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: Institute of Psychiatry, Psychology & Neuroscience, Division of Neuroscience, King’ College London, Maurice Wohl Clinical Neuroscience Institute, Ground Floor (G.39), 5 Cutcombe Road, Camberwell, London, SE5 9RX, UK.
The pooled incidence of pre-ictal heart rate increase (HRI) was 36/100 seizures.
•
Higher HRI estimates were found in TLE, adults and people on AEDs.
•
Pre-ictal HR reduction incidence was 0.5% in the paediatric population.
•
Bias and methods assessment disclosed limitations in the evidence-base.
•
Monitoring of HR changes could be attempted to early identify seizures.
Abstract
Purpose
To estimate the incidence of pre-ictal heart rate (HR) manifestations and to identify clinical and study-related factors modulating the estimate.
Methods
We searched articles recording concurrent pre-ictal EEG and HR in adults and children with epilepsy. Pre-ictal HR changes were classified as HR reduction (HRR) or increase (HRI). Studies reporting the total number of seizures and the number of seizures with pre-ictal HR changes were included in a random-effects meta-analysis. A random-effects meta-regression was used to identify variables affecting study heterogeneity.
Results
Thirty studies, including 1110 participants and 2957 seizures, were included. The meta-analysis showed a pooled incidence of pre-ictal HRI of 36/100 seizures (95% CI 22–50). The pre-ictal HRI incidence was 44/100 seizures (95% CI 33–55) in studies including temporal lobe epilepsy, 55/100 seizures (95% CI 41–68) in studies enrolling adults and 35/100 seizures (95% CI 16–58) when patients on antiepileptic drugs were included. The meta-regression showed that the age group, the length of the pre-ictal period, the incidence of ictal tachycardia and the time of onset of the pre-ictal HRI had a significant impact on estimates variability. The pooled incidence of pre-ictal HRR was 0/100 seizures (95% CI 0–1).
Conclusion
Review of bias evaluation and methods assessment disclosed several major limitations in the evidence-base. HR monitoring could be valuable to identify seizures prior to their apparent onset, opening the possibility to early interventions. Additional effort is necessary to delineate the target population who might benefit from its use and the mechanisms sustaining the pre-ictal cardiac changes.
]. Ictal cardiac changes have been attributed either to the activation of cortical structures connected to the autonomic centres, or to peripheral mechanisms regulating reflex responses, driving various cardiac manifestations [
]. Although a reduction of the heart rate (HR) can occur, the most commonly observed pattern associated with seizures is represented by an increased HR [
]. The possibility to detect alterations of cardiac parameters during a seizure has led to consider the HR as a potential extra-cerebral seizure biomarker [
]. In comparison to other more invasive measurements, HR assessment has the potential of providing a reliable and easily measurable signal which may be recorded continuously in a daily life environment. In addition, HR alterations have been reported to occur early during a seizure or even pre-ictally [
], demonstrating a theoretical role in seizure prediction. However, studies investigating ictal and pre-ictal HR changes have yielded controversial results and our understanding of the mechanism supporting these alterations is still limited [
]. In fact, the frequency of ictal HR manifestations in patients with epilepsy varies considerably across studies and estimates range between 38 and 100% for IT and <5 and 66.7% for IB [
], with higher variability when a pre-ictal onset is considered. This variance likely reflects differences in measurements along with clinical variables, including epilepsy and seizure type, age groups, use of antiepileptic drugs (AEDs), lobe of seizure onset and presence of cardiac co-morbidities. In addition, methodological issues can also be responsible for the mixed findings, including methods used to assess seizure onset and to define HR changes. Understanding the distribution of pre-ictal HR alterations and the temporal relation between these changes and seizure activity might represent a pivotal point to shed new light on the mechanisms shaping the ictal autonomic changes and influencing seizure spreading patterns. Moreover, this could also help to explore the utility of this measure as a seizure detection tool and to identify the target population who might benefit from its use. We aimed to perform a meta-analysis to assess the incidence of pre-ictal HR alterations and to identify clinical and study-related factors modulating this phenomenon.
2. Methods
This systematic review and meta-analysis was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) and Cochrane Handbook guidelines [
2.1 Criteria for considering studies for this review
Retrospective and prospective studies (case-control, cohort or case series) recording concurrent pre-ictal EEG and HR in adults and children with epilepsy were considered for inclusion. Studies focusing on neonatal seizures only were excluded. Pre-ictal HR changes were classified as HR reduction (HRR) or increase (HRI) as compared to the baseline HR. Methods used to measure the HR and definitions of HR alterations were extracted and considered as separate variables assessed in the meta-regression as potential sources of between studies heterogeneity.
2.2 Search methods for identification of studies
A systematic search with no language restrictions was carried-out to identify all relevant published and unpublished studies. The search strategy included the terms “epilepsy”, “heart rate”, “ictal tachycardia” and “ictal bradycardia” and is reported in Appendix S1. The search was conducted from the first date available (1958) up to May 30, 2017 in MEDLINE, Embase and the Cochrane Central Register of Controlled Trials (CENTRAL). The search strategies for each database were based on the strategy developed for MEDLINE, taking into account the differences in controlled vocabulary and syntax rules. In addition to the electronic searches, we hand-searched reference lists of all available review articles and primary studies and hand-searched the references quoted in the most recent congress proceedings (e.g. International Epilepsy Congress, European Congress on Epileptology).
2.3 Data collection and analysis
Two review authors (EB and AB) independently assessed the titles and abstracts of all the studies identified by the electronic searching or hand-searching. Full texts of potentially relevant studies were obtained and screened. We resolved any disagreements concerning study inclusion and exclusion by discussion. For each study included, two review authors (EB and AB) independently extracted the following data on an ad-hoc created data collection form: study design (prospective, retrospective) and setting (inpatients, outpatients); demographic and clinical data of the population (number of patients, age group, gender, type of epilepsy, drug-resistance, cardiac comorbidities, AEDs administration/withdrawal during the recording); number of seizures recorded, seizure type and focus (including focus side), definition of seizure onset (EEG, clinical, either EEG or clinical); EEG characteristics: type (scalp, intracranial), number of electrodes used (standard, non-standard), duration (continuous, intermittent, number of hours recorded), use of video recording; HR assessment methods (automatic versus manual count of R–R intervals or QRS complexes or beats) and length of ECG epoch used for HR analysis (in seconds); duration of the pre-ictal period assessed (in seconds); number of seizures presenting with ictal HR change (IT, IB and definitions adopted) and ictal HR peak/minimum reached; number of seizures with pre-ictal HR change (pre-ictal HRI and HRR) and onset time (seconds before seizure onset).
2.4 Quality assessment
The quality of included studies was evaluated using a standard assessment tool, that was slightly re-adapted (Appendix S2), and included sample representativeness, condition assessment, and statistical methods [
]. Each study was given a quality score of 0 to 8 based on fulfilment of the quality criteria. The quality score was considered as a separate variable in the meta-regression.
For studies included in the meta-analysis, two review authors (EB and AB) independently assessed the study methods and the risk of bias related to them. The following domains were considered and compared across studies: definition of HR change, definition of pre-ictal period, time interval used to calculate the HR change, assessment of onset of HR change.
2.5 Data synthesis and analysis
Studies reporting the total number of seizures recorded and the number of seizures with pre-ictal HR changes (including 0), were included in a meta-analysis.
Pre-ictal HRI and pre-ictal HRR were separately analysed. In addition, meta-analysis was separately fitted according to epilepsy type, age group and AEDs administration/withdrawal, when these variables were specified by an adequate number of studies. We used the ‘metaprop_one’ command in Stata 14.0 to estimate crude incidence rates along with their 95% confidence intervals (CI) and we expressed the estimates as the number of seizures with pre-ictal HR changes per 100 seizures. We reported the pooled, weighted estimate generated by random-effects models. To handle the studies with zero events, we used Freeman–Tukey double arcsine transformation which stabilizes the variance of the proportion restricting the 95% CI within the range of 0 and 1, even in the presence of zero events [
]. As a sensitivity analysis, the pooling process was repeated after the successive removal of incidence studies judged at high risk of bias in all the domains considered. The I2 was used to quantify the magnitude of between-study heterogeneity and the Cochrane Q statistic was calculated to determine significance. Publication bias was investigated statistically using Begg’s and Egger’s tests. To determine the influence of the clinical variables and of the study-level factors on the observed variability, we used random-effects meta-regression. We regressed one variable at a time. Significance level was established at p < 0.05. All analyses were performed using STATA version 14.0 (StataCorp, College Station, TX, U.S.A.).
3. Results
3.1 Study selection and quality assessment
The search of electronic databases yielded 1130 references (Fig. 1). One additional study was identified by hand-searching. After duplicates and non-relevant studies were removed, the titles and abstracts of the remaining studies were reviewed and the full-text of 98 articles with potentially relevant studies was assessed. Finally, 30 published studies were considered eligible for qualitative synthesis. All the included studies assessed the occurrence of pre-ictal HRI, while pre-ictal HRR was investigated in 20 studies only. The characteristics of all the 30 studies included are summarized in Table 1 and more extensively detailed in Table S1 (Supporting information). Quality assessment revealed a median study quality score of 6/8 (range 1–8).
Fig. 1Flowchart of the study selection process. HRI: heart rate increase; HRR: heart rate reduction.
Thirty studies, enrolling 1110 people living with epilepsy (39.0% female; 38.2% male; 22.8% not reported) and a total of 2957 seizures, assessed the occurrence of pre-ictal HRI. Eighteen included adults [
], and 14 did not report the AEDs status. Twenty-one studies included complex partial seizures (CPS) alone or in combination with other seizure types [
]. The majority were based on prolonged EEG recording, using a standard number of electrodes and additional video recording. Only studies published before 1997 used less electrodes than standard [
]. The mean time-epoch considered for HR assessment was on average 23.1 s (±23.3; range 5–60) and a mean pre-ictal period of 99.2 s prior to seizure onset (±156.5; range 15–600) was assessed. One study considered multiple pre-ictal periods, starting from 240 min before seizures onset [
The mean baseline HR was 74.35 bpm (±4.8; range 65.9–86.4). Although all the studies described the occurrence of IT, only 18 studies clearly reported its frequency, giving a cumulative incidence of 1556/2957 seizures (52.6%) [
]. The mean ictal HR peak was 116.5 bpm (±27.0; range 81.2–201.0), demonstrating a mean increase of 56.8% as compared to the baseline HR. Pre-ictal HRI was found in 623/2957 seizures (21.1%) and it was reported in 14 studies [
], seizures originating from the mesial temporal lobe were more often associated with pre-ictal HRI as compared to lateral and extra-temporal onset, with two studies reporting more significant pre-ictal HRI in right-sided mesial seizures [
]. Patients with secondarily generalized tonic-clonic (SGTC) seizures were associated with increased pre-ictal HR in one study compared to those with localized seizures only [
], no data were available on the influence of gender on HRI onset time. The peri-ictal HR pattern was considered consistent or predictable in the same patient across multiple seizures in 2 studies [
]. For studies included in the meta-analysis, the quality assessment yielded higher scores (7/8). The assessment of study methods and risk of bias is detailed in Table 2. A high variety in the four methodological domains considered was demonstrated across studies, and the risk of bias was frequently rated as unclear or high. The pooled incidence of pre-ictal HRI was 36/100 seizures (95% CI 22–50) (Fig. 2A), with a significant between study heterogeneity (I2 = 97.25%, Q p-value = 0.000). Egger’s and Begg’s test were not significant (p = 0.74, p = 0.92 respectively) indicating a low risk of publication bias. The analysis performed according to epilepsy type (Fig. 3A) showed that studies including TLE had a pre-ictal HRI incidence of 44/100 seizures (95% CI 33–55) while the incidence was 36/100 seizures (95% CI 6–75) for studies including mixed focal types and 16/100 seizures (95% CI 4–34) for those including generalized and focal epilepsy combined. Studies including adults presented with an incidence rate of 55/100 seizures (95% CI 41–68) while the incidence was 25/100 seizures (95% CI 9–44) for studies including children and 9/100 seizures (95% CI 0–33) for studies including both children and adults (Fig. 3B). Studies including patients recorded while on AED treatment had a pooled incidence rate of 35/100 seizures (95% CI 16–58) as compared to 22/100 seizures (95% CI 4–47) for those studies including patients withdrawing the treatment (Fig. 3C). The treatment status was not reported in 5 studies [
Table 2Methods and risk of bias of studies included in the meta-analysis (N = 15) (For interpretation of the references to colour in the table legend, the reader is referred to the web version of this article.).
Risk of bias assessment: red = high risk of bias, yellow = unclear risk of bias, green = low risk of bias. Risk of bias assessment criteria: Definition of HR change: Baseline not defined or overlapping with pre-ictal periods (high risk of bias), Unclear HRI definition (unclear risk of bias), Well defined HR change (low risk of bias). Definition of pre-ictal period: ≤30 seconds or not defined (high risk of bias), >30 and <60 seconds (unclear risk of bias), ≥60 seconds (low risk of bias). Time interval to calculate HR change: Fixed time interval (average of RR intervals) ≥ 10 sec or not defined (high risk of bias), Fixed time interval (average of RR intervals) <10 sec (unclear risk of bias), Dynamic (instant RR intervals plot) (low risk of bias). HR change onset time: Not reported (high risk of bias), Time interval of HR change as per definition (unclear risk of bias), Exact onset (first instant HR change) reported (low risk of bias). Bpm: beat per minute; CIs: confidence intervals; HR: heart rate; HRI: heart rate increase; HRR: heart rate reduction.
The meta-regression (Table S2, Supporting information) showed that, among clinical variables, the age group of study participants had a significant impact on variability of studies (p = 0.004), accounting for 23.44% of the observed heterogeneity. Among methodological covariates, the length of the pre-ictal period during which the HR was assessed (p = 0.016), explained an additional 33.44% of the observed heterogeneity, with higher incidence in studies using a longer time frame (60 s or more). In addition, the pre-ictal HRI estimates were significantly correlated with the IT frequency (p = 0.004) and with the time of onset of the pre-ictal changes (p = 0.04), and higher pre-ictal estimates were reported in studies observing higher IT estimates or earlier HRI presentations (Fig. 4). Finally, the sensitivity analysis demonstrated comparable results when two studies considered at high risk of bias in all the domains were excluded (pre-ictal HRI incidence 36/100 seizures, 95% CI 23–49; I2 = 96.58%, Q p-value = 0.000) and the meta-regression found no effect of study quality on the estimates of the pre-ictal HR (p = 0.2).
Fig. 4Meta-regression bubble plot showing the variation of the incidence of pre-ictal heart rate increase (HRI) regressed against the (A) proportion of seizures with ictal tachycardia (IT) and the (B) time of onset of pre-ictal HRI reported in the studies included.
]. The mean baseline HR was 76.2 bpm (±4.9; range 69.5–86.4) and the mean ictal HR minimum was 62.2 bpm (±13.4; range 50.0–76.5), demonstrating a mean decrease of 18.4% as compared to the baseline HR. Pre-ictal HRR was observed in 13/2346 seizures (0.5%) and was reported in four studies, exclusively including a paediatric population [
], one reporting a mild pre-ictal bradycardia progressing into IT in seizure originating from the lateral temporal areas or from the right hemisphere [
30734-3&id=gr1.jpg){kind=link}
30734-3&id=gr2.jpg){kind=link}
30734-3&id=gr3.jpg){kind=link}
30734-3&id=gr4.jpg){kind=link}