Seizure Detection Using SEDline During Therapeutic Hypothermia in Cardiac Arrest Victims: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Seizure Detection Using SEDline During Therapeutic Hypothermia in Cardiac Arrest Victims (NCT01946802), a registered phase not stated study. The primary outcome is Percentage of Seizure-positive and Seizure-negative That Were Correctly Classified by the SEDline in All Evaluated Cases.

2. Clinical Question

The clinical problem described in the registry record is that sedatives and neuromuscular blocking agents used to limit shivering during therapeutic hypothermia can make seizure detection difficult. Frequent or continuous EEG monitoring is recommended in that setting, but the record notes practical barriers related to specialized equipment and personnel. The study asked whether a frontal 4-channel device could detect seizures during this period.

This is a diagnostic question, not a treatment comparison. The study evaluates how SEDline classifications correspond with conventional EEG interpretations; it does not compare clinical outcomes under different monitoring strategies. The record says the EEG methods were performed simultaneously for 30 minutes during therapeutic hypothermia and rewarming, within 72 hours after cardiac arrest. It does not report the number of seizure-positive cases, the number of seizure-negative cases, or how often the two methods agreed. Consequently, the diagnostic question is identifiable, but its answer cannot be quantified from the registry record.

3. Trial Design

This was a completed device study of frontal 4-channel EEG monitoring for seizure detection in patients receiving postresuscitation care after cardiac arrest. The intervention is listed as “Frontal 4 channel EEG.” The registry describes the enrolled group as consecutive comatose patients admitted to an emergency ICU after successful cardiopulmonary resuscitation for nontraumatic out-of-hospital or in-hospital cardiac arrest. The reported enrollment was 39.

The study used conventional EEG as the reference method and SEDline as the device under evaluation. The registry describes simultaneous monitoring for 30 minutes during therapeutic hypothermia and rewarming, within 72 hours after cardiac arrest. Data from both devices were to be interpreted and analyzed for the presence of seizure. No random allocation or parallel treatment-control comparison is described. The allocation method is not stated in the registry record, and the study should not be read as a randomized comparison of clinical interventions.

The record reports that 44 people provided informed consent and that 5 were withdrawn. The listed reasons were death before EEG, a relative withdrawing the patient, combined septic shock, and failure to perform EEG within 72 hours. The breakdown is 1 death before EEG, 2 relative withdrawals, 1 combined septic shock, and 1 patient without EEG within 72 hours. The record then states that 39 patients were finally enrolled. It does not provide further detail on whether all enrolled patients had interpretable paired recordings or were included in every outcome analysis.

The study started in 2014-12, had primary completion in 2016-10, and is marked completed. The sponsor is listed as Gil Joon Suh. Although the registry lists six outcome measures, it reports no analysis count. Details such as assessor masking, rules for interpreting recordings, prespecified diagnostic thresholds, and handling of technically inadequate EEG data are not reported in the registry record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: 53 were screened for eligibility. 9 were excluded (4 Advanced directives to withhold or withdraw life-sustaining treatment, 3 No informed consent, 1 Known intracranial pathology, and 1 Visible generalized seizure). 44 provided informed consents. 5 were withdrawn(see baseline characteristics). Finally 39 patients were enrolled.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: 44 provided informed consents. Among them 5 were withdrawn. 1 death before EEG, 2 relative withdrew patient, 1 combined septic shock, and 1 patient did not perform EEG within 72 hours. Finally 39 patients were enrolled in this study. Conventional EEG (gold standard for seizure detection) and SEDline monitoring were conducted simultaneously for 30 minutes at During therapeutic hypothermia and rewarming.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

Population: 44 provided informed consents. Among them 5 were withdrawn. 1 death before EEG, 2 relative withdrew patient, 1 combined septic shock, and 1 patient did not perform EEG within 72 hours. Finally 39 patients were enrolled in this study.

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Seizure-positive and Seizure-negative That Were Correctly Classified by the SEDline in All Evaluated Cases, measured over Within 72 hours after cardiac arrest.

The primary endpoint is described as the percentage of seizure-positive and seizure-negative cases correctly classified by SEDline among all evaluated cases. In practical terms, this concerns whether the device's seizure classification agrees with the reference interpretation from conventional EEG. The stated timeframe is within 72 hours after cardiac arrest, with simultaneous conventional EEG and SEDline monitoring described for 30 minutes during therapeutic hypothermia and rewarming.

Diagnostic classification is useful because an overall percentage can describe agreement across evaluated cases. However, an overall percentage alone can conceal different performance for seizure-positive and seizure-negative cases. The registry also lists sensitivity, specificity, positive predictive value, negative predictive value, and area under the receiver operating characteristics curve among the secondary outcomes. These measures address different aspects of classification: sensitivity concerns detection among cases with seizures, while specificity concerns correct classification among cases without seizures. Predictive values describe the meaning of a positive or negative device result in the evaluated sample, and the area under the curve describes discrimination across possible classification thresholds.

The record does not state the numerical primary endpoint result, the counts in the seizure-positive and seizure-negative groups, or the rule used to define a correct SEDline classification. It therefore does not establish how often the device agreed with conventional EEG or how precise that estimate was. The endpoint is clinically relevant to the stated diagnostic purpose, but its registry entry is insufficient to assess performance quantitatively.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does diagnostic accuracy mean?

In general, diagnostic accuracy describes how closely a test's classifications correspond to a reference method or reference diagnosis. In this study, conventional EEG is identified as the gold standard, while SEDline is the device being evaluated. A correct classification would therefore depend on the SEDline interpretation matching the conventional EEG classification under the study's rules. The registry record does not specify those rules or report the observed classifications, so the meaning can be explained conceptually but the study's accuracy cannot be reconstructed.

The listed primary endpoint concerns correct classification among seizure-positive and seizure-negative cases. Sensitivity and specificity separate these two types of classification. In general, sensitivity is the proportion of reference-positive cases classified as positive by the test; specificity is the proportion of reference-negative cases classified as negative. A single overall percentage can be influenced by how many evaluated cases belong to each reference category. Reporting the category-specific measures alongside an overall measure helps show whether performance differs between seizure detection and exclusion. The registry lists those measures as outcomes, but does not report their estimates or denominators.

What do predictive values and the area under the curve add?

In general, positive predictive value describes the proportion of positive test classifications that are reference-positive, while negative predictive value describes the proportion of negative test classifications that are reference-negative. These values depend in part on the mix of reference-positive and reference-negative cases in the population being assessed. They may not transfer directly to a setting with a different prevalence of seizures. The registry lists both predictive values but does not report them or the prevalence of seizure classifications in the evaluated sample.

In general, a receiver operating characteristics curve displays sensitivity against the false-positive rate across possible thresholds. Its area under the curve is a summary of discrimination across those thresholds. Interpretation depends on how the device produces or categorizes results and on whether thresholds were prespecified or selected during analysis. The registry lists an area under the curve outcome, but does not give the estimate, the underlying data, or the threshold procedures.

Why does paired measurement matter?

General statistical explanation: when the same clinical episode is assessed by a device and a reference method, the observations are paired rather than independent. A sound comparison must preserve that pairing, because the key information is which classifications agree and which disagree for the same evaluated case. The registry says conventional EEG and SEDline were conducted simultaneously, which is consistent with paired assessment. It does not report the paired classification counts, reader masking, adjudication, or the statistical method used to analyze agreement.

The monitoring description specifies 30 minutes during therapeutic hypothermia and rewarming within 72 hours after cardiac arrest. This time-limited assessment does not, by itself, establish performance during other monitoring durations or outside the described clinical period. The record does not indicate whether multiple EEG segments from an individual were treated as separate observations or whether the unit of analysis was always the patient. That distinction matters in general because observations from the same person can be correlated.

What would a confidence interval tell the reader?

General statistical explanation: a confidence interval conveys the uncertainty around an estimated quantity. A narrow interval indicates greater precision under the analysis assumptions, while a wide interval indicates less precision. For diagnostic measures, precision depends on the relevant classification counts, especially the number of reference-positive cases for sensitivity and reference-negative cases for specificity. The registry record does not report confidence intervals or the classification counts needed to assess precision. No interval should be inferred from enrollment alone.

How should planning calculations be distinguished from observed evidence?

General statistical explanation: a sample-size planning calculation describes a hypothetical design under stated assumptions; it is not an observed result and does not show that a completed study had a particular level of power. The registry record's planning description refers to a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation, with the target effect expressed as a fraction of an observed effect. It also states that time-to-event calculations refer to total events and other calculations to participants per arm.

Those assumptions describe a two-arm planning illustration, whereas the study description is a diagnostic device assessment and does not describe a randomized two-arm comparison. Such an illustration should not be read as the trial's actual design or as evidence of observed diagnostic performance. The registry does not report a sample-size justification, a prespecified design target for diagnostic accuracy, or the observed effect used for such planning. It is therefore not possible to judge from the registry record whether enrollment was adequate for the sensitivity, specificity, predictive-value, or area-under-the-curve estimates.

What does over-dispersion mean in general?

General statistical explanation: over-dispersion occurs when observed variability is greater than a model assumes. It can make standard errors too small and confidence intervals too narrow if the model does not account for the extra variability. It is relevant to some count or repeated-measure analyses, but the registry record does not report a statistical model or an over-dispersion assessment for this study. There is no basis for claiming that over-dispersion was present or that any particular adjustment was used.

The registry reports no analysis count and does not describe an inferential statistical plan. The definition of the endpoint, diagnostic threshold, handling of missing recordings, and uncertainty estimates are not reported in the registry record. These omissions prevent a detailed evaluation of model assumptions and analytic reproducibility.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists six outcome measures, including the primary classification endpoint and secondary diagnostic measures. It does not report an analysis count or describe a plan for handling multiple outcomes. The registry record therefore does not say whether a particular outcome hierarchy was prespecified beyond identifying a primary outcome, whether any adjustment for multiple testing was planned, or how emphasis would be assigned across the secondary measures. This is not evidence that an adjustment was or was not performed; the analysis details are not reported in the registry record.

Missing-data handling is also not described. The record reports 44 people provided informed consent, 5 withdrawals, and 39 patients finally enrolled. It gives reasons for the withdrawals: 1 death before EEG, 2 relative withdrawals, 1 combined septic shock, and 1 patient who did not perform EEG within 72 hours. These figures explain the listed pre-enrollment losses but do not establish whether every enrolled patient contributed evaluable recordings to each analysis. The record does not report incomplete, uninterpretable, or discordant recordings after enrollment, nor does it state how such cases were handled.

The analysis population is not defined beyond the description of enrolled patients and evaluated cases. It is not reported whether the primary outcome included all enrolled patients, only patients with paired interpretable EEG, or another subset. The registry does not provide outcome-specific denominators, a participant flow for analysis, or a method for addressing missing classifications. These gaps matter because diagnostic estimates can change if the cases without evaluable results differ from those included in the analysis. The record does not provide enough information to assess that possibility.

11. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

The registry's safety entry lists 0 serious adverse events and 0 other adverse events for the Frontal 4 Channel EEG group, with 39 at risk in each reporting category. It also records 17 deaths. These are event counts in the registry; the entry does not provide timing, event descriptions, attribution to the device, or an explanation of the reporting window. The figures therefore should not be interpreted as establishing whether the device caused, contributed to, or prevented any event.

The record separately lists 1 death before EEG among the reasons for withdrawal from the consented group. It does not clarify how that pre-EEG death relates to the 17 deaths in the safety entry, including whether the reporting populations and periods overlap. The registry record also does not describe follow-up duration for safety reporting or provide comparative safety data from a control group. With no randomized or like-for-like comparison described, the safety counts cannot support a comparative risk estimate.

A zero count in a registry category means no events were recorded in that category for the stated group; it does not establish that an event is impossible or that all events were captured. The event counts should be read descriptively, with the missing attribution and timing information kept in view.

12. What the Result Does, and Does Not, Mean

The study addresses whether SEDline, a frontal 4-channel EEG device, can classify seizure presence during therapeutic hypothermia and rewarming after cardiac arrest. Conventional EEG is named as the reference standard, and the registry describes simultaneous monitoring for 30 minutes within 72 hours after cardiac arrest. The question is focused and clinically situated, but the registry record does not report the primary endpoint value or the secondary diagnostic estimates.

Accordingly, the record does not establish the percentage of cases correctly classified, how often seizures were detected, how often seizure-negative cases were correctly classified, or how well positive and negative device readings predicted the reference classification. It also does not provide uncertainty intervals. The study's enrollment figure alone cannot substitute for those results, because diagnostic precision depends on the number and pattern of reference classifications and discordant readings.

The record does establish that the study was completed and that it enrolled 39 patients. It describes a group of comatose patients receiving postresuscitation care after nontraumatic out-of-hospital or in-hospital cardiac arrest. It also lists outcomes intended to characterize several aspects of diagnostic performance. These registry entries explain the study's purpose and scope, but they are not evidence that the device met a performance threshold.

The entry is not a randomized comparison of monitoring strategies or patient outcomes. It cannot show that using SEDline changes care, improves outcomes, or is equivalent to conventional EEG. Even if diagnostic accuracy estimates were available, conclusions about clinical consequences would require evidence connecting device classifications to decisions and outcomes. Such evidence is not reported in the registry record.

The safety counts are similarly limited. The registry records 0 serious adverse events, 0 other adverse events, and 17 deaths, but does not provide timing or attribution. These figures cannot establish a causal safety profile. The most defensible interpretation is therefore descriptive: the registry records a completed diagnostic device study with a clearly stated clinical question and a set of intended diagnostic outcomes, but it does not report the performance results needed to answer that question.

13. Why This Trial Matters Statistically

This study matters statistically because diagnostic performance is not captured fully by a single overall classification percentage. A device may classify seizure-positive cases differently from seizure-negative cases, so sensitivity and specificity provide complementary views. Predictive values add information about the meaning of positive and negative readings in the evaluated sample, while the area under the receiver operating characteristics curve can describe discrimination across thresholds. The registry lists these outcomes, making clear that the intended assessment had several dimensions.

The simultaneous use of SEDline and conventional EEG is also important to the comparison. Paired assessment can distinguish agreement and disagreement for the same clinical episode, rather than comparing separate groups exposed to different methods. In general, that design makes it especially important to report the paired classification data, the interpretation rules, and the approach to discordance. Those details and the resulting estimates are not reported in the registry record.

Clinical context affects interpretation. The monitoring is described during therapeutic hypothermia and rewarming after cardiac arrest, when the record notes that sedatives and neuromuscular blocking agents can make seizure detection difficult. Results from this setting, if reported, would need to be interpreted in relation to the stated population and monitoring period rather than assumed to apply to other situations. The registry record does not report enough performance data to judge even this defined setting.

The study also illustrates why enrollment and outcome reporting are not interchangeable. The record reports 39 enrolled patients, but does not report the numbers of seizure-positive and seizure-negative cases or the denominators behind the diagnostic outcomes. Without those details, neither the estimates nor their precision can be assessed. Statistical importance therefore lies both in the study's paired diagnostic question and in the reporting gap that prevents readers from evaluating its answer.

14. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The primary limitation is that the registry record does not report the numerical primary endpoint result or any of the listed secondary diagnostic estimates. It gives no classification counts, confidence intervals, or outcome-specific denominators. Thus, accuracy and precision cannot be evaluated from the registry entry.

The endpoint description does not specify the operational definition of a correct SEDline classification, the device threshold, or how seizure interpretations were made. The record identifies conventional EEG as the gold standard but does not describe assessor masking, adjudication of disagreements, or the handling of ambiguous readings. Those procedural details can affect diagnostic estimates and their reproducibility.

The monitoring period is described as 30 minutes during therapeutic hypothermia and rewarming, within 72 hours after cardiac arrest. The registry does not explain how that recording period was selected or whether the device was assessed over additional intervals. The results, if available, would therefore need to be interpreted within the described recording conditions rather than generalized to other monitoring durations or settings.

The participant accounting leaves the analysis population unclear. The registry reports 44 consents, 5 withdrawals with stated reasons, and 39 enrolled patients, but does not report how many contributed usable paired recordings or were included in each endpoint analysis. Post-enrollment missingness and exclusions are not described. The absence of a control group also limits the study to diagnostic comparison rather than comparative clinical outcomes. Finally, safety counts lack timing and attribution, so they cannot establish device-related risk.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Seizure Detection Using SEDline During Therapeutic Hypothermia in Cardiac Arrest Victims — NCT01946802.
• Registry record last updated 2021-03-23; status last verified 2021-03.
• Results first posted 2021-03-23.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

18. Record Summary

NCT01946802 is a completed device study of SEDline frontal 4-channel EEG for seizure detection during therapeutic hypothermia and rewarming after cardiac arrest. The registry describes comatose patients receiving postresuscitation care after nontraumatic out-of-hospital or in-hospital cardiac arrest and reports 39 enrolled patients. Conventional EEG is identified as the gold standard, with simultaneous monitoring described for 30 minutes within 72 hours after cardiac arrest.

The primary outcome concerns the percentage of seizure-positive and seizure-negative cases correctly classified by SEDline. Sensitivity, specificity, positive predictive value, negative predictive value, and area under the receiver operating characteristics curve are also listed. The registry record does not report their numerical results, confidence intervals, or classification counts, so diagnostic performance cannot be judged from this entry. Safety counts list 0 serious adverse events, 0 other adverse events, and 17 deaths, without timing or attribution details.