SEARCH

SEARCH BY CITATION

Keywords:

  • Sudden unexpected death in epilepsy;
  • Hypercapnia;
  • Hypoxemia;
  • Seizure;
  • Localization-related epilepsy

Summary

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

Purpose:  The rate of sudden unexpected death in epilepsy (SUDEP) approaches 9 per 1,000 patient-years in patients with refractory epilepsy. Respiratory causes are implicated in SUDEP. We reported that ictal hypoxemia occurs in one-third of seizures in localization-related epilepsy. We now report on respiratory changes in the ictal/postictal period including changes in end-tidal CO2 (ETCO2) that correlate directly with alveolar CO2, allowing a precise evaluation of seizure-related respiratory disturbances.

Methods:  One hundred eighty-seven seizures were recorded in 33 patients with localization-related epilepsy, with or without secondarily generalized convulsions, undergoing video-electroencephalography (EEG) telemetry with recording of respiratory data.

Results:  The ictal/postictal ETCO2 increase from baseline was 14 ± 11 mm Hg (11, −1 to 50) [mean ± standard deviation (SD) (median, range)]. ETCO2 peak was at or above 50 mm Hg with 35 of 94 seizures, 60 mm Hg with 15, and 70 mm Hg with five seizures. Eleven of the 33 patients had seizures with ETCO2 elevation above 50 mm Hg. The duration of ictal/postictal ETCO2 increase above baseline was 424 ± 807 s (154, 4 to 6225). The duration of ictal apnea was 49 ± 46 s (31, 6–222); most ictal apneic events were central. Oxygen desaturation to 60% or less occurred with 10 seizures, including five that did not progress to generalized convulsions. Respiratory rate and amplitude increased postictally. The peak ictal ETCO2 change and duration of change were not associated with apnea duration or seizure duration. Peak ETCO2 change was significantly associated with contralateral seizure spread.

Conclusions:  Severe and prolonged increases in ETCO2 occur with seizures. Postictally, respiratory effort is not impaired. Ictally triggered ventilation–perfusion inequality from pulmonary shunting or transient neurogenic pulmonary edema may account for these findings.

Sudden unexpected death in epilepsy (SUDEP) has an incidence of 0.09–9 per 1,000 patient years, with the highest incidence in patients with refractory epilepsy (Tomson et al., 2008). Both cardiac and respiratory mechanisms are implicated in SUDEP (Surges et al., 2009). Seizures are associated with hypoxemia (Hewertson et al., 1996; Nashef et al., 1996; Blum et al., 2000; Bateman et al., 2008). We demonstrated a high incidence of ictal/postictal hypoxemia in patients with localization-related epilepsy undergoing inpatient video-EEG telemetry (VET) (Bateman et al., 2008). Ictal hypoxemia may be severe and prolonged in partial-onset seizures (Bateman et al., 2008). Respiratory changes in the ictal and postictal period are not well characterized. End-tidal CO2 (ETCO2) measurements correlate directly with changes in alveolar PCO2 and may, therefore, allow a precise evaluation of seizure-related respiratory disturbances. In contrast, oxygen saturation (SaO2) recorded by digital pulse oximetry has a nonlinear relationship with the partial pressure of blood oxygen, and there is little change in SaO2 until PO2 values have dropped below 60 mm Hg (West, 1977).

We now expand our observations and report on respiratory changes in the ictal/postictal period in patients with localization-related epilepsy undergoing VET. We studied the magnitude and duration of ETCO2 change associated with seizure type, seizure duration, duration of ictal apnea, and with changes in SaO2. Changes in airflow amplitude and respiratory rate (RR) in the postictal period were characterized. These findings provide the first detailed analysis of seizure-related respiratory changes and expand on potential respiratory mechanisms of SUDEP.

Methods

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

Thirty-three consecutive patients (13 female) with medically refractory localization-related epilepsy were admitted for VET as part of their presurgical work-up and studied using standard recordings at this epilepsy monitoring unit. None of the patients had preexisting known cardiac or pulmonary disease. Video data, SaO2 using digital pulse oximetry (Nellcor N-395 pulse oximeter, Covidien-Nellcor, Boulder, CO, U.S.A.), ETCO2 using nasal cannulae and a capnograph (BCI Inc., Waukesha, WI, U.S.A.), nasal airflow using a pressure transducer, and abdominal respiratory excursions using inductance plethysmography were recorded synchronized with the EEG. SaO2 values were generated each second and ETCO2 values every 4 s. The capnograph has an average delay time of 1.8 s. Antiepileptic drugs (AEDs) were either reduced or stopped in an attempt to provoke seizures. Medication taper was individualized for each patient. The study was approved by the local institutional review board.

Seizure characteristics were noted, including seizure localization at onset, time of onset, time to contralateral spread, time to onset of generalized convulsion, and seizure duration. The following were also noted: onset of desaturation (defined as a drop in oxygen saturation <90%), duration of desaturation (defined as the time from drop in saturation <90% to return of saturation ≥90%), duration of recovery of saturation to within 2% of preictal baseline (defined as the mean saturation in the interval between 2 and 1 min prior to seizure onset), oxygen saturation nadir and time of nadir, preictal ETCO2 (mean of ETCO2 values between 2 and 1 min prior to onset of seizure), peak ictal/postictal ETCO2, time of return of ETCO2 to within 2 mm Hg of the preictal level (defined as the first of five consecutive ETCO2 values within 2 mm Hg of preictal baseline), the preictal RR (defined as the mean RR recorded over a 20 s period starting at 2 min before seizure onset), the RR on resumption of regular respirations in the immediate postictal period averaged over 20 s, the peak-to-peak amplitude of the airflow signal at baseline and in the immediate postictal period, time of onset of apnea, and duration and type of apnea (central, obstructive, or mixed) when present. The mean preictal heart rate (HR) averaged over 10 s starting at 2 min before seizure onset, time of onset of tachycardia defined as increase in HR above 100 beats per minute (bpm), peak HR, and duration of tachycardia were recorded. Epochs with muscle and movement artifact obscuring the electrocardiographic channel were excluded. The peak-to-peak amplitude of the airflow signal was transformed by obtaining square root values of the amplitude signal at each data point. This transformation corrects for nonlinearities and approximates airflow values obtained from a pneumotachograph (Montserrat et al., 1997). Brief electrographic seizures lasting 10 s or less were excluded from analysis. Probable simple partial seizures without a clear electrographic change were excluded from analysis.

Statistical methods

The primary endpoint for determination of sample size was a demonstration of a rise of ETCO2 of 3 mm Hg or more with a desaturation below 90% for each seizure. It was determined that with a two-sided two-sample t test with the assumption of equal variances at a significance level = 0.05 and a power of 0.9, a sample size of 120 seizures (60 in each group) would be required.

Seizure duration, duration of apnea, and contralateral spread of seizures were initially treated as independent variables, and the relationship of these to changes in ETCO2 was determined. Because outliers and leverages were detected, a univariate robust linear regression model with MM estimation method was used to study the relationship between a response and independent variable in each case (Yohai, 1987). A p-value < 0.05 was considered statistically significant. Next, the independent variables with a p < 0.3 were used as candidate variables and included in a multivariable robust regression model for multivariable analysis. All analyses were performed with SAS version 9.2 (SAS Institute Inc, Cary, NC, U.S.A.).

Locally weighted scatterplot smoothing (LOESS) (Cleveland, 1979; Cleveland & Devlin, 1988) was used to fit ETCO2 data points following the peak ETCO2 value. The 95% point-wise confidence interval was determined. The LOESS fit was generated with R 2.8.1, which is available at: http://cran.us.r-project.org/.

Results

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

Summary statistics

One hundred eighty-seven seizures were captured in 33 patients (13 female). The mean age was 36.4 ± 13.3 years (35, 18–66) [mean ± SD (median, range)]. The body mass index (BMI) was 27.3 ± 5 (25.6, 20.5–37.6). One hundred sixty seizures were of temporal onset, 16 of frontal onset, 1 of central onset, and in 10 seizures the onset was undetermined. Seizure lateralization was left in 100, right in 74, undetermined in 9, and had a near simultaneous bilateral onset in 4. Additional clinical details are provided in Table S1. One patient was studied twice, initially with noninvasive scalp monitoring and several months later again with intracranial electrodes.

Five patients were on AED monotherapy; the remaining patients were on two or more AEDs prior to reductions of AEDs. Forty-one of the 187 seizures progressed to generalized convulsions. Ictal/postictal ETCO2 peak values were available for 94 seizures. SaO2 nadir values were available for 163 seizures, including 36 seizures that progressed to generalized convulsions.

For all 187 seizures the seizure duration was 111 ± 95 s (84, 17–724). The time to contralateral spread of seizure was 36 ± 52 s (19, 0–375). The time to secondarily generalized convulsion was 64 ± 47 s (45, 13–240).

The baseline ETCO2 was 35 ± 6 mm Hg (37, 19–47). The patient with the baseline ETCO2 of 19 mm Hg had been hyperventilating to provoke a seizure. The peak ictal/postictal ETCO2 was 49 ± 11 mm Hg (47, 30–94). The ictal/postictal ETCO2 increase from baseline was 14 ± 11 mm Hg (11, −1 to 50). For the 94 seizures in which ETCO2 information was available, peak ETCO2 increases to 50 mm Hg or above occurred with 35 seizures with or without secondarily generalized convulsions. ETCO2 was at or above 60 mm Hg and 70 mm Hg with 15 and 5 seizures, respectively (Fig. S1A). Eleven of the 33 patients had seizures with ETCO2 elevation above 50 mm Hg. Of these 11 patients, in 6 the ETCO2 was above 60 mm Hg and in 2 above 70 mm Hg for at least one seizure. The ETCO2 peak occurred 175 ± 118 s (158, 10–660) after seizure onset and 56 ± 111 s (27, −318 to 552) after seizure termination. The ETCO2 peak occurred 45 ± 86 s (32, −224 to 271) after the SAO2 nadir. The duration of the ictal/postictal ETCO2 increase was 424 ± 807 s (154, 4 to 6225) (Fig. S1B).

The baseline SaO2 on room air was 97 ± 2% (98, 92–100). The SaO2 nadir on room air for seizures that did not progress to generalized convulsions was 88 ± 11% (92, 43–100). With secondary generalized convulsions the SaO2 nadir on room air was 74 ± 11% (75, 50–91). The time from seizure onset to onset of desaturation below 90% was 81 ± 62 s (67, −2 to 431). The time from seizure onset to SaO2 nadir was 111 ± 70 s (96, 22−469). The duration of desaturation (<90% to ≥90%) was 80 ± 89 s (59, 4–712). The time from onset of desaturation below 90% to return of saturation to within 2% of baseline SaO2 was 176 ± 281 s (103, 13 to 2036).

Following partial seizures that did not evolve into generalized convulsions, peak ETCO2 values declined rapidly over 500 s, followed by a more gradual decline over the next 1000 s. The ETCO2 values remained at least 3 mm Hg above preictal levels for 1,500 s. A similar decline from peak ETCO2 values occurred following secondarily generalized convulsions (Fig. 1).

image

Figure 1.   Locally weighted scatterplot smoothing (LOESS) plots for the change in end-tidal CO2 (ETCO2) values relative to preictal baseline. The tracings show the postictal decline in ETCO2 values starting with the greatest ETCO2 change recorded in the ictal/postictal period. The fitted curve (red) and 95% confidence intervals (blue) are shown. (A) shows the decay from peak ETCO2 change with partial onset seizures that progressed to generalized convulsions. (B) shows the decay from peak ETCO2 change with partial seizures that did not progress to generalized convulsions.

Download figure to PowerPoint

The duration of apnea was 49 ± 46 s (31, 6–222). There were 67 central, 8 obstructive, and 9 mixed apneas (apneas with both central and obstructive components). There was inadequate information to characterize the apnea in six seizures, as abdominal movement information was not available. There were three hypopneas. Both airflow and abdominal movement information was unavailable with 58 seizures. No apnea was detected with the remaining seizures.

The RR in the preictal baseline was 18.6 ± 5.2 breaths per minute (18, 8–36) and increased to 25.7 ± 10.0 breaths per minute (24, 12–60) in the immediate postictal period. The increase in RR was statistically significant (p < 0.001). The square root transformed amplitude of the peak to peak airflow signal at baseline was 3.2 ± 2.6 (2.7, 0.04–7.6) and in the immediate postictal period was 11.7 ± 7.7 (13.0, 0.36–28). The postictal airflow amplitude was significantly higher than the amplitude during preictal baseline (two-sided Wilcoxon signed-rank test, p = 0.001) (Fig. S2).

The preictal heart rate was 85 ± 19 bpm (84, 54–192). The peak ictal heart rate was 131 ± 27 bpm (126, 72–210). The ictal heart rate change was 45 ± 27 bpm (42, −12 to 120). The duration of ictal tachycardia was 329 ± 737 s (80, 10−4,251).

Associations between seizure characteristics and respiratory changes

In all seizures there was a significant association between the peak ictal/postictal change in ETCO2 relative to baseline and the peak change in SaO2 (p = 0.0001). In the subgroup of seizures that evolved to generalized convulsions, there was a significant association between the SaO2 nadir and peak ETCO2 in the ictal/postictal period (p = 0.0001). In the seizures that did not evolve into secondarily generalized convulsions, there was no association between the SaO2 nadir and peak increase in ETCO2 in the ictal/postictal period relative to baseline (p = 0.287).

Duration of apnea and change in ETCO2 and SaO2

For all seizures, there was no statistically significant association between the peak ictal/postictal change in ETCO2 and duration of ictal apnea (p = 0.398). The duration of ETCO2 increase above preictal baseline was not associated with the duration of ictal apnea (p = 0.686). The peak increase in ETCO2 was marginally associated with the duration of seizures that did not progress to generalized convulsions (p = 0.052). With secondarily generalized convulsions, the peak oxygen desaturation was not associated with the duration of ictal apnea (p = 0.721). For seizures that did not evolve to generalized convulsions, the SaO2 nadir was associated with apnea duration (p = 0.004). The duration of ictal/postictal oxygen desaturation relative to preictal baseline was not associated with the duration of ictal apnea (p = 0.825).

Relationship of ETCO2 change to contralateral seizure spread

There was a statistically significant association between ictal/postictal change in peak ETCO2 and contralateral spread of seizures, whether generalized convulsions occurred or not, compared with seizures that did not spread to the contralateral hemisphere (p < 0.0001).

Relationship of seizure duration and change in ETCO2 and SaO2

For all seizures, the peak ictal/postictal increase in ETCO2 did not correlate with seizure duration (p = 0.074). The peak change in ETCO2 did not correlate with the duration of secondarily generalized seizures (p = 0.138). The duration of ETCO2 increase above baseline did not correlate with seizure duration in seizures that did not progress to generalized convulsions (p = 0.103). The duration of oxygen desaturation was significantly associated with seizure duration (p = 0.005).

There was no correlation between seizure duration and apnea duration (p = 0.137, Spearman rank correlation coefficient 0.203). No association was demonstrated between apnea duration and contralateral spread of seizures (p = 0.515, two-sided Wilcoxon rank-sum test). An association was demonstrated between seizure duration and contralateral spread of seizures (p = 0, two-sided Wilcoxon rank-sum test). Contralateral spread of seizures and seizure duration that had p < 0.3 from univariate analysis were used as candidate variables and included in the multivariable robust linear regression model. Multivariable analysis showed the p-values of contralateral spread of seizures and seizure duration of p ≤ 0.0001 and p = 0.006, respectively, indicating that the contralateral spread of seizures and seizure duration are jointly associated with ETCO2 change.

Discussion

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

Oxygen desaturation below 90% occurs in one-third of seizures in patients with localization-related epilepsy and occurs with partial seizures that do not progress to generalized convulsions as well as with secondarily generalized convulsions (Bateman et al., 2008). Patient and seizure characteristics associated with ictal hypoxemia have been characterized (Bateman et al., 2008; Seyal & Bateman, 2009). We have shown that pronounced ictal/postictal peak ETCO2 elevations, in some cases exceeding 90 mm Hg, occur with partial-onset seizures (Fig. S1A). The increase above baseline in ETCO2 persisted for a mean of >7 min, and with some seizures the ETCO2 did not return to baseline for >15 min following the end of the seizure (Figs. 1 and S1B). Severe acidosis could potentially result from such marked seizure-related increases in ETCO2. An abrupt increase in ETCO2 to 90 mm Hg would result in a pH of about 7.05 as predicted by the Henderson-Hasselbalch equation. Although the most marked changes in ETCO2 and SaO2 occurred with secondary generalized convulsions, severe oxygen desaturations (<60%) and ETCO2 elevations (>50 mm Hg) occurred with some seizures that did not progress to generalized convulsions.

Peak ictal/postictal elevation in ETCO2 and duration of ETCO2 increase above baseline were not significantly associated with the duration of ictal apnea with seizures that progressed to generalized convulsions. The brevity of ictal apnea relative to the duration of ETCO2 increase and oxygen desaturation makes it likely that other factors, such as increased metabolic activity with convulsive seizures and/or ictal-related pulmonary dysfunction, may contribute to the transient rise in ETCO2 disproportionate to the decrease in SaO2. In the subgroup of seizures that did not progress to generalized convulsions, the peak ictal/postictal elevation in ETCO2 was marginally associated with the duration of ictal apnea.

The drop in SaO2 was significantly associated with the duration of ictal apnea in partial seizures that did not generalize. Therefore, in nonconvulsive seizures, the duration of ictal apnea rather than an increase metabolic activity or pulmonary dysfunction may be a major contributor to ETCO2 elevation and oxygen desaturation. Seizure duration affects the degree of respiratory dysfunction, as seizure duration is associated with the depth of oxygen desaturation. A similar trend was present for an association of seizure duration and peak elevation of ETCO2, but this did not reach statistical significance. Multivariable analysis indicates that contralateral seizure spread and seizure duration jointly affect ETCO2 change.

Although total ventilation was not directly assessed, the postictal increase in RR and the increase in amplitude of the peak-to-peak airflow signal suggest that postictal depression of ventilatory effort is not the likely cause of hypercapnia. The square root of nasal flow measured from nasal prongs connected to a pressure transducer acceptably fits flow data measured with a pneumotachograph (Montserrat et al., 1997). Therefore, the prolonged postictal elevation of ETCO2 may be a consequence of pulmonary dysfunction rather than a decrease in postictal ventilatory effort.

Hypoventilation and ventilation–perfusion inequality are the two major causes for CO2 retention (West, 2008). Ventilation–perfusion inequality may arise from intrapulmonary right-to-left shunting of blood. Pulmonary shunting of venous blood results in a drop in PaO2 and consequent decrease in SaO2. This hypoxemia cannot be completely abolished by giving the patient 100% O2 to breathe. Fixed shunts usually do not result in raised PCO2 because chemoreceptor responses to hypercapnia result in an increase in minute ventilation until PaCO2 is normalized. Transient PaCO2 increases may be expected if neurogenic mechanisms result in acute but temporary pulmonary shunting of venous blood. Transient intrapulmonary shunting can occur in normal humans. Using bubble contrast echocardiography, it was shown that normoxic and hypoxic exercise results in intrapulmonary shunting with the appearance of bubbles in the left heart. The shunting persisted for up to 5 min postexercise and was associated with significantly worsened gas exchange. No shunting occurred preexercise in these healthy adult subjects (Lovering et al., 2008). Exercise-induced arteriovenous shunting via large-diameter arteriovenous channels occurs in dogs, and these shunts accounted for up to 3.1% of cardiac output (Stickland et al., 2007). It is possible that similar shunting occurs with seizures, which may elicit similar metabolic demands and/or physiologic responses to vigorous exercise. Pulmonary corner vessels are large-diameter vessels in corner pleats of the alveolar wall. The corner vessels represent functional, not anatomically fixed, shortcuts between arteries and veins, which permit bypass of parts of the alveolar capillary network (Ciurea & Gil, 1996). Seizures may directly influence pulmonary blood flow, possibly by opening these vessels, as there is evidence for autonomic neural regulation of pulmonary blood flow. A variety of neurotransmitters have been implicated in the control of pulmonary vascular tone (Barnes & Liu, 1995).

Transient pulmonary edema is another potential explanation for the respiratory disturbances observed in this study. Hypercapnia (with respiratory acidosis) occurs in about 20% of cases of pulmonary edema, and in these cases respiratory muscle fatigue or severe ventilation perfusion mismatch may be responsible mechanisms (Prichard & Firth, 2005). Neurogenic pulmonary edema can occur following seizures. About 40 cases of seizure-related pulmonary edema have been reported (Cho et al., 2002) since the first description in 1908 (Shanahan, 1908). It is likely that the incidence of seizure-related pulmonary edema is higher than suggested by the reported cases, as cases that are mild and transient are less likely to be recognized. The alveolar infiltrates of acute neurogenic pulmonary edema most often occur immediately after the seizure and resolve quickly, although the edema may be delayed in some cases (Fredberg et al., 1988). Recurrent seizure-related pulmonary edema has been reported (Darnell & Jay, 1982). The mechanisms of neurogenic pulmonary edema remain speculative. Two mechanisms have been suggested: (1) a hemodynamic mechanism from intense pulmonary vasoconstriction secondary to the adrenergic response associated with a seizure, resulting in increased pulmonary hydrostatic pressure and increased permeability of pulmonary capillaries and (2) an inflammatory mechanism inducing an increase in the permeability of pulmonary capillaries (Baumann et al., 2007). Ictal neurogenic pulmonary edema has been reported in cases of SUDEP (Terrence et al., 1981; Swallow et al., 2002). None of our patients reported dyspnea; however it is likely that in the postictal state transient symptoms of respiratory distress may not be recognized and reported.

We have shown that severe transient respiratory changes with hypercapnia and hypoxemia can occur with seizures. Animal data indicate that these changes may predispose to SUDEP. In a sheep model of seizure-associated sudden death, severe elevations in PaCO2 and drops in PaO2 occurred in animals dying early after seizures without the development of malignant arrhythmias (Johnston et al., 1997). Ictal-related acute hypoxemia and hypercapnia may also influence the cardiac rhythm. There is prolongation of the QTc and increased QT dispersion with acute hypercapnia, which could represent a mechanism for ictal cardiac arrhythmia (Kiely et al., 1996). Studies in the rat heart have suggested that acidosis slows cardiac pacemaker activity, particularly at the A-V node (Aberra et al., 2001). Ventricular fibrillation has been demonstrated in a canine model of anoxia or asphyxia. It was also shown that in the presence of hypercarbia and acidosis, the animals terminated in asystole (Kristoffersen et al., 1967).

Hypercapnic acidosis may limit seizure discharges in patients with epilepsy and in experimental models of epilepsy (Lennox, 1929; Mitchell & Grubbs, 1956; Woodbury et al., 1956). However, the severe hypoxemia and hypercapnia that occur with some seizures may be sufficient to lead to SUDEP directly or indirectly, for example, by precipitating cardiac arrhythmias. The amelioration of seizure-associated respiratory dysfunction should be considered a potential therapeutic target for SUDEP prevention.

Acknowledgments

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

Statistical support for this publication was made possible by Grant Number UL1 RR024146 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH), and NIH Roadmap for Medical Research. Its contents are solely the responsibility of the authors and do not necessarily represent the official view of NCRR or NIH. Information on Re-engineering the Clinical Research Enterprise can be obtained from http://nihroadmap.nih.gov/clinicalresearch/overview-translational.asp.

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.

References

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure
  8. References
  9. Supporting Information

Figure S1. (A) shows the recorded peak ictal/postictal ETCO2 (mmHg) in 94 of 187 consecutive seizures. Missing values are left blank. (B) shows the duration of ETCO2 increase above baseline for these 94 seizures.

Figure S2. Square root transformed airflow amplitude signal in three seizures from three patients. The time base is 400  s. The bar at the bottom of each tracing indicates the seizure duration. The hatched portion of each bar indicates the time for which the seizure remained ipsilateral. The black portion of the bar indicates the duration of contralateral spread. In each case the amplitude of the airflow signal increases in the postictal period. The top trace is from a right temporal onset seizure that progressed to a generalized convulsion 39  s after onset. There is cessation of airflow with a central apnea. The postictal oxygen desaturation nadir was 52%. The middle trace is from a right temporal seizure. There is cessation of airflow with contralateral seizure spread. The seizure did not progress to a generalized convulsion. Oxygen saturation dropped to 52%. The bottom trace is from a right temporal seizure. There is cessation of airflow with a central apnea. The postictal desaturation nadir was 91%.

Table S1. The relevant clinical details for the 33 patients are shown in this table.

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
EPI_2518_sm_FigS1.tif1019KSupporting info item
EPI_2518_sm_FigS2.tif4590KSupporting info item
EPI_2518_sm_TableS1.doc223KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.