Scales of Infant and Toddler Development 2nd edition (BSIDII) was used after 2007, considering mental developmental index (MDI) and psychomotor development index (PDI) scores < 85 [
] used the term neurodevelopmental impairment (NDI), defined as a composite outcome including CP evaluated by Gross Motor Function Classification System (GMFCS), Bayley III (< 85 in NDI and < 70 in significant NDI), sensorineural/mixed hearing loss, or visual impairment (need for hearing aids or cochlear implant in significant NDI), or severe developmental delay precluding Bayley-III assessment (significant NDI).
Three studies rely on an exclusively clinical definition of NS [
] have preterm infants without seizures as a comparison group and find lower MDI scores in ex-preterm infants with seizures, (20–42.7 %; median: 38.5 %; IQR: 11.4 % versus 14–22.6 %; median: 14 %; IQR: 4.5 %), although in the study of West et al. [
], severe developmental delay is considered. The remaining studies compare preterm with full-term newborns with NS, finding comparable rates (23–42.4 %; median: 29.2 %; IQR: 14.8 % versus 20.9–54.4 %; median: 32.7 %; IQR: 10.9). Finally, one study do not have a comparison group (35.5 % of DD) [
]. Overall, the risk of developmental delay was higher in preterm with NS than in preterm without NS (OR: 2.73; 95 % CI: 2.26–3.30; heterogeneity assessment: I2: 30 %, p: 0.24, Chi2: 2.87), while no significant difference resulted from the comparison of preterm and full-term with NS (OR: 0.89; 95 % CI: 0.54–1.48; heterogeneity assessment: I2: 0 %, p: 0.53, Chi2: 1.27) (Fig. 5a, b). five studies (5/7, 71.4 %) did not assess the independent risk factors for intellectual disability/developmental delay separately from other unfavourable neurological outcomes [
] did not analyze risk factors at all, while one (14.3 %) study found etiology (metabolic disorder, intracranial bleeding, perinatal asphyxia), seizure type (myoclonic seizures or >1 seizure type), and duration of treatment >24 months to be associated with developmental delay in a mixed cohort of full-term and preterm newborns [
Fig. 5a. Forest plot of the odds ratio for intellectual disability/developmental delay in preterm babies with NS versus preterm babies without NS. b. Forest plot of the odds ratio for intellectual disability/developmental delay in preterm versus full-term babies with NS.
3.6 Independent predictors of composite unfavourable outcome in preterm infants with NS
Some of the addressed studies did not analyze outcomes separately, but rather addressed predictors of unfavourable outcome. Apgar score at 1 min, and severely abnormal background EEG activity were found as independent predictors of abnormal outcome in a cohort of preterm newborns with NS [
] and abnormal neurologic examination and abnormal polysomnographic recording by a different research group as independent predictors of unfavourable outcome [
All of the reviewed studies have implemented our understanding on the disease evolution. Evaluated studies have differing study designs: none of the studies has a randomized controlled design, and the majority have a retrospective design. There are three population-based studies. Depending on the evaluated outcome, the percentage of studies including a comparison group varies between 83.3 % and 100 %. Another noted source of variability is that definition criteria of NS vary between clinical-only, EEG-defined and mixed. This heterogeneity represents a huge source of risk for a selection bias, as a wide body of evidence documents the unreliability of clinically-based differential diagnosis between seizures and other paroxysmal events in newborns [
]. According to the American Clinical Neurophysiology guidelines, continuous EEG monitoring is considered the best recommended method of ascertainment and quantification of NS [
Conclusions drawn by each study are not always comparable because of the variable follow-up length.
Although studies include a categorization of NS according to etiologies, the growing knowledge on neonatal-onset epilepsies of genetic origin accumulating in the last few years might have resulted in the inclusion of non-ascertained genetic diagnoses. This would result in the inclusion of a subpopulations of patients with specific, and different, clinical characteristics and evolution compared to acute symptomatic NS.
Finally, as all of the reviewed studies are observational (and mainly retrospective), evaluation of risk factors for each of the considered outcomes is potentially hampered by confounding biases. Although multiple logistic regression models have been applied in the studies evaluating independent predictors of outcome, still the retrospective design theoretically might not allow for the complete exclusion of potential, additional, undetermined or non-evaluated confounders. However, considering the study populations consisting of preterm (often critically-ill) newborns it must be acknowledged that there might be significant intrinsic limitations to the construction of alternative study designs.
4.2 Main outcomes
Studies addressing subsequent normal neurological outcome in preterm newborns with NS show highly variable rates. However, when a comparison group is provided, they are concordant on a higher rate of normal development in preterm infants without NS compared to preterm infants suffering from NS. Of note, these conclusions are based on variable follow-up lengths.
Death is unanimously reported as higher in preterm infants with NS with a 3- to 4-folds increased risk in comparison with preterm without NS. However, percentages are highly variable between studies: 22–46.2 % with clinically-defined NS [
]. These figures seem to corroborate two considerations: the first being that NS are associated with higher mortality, and the second that the increased specificity provided by EEG confirmation of NS might enable a more correct selection of patients at increased risk of unfavourable outcome. Improvements in neurocritical care have resulted in progressively lower rates of death compared to older series [
Rates of epilepsy are variable and studies provide conflicting results regarding the role of gestational age in the development of epilepsy, while the comparison among preterm-born infants discloses a correlation between NS and subsequent epilepsy. From analyzed studies, cerebral palsy is more prevalent in preterm survivors of NS. Raw percentages seem to be similar irrespective of the modality of ascertainment of NS.
Data on intellectual disability/developmental delay can also be difficult to compare or generalize. In fact, different standardized scales have been used. Comparison with preterm infants without seizures seems to suggest a correlation between seizures and worse outcome for cognitive domains. The development of cognitive impairment (developmental delay or intellectual disability) in preterm newborns is higher if they experienced seizures in the neonatal period, but it seems to be affected by the ascertainment method of NS as well, being higher in case of clinically-defined seizures, apparently irrespective of follow-up length (Table 5). A common observation from different studies is that the majority of patients with developmental delay are severely affected and the presence of cognitive dysfunction is usually associated with other disabilities [
4.3 Prognostic factors in preterm newborns with seizures
In the majority of the reviewed studies, independent risk factors for abnormal outcomes in preterm infants with NS were not assessed specifically. Some of the studies did not look for predictors of outcome at all, while others did not distinguish between preterm and full-term babies.
Some studies merged together various unfavourable outcomes, documenting the detrimental role of Apgar score at 1 min, severely abnormal background EEG in preterm newborns with NS, and of birth weight, severely abnormal and cerebral ultrasound scan, status epilepticus [
Risk factors for death were retrieved from two different studies from the same research group: documenting a role of low birth weight (<1000 g), need for cardiopulmonary resuscitation, moderately/severely abnormal cerebral ultrasound scan [
No papers evaluated the risk for cerebral palsy following NS exclusively in preterm infants. In a mixed cohort, perinatal asphyxia, severely abnormal background EEG, and need for resuscitation at birth predicted CP [
]. The same study also provided risk factors for developmental delay in the same cohort: etiology, seizure type, and duration of treatment >24 months.
These results favor the view that the most prominent indicator of subsequent outcome is the severity of brain injury (neuroimaging data, especially ultrasound scans), and dysfunction (neurologic examination, background EEG) and of the overall clinical condition of preterm babies (perinatal variables: lower gestational ages and birth weights, lower Apgar scores).
A prognostic role of specific seizure types is on the contrary more controversial [
]. However, an association between multiple seizure types and outcome might be linked to more widespread brain dysfunction.
Preterm infants with SE have an unfavorable outcome. However, in a mixed cohort of preterm and full-term newborns with SE, no variables differentiated the two groups according to gestational age [
]. SE is more frequent in preterm newborns with a birth weight < 1000 g and between 1500 g and 2499 g, severely abnormal EEG and neurologic examination [
In the last 19 years, a total of 22 papers fulfilling our inclusion criteria addressed the question of neurologic outcome and mortality following NS in preterm newborns. None of these papers provides the best quality of evidence, as they all represent observational studies, mainly retrospectively conducted. However, from a clinical point of view, the take-home message is that preterm infants with seizures do worse than preterm infants not sustaining seizures and have higher mortality rates of both preterm newborns without seizures and full-term newborns with seizures.
Additionally, even with methodological limitations, the studies investigating independent risk factors for adverse outcome in preterm infants with seizures, identified cot-side variables which can be readily applied in clinical practice. Our review also suggests future clinical research directions, as long-term prospective, population-based studies collecting data on preterm newborns with EEG-defined NS, ideally with continuous or prolonged EEG monitoring, exclusive inclusion of acute symptomatic NS, precise definition of outcome measures and inclusion of a comparison group would be the best way to provide the robust evidence is needed to generate better monitoring and treatment practices.
Study funding
No funding was secured for this study.
Disclosure
The authors report no disclosures.
Declaration of Competing Interest
The Authors declare that they do not have any conflict of interest to disclose and that no funding was secured for the completion of this study.
Appendix A. Supplementary data
The following is Supplementary data to this article:
30481-9&id=gr1.jpg){kind=link}
30481-9&id=gr2.jpg){kind=link}
30481-9&id=gr3.jpg){kind=link}
30481-9&id=gr4.jpg){kind=link}
30481-9&id=gr5.jpg){kind=link}