The association of brainstem and motor recovery with awakening after out-of-hospital cardiac arrest

Abstract

Background

Coma is common following resuscitation from cardiac arrest. Few data describe the trajectory of recovery the first days following resuscitation. The objective of this study is to describe the evolution in neurological examination during the first 5 days after resuscitation and test if subjects who go on to awaken have different patterns of early recovery.

Methods

Prospective study of adult subjects resuscitated from out-of-hospital cardiac arrest. We abstracted demographic information and trained clinicians completed daily neurologic examinations using the Glasgow Coma Scale (GCS) and Full Outline of UnResponsiveness brainstem (FOUR-B) and motor (FOUR-M) scores during daily sedation interruption. The change in scores between Day 1 and Day 5 was analyzed using the Kruskal-Wallis Test and logistic regression models. The relationship of FOUR-B, FOUR-M, and GCS with time to death was estimated by fitting cox proportional hazard models.

Results

FOUR-M and GCS did not differ over time (p = 0.10; p = 0.07). FOUR B increased over time (p < 0.01). Time to recovery of brainstem or motor function differed between those treated at 33 °C and 36 °C (p = 0.0023 and p = 0.0032, respectively). FOUR-B, FOUR-M, and GCS differed between survivors and non-survivors (p < 0.01). Time to recovery of brainstem and motor function differed between survivors and non-survivors. FOUR-M and FOUR-B differed between those with good outcome and poor outcome.

Conclusions

The brainstem clinical examination improved during the first 5 days following resuscitation. Brainstem recovery was common in entire cohort and did not differentiate between survivors and non-survivors. Recovery of motor function, however, was associated with survival.

Introduction

Out-of-hospital cardiac arrest (OHCA) affects more than 345,000 Americans each year, but only 9 % of patients survive to hospital discharge.¹ Many patients with return of spontaneous circulation (ROSC) are comatose on arrival to the hospital. Failure to awaken strongly predicts in-hospital mortality.2, 3 Previous prognostic studies focus on accurately identifying patients who will not recover.4, 5, 6 However, identification of early clinical characteristics that predict a likely recovery and good outcomes after cardiac arrest is arguably equally important to inform medical decision making and families. To this end, we describe the trajectory of neurological recovery in patients resuscitated from OHCA and test the hypothesis that specific patterns of change in the neurological examination during the first five days after OHCA that predict awakening (the ability to follow commands). Secondarily, we examined whether brainstem and cortical function vary independently. Target temperature and its association with the trajectory of neurological exam was also analyzed.

Methods

The University of Pittsburgh Institutional Review Board approved all aspects of this work. We completed a prospective cohort study enrolling consecutive adult subjects treated at a single academic medical center after resuscitation from OHCA from June 2015 to December 2015 and during September 2016. We excluded patients who were awake (i.e. following verbal commands), not living independently prior to arrest, pregnant, incarcerated, arrested secondary to presumed drug overdose (as it may confound the exam), in-hospital cardiac arrests, as well as those with a surgical etiology of arrest (considered in-hospital arrest), evidence of brain edema on CT scan (poor gray-white matter differentiation within 24 h of arrest correlated with poor neurological outcome and mortality),6, 7, 8, 9 active do not resuscitate order, or withdrawal of life-sustaining therapy within 6 h of hospitalization.

Following enrollment, a neurological exam was performed each morning for the first 5 days during a daily sedation interruption. During the study, it was routine clinical practice to discontinue all sedation on every comatose subject at least daily, unless there was a clinical indication that made sedation interruption unsafe. We performed neurological assessments after sedation was stopped for at least 40 min unless this was not tolerated by the patient. Contraindications or clinical indications for early termination of sedation interruption were: overt hemodynamic instability, severe hypoxemic respiratory failure or ventilator dysynchrony, sedation needed for seizure control, neuromuscular blockade or comfort measure only care.

We evaluated the motor response beginning with a normal verbal stimulus, which consisted of the clinician greeting the patient and introducing themselves. If the patient did not respond, we provided a loud verbal stimulus by shouting the patient’s name. If the patient did not respond to both normal and loud verbal stimuli, the clinician would provide noxious stimuli to each of the four extremities by either pinching the skin or applying pressure to a nailbed. If the patient did not respond to peripheral painful stimuli, we provided central noxious stimulation by pinching the trapezius muscle or applying supraorbital pressure. We tested brainstem function with the following maneuvers: pupil response to light, corneal response to saline or sterile gauze, gag by suctioning of pharynx, cough by advancing a suction catheter through the endotracheal tube into the lower airways, and oculocephalic stimuli by rapid rotation of the head. We recorded the results of the best examination obtained during each days’ sedation interruption. We summarized these findings using the Glasgow Coma Scale (GCS) and the motor and brainstem components of the Full Outline of UnResponsiveness Score (FOUR Score).10, 11, 12, 13 See Table 2a for reference on components of FOUR Score.

Table 2.

General Results (Median and Interquartile Range). FOUR M- Full Outline of UnResponsiveness Motor exam; FOUR B- Full Outline of UnResponsiveness Brainstem exam; GCS- Glasgow Coma Scale.

Day Post-Arrest	Cohort FOUR M	Cohort FOUR B	Cohort GCS	Survived FOUR M	Survived FOUR B	Survived GCS	Died FOUR M	Died FOUR B	Died GCS
Day 1	1 (0,3)	2 (2,4)	6 (3,10)	2 (0,3)	4 (2,4)	9 (3,12)	0 (0,2)	2 (1,3)	3 (3,7)
Day 2	2 (0,4)	4 (2,4)	6 (3,14)	4 (2,4)	4 (4,4)	13 (7,15)	0 (0,2)	3 (1,4)	3 (3,6)
Day 3	3 (0,4)	4 (4,4)	10 (4,15)	4 (4,4)	4 (4,4)	15 (12,15)	1 (0,2)	4 (2,4)	4 (3,7)
Day 4	4 (1,4)	4 (4,4)	10 (4,15)	4 (3,4)	4 (4,4)	15 (8,15)	1 (0,2)	4 (2,4)	4 (3,7)
Day 5	4 (0,4)	4 (4,4)	13 (4,15)	4 (4,4)	4 (4,4)	15 (10,15)	0 (0,1)	4 (2,4)	4 (3,7)

We obtained baseline demographic information from the medical record, including patients’ age, sex, initial arrest rhythm, and Pittsburgh Cardiac Arrest Category (PCAC). PCAC is a validated measure of post-arrest illness severity assessed on hospital presentation. There are four categories: I-Awake, II- Coma without severe cardiopulmonary failure, III- Coma with severe cardiovascular or pulmonary failure, and IV- Coma with loss of at least some brainstem reflexes.10, 11 We treated all subjects with a standardized care plan including targeted temperature management and continuous electroencephalography (EEG).14, 15, 16 Locally, we choose 36C for patients with PCAC II illness severity, 33C for patients with PCAC IV illness severity and a combination of 33C or 36C for patients with PCAC III illness severity. In our facility, a multimodal approach to neurologic prognostication performed over the first 3–5 days in patients who have not yet awakened from coma incorporates clinical examination, brain imaging (computerized tomography on all patients on day one and magnetic resonance imaging when indicated if the patient did achieve awakening by day 3), and somatosensory evoked potentials.¹⁷

Our primary outcome was awakening from coma, which we defined as the ability to follow verbal commands. The secondary outcomes were survival to hospital discharge and good outcome at hospital discharge, which we measured using Cerebral Performance Category (CPC) and Modified Rankin Scale (mRS). We considered a good outcome to be a CPC of 1–2 or mRS of 0–3 at discharge.

We summarize baseline demographics using descriptive statistics. We used Kruskal-Wallis tests to analyze exam trends and constructed Kaplan Meier survival curves to compare time to motor or brainstem recovery stratified by survival status and by target temperature during initial care (33 °C vs 36 °C). We used Stata v13.0 (StataCorp, College Station, TX). To test discrimination of the FOUR-B, FOUR-M, and GCS, we fit logistic regression models. We took the medians of FOUR-B, FOUR-M, and GCS per person and used them as predictors in separate models. Survival and awakening from coma were outcomes. We performed 5-fold cross validation to gain estimates for AUC, sensitivity, and specificity. Results were aggregated and reported as means and standard deviations.

We estimated the relationship of FOUR-B, FOUR-M, and GCS with time to death, by fitting cox proportional hazard models. The outcome was death and the primary predictors were treated as time dependent. We clustered by individual. Spans are right censored at day 5 for those who survived. Hazard ratios and confidence intervals are reported.

Logistic regression cross validation and Cox hazard models were conducted using R software (version 4.2.1, R Core Team, 2022). Cross validation was conducted using the caret package. Survival analysis was conducted using the survival package.

Results

We screened 130 patients and enrolled 47. Reasons for excluding the other 83 are listed in Fig. 1. Details of subject exams by day for the 47 who were enrolled are in Fig. 2. We treated 16 subjects targeting a core temperature of 33 °C and 31 subjects targeting a core temperature of 36 °C. Median age was 62 years, 57 % were male, 49 % were shockable, and median LOS was 9 days [IQR 9, 15] (Table 1). PCAC IV illness severity was the most common (40 %). All but one subject who awakened survived to hospital discharge, and one subject who did not awaken survived to hospital discharge (awakened n = 18, survived n = 18). Fentanyl and Propofol were the most used sedation agents (Table 1 and Supplemental Data).

Fig. 1.

Subject Screening. IHCA- in-hospital cardiac arrest; OD- overdose; CMO- comfort measures only; CT- computer tomography; DNR- do not resuscitate order.

Fig. 2.

Subject Exam by Day. EEG- electroencephalogram; CMO- comfort measures only; OR- operating room.

Table 1.

Subject Demographics. PCAC- Pittsburgh Cardiac Arrest Category; LOS- Length of Stay; CPC- Cerebral Performance Category; mRS- Modified Rankin Scale.

	Cohort	Survivors	Non-survivors
Age, Years (SD)	62 (16)	57 (14.5)	65 (16)
Sex, Male	27 (57%)	10 (56%)	17 (59%)
Shockable Rhythm	23 (49%)	12 (67%)	11 (38%)
PCAC I	0 (0%)	0 (0%)	0 (0%)
PCAC II	17 (36%)	11 (61%)	6 (21%)
PCAC III	7 (15%)	3 (17%)	4 (14%)
PCAC IV	19 (40%)	3 (17%)	16 (55%)
Median sedative agents (IQR)
Day 1	2 (1,3)	2 (1, 2)	1 (1, 2)
Day 2	1 (0, 2)	2 (0, 2)	1 (0, 2)
Day 3	0 (0, 2)	1.5 (0, 2)	0 (0, 1)
Day 4	0 (0, 1)	0 (0, 2)	0 (0, 1)
Day 5	0 (0, 1)	0 (0, 1)	0 (0, 0.5)
Hospital LOS, Days (IQR)	9 (3, 15)	16 (10, 20)	3 (2, 7)
Awaken	18 (38%)	15 (83%)	3 (10%)
Survive to Hospital Discharge	18 (38%)	18 (100%)	0 (0%)
Function at Hospital Discharge
CPC 1	1 (2%)	1 (2%)	0 (0%)
CPC 2	0 (0%)	0 (0%)	0 (0%)
CPC 3	14 (34%)	14 (78%)	0 (0%)
CPC 4	3 (7%)	3 (17%)	0 (0%)
CPC 5	29 (62%)	0 (0%)	29 (100%)
mRS 0	0 (0%)	0 (0%)	0 (0%)
mRS 1	1 (2%)	1 (6%)	0 (0%)
mRS 2	0 (0%)	0 (0%)	0 (0%)
mRS 3	3 (7%)	3 (17%)	0 (0%)
mRS 4	8 (20%)	8 (44%)	0 (0%)
mRS 5	6 (15%)	6 (33%)	0 (0%)
mRS 6	29 (62%)	0 (0%)	29 (100%)

Overall, neither FOUR-M nor GCS differed over time (p = 0.10; p = 0.07) while FOUR-B increased over time (p < 0.01). FOUR-B, FOUR-M, and GCS differed between survivors and non-survivors over time (p < 0.01) (Table 2). Time to recovery of brainstem and motor function differed between survivors and non-survivors (p < 0.001 for both analyses). [Fig. 2a] FOUR-M and FOUR-B differed between those with good outcome and poor outcome. [Table 2, Fig. 3a] Survival did not differ between those treated with a goal temperature of 33 °C [3/11; 27 %] versus those treated with a goal temperature of 36 °C [15/30; 50 %] (p = 0.19). However, time to recovery of brainstem or motor function differed between those treated at 33 °C and 36 °C (p = 0.0023 and p = 0.0032, respectively) [Fig. 3a].

Logistic regression results using Area Under the Curve of Receiver Operating Characteristics (AUC of ROC), sensitivity, and specificity for survival and awakening results provide more context for clinical decision-making. The mean AUC values regarding survival and awakening were all greater than 0.5, meaning the classification accuracy of FOUR M, FOUR B, and GCS are greater than random probability among this cohort (Table 3). Sensitivity and specificity analysis results indicate that FOUR M has higher negative predictive value than positive predictive value for both survival and awakening (Table 3). Hazard ratios results consistent with FOUR M, FOUR B, and GCS clinical examination results being predictive of survival the first five days following OHCA (Table 4).

Table 3.

FOUR M, FOUR B, & GCS predicting survival and awakening in subjects: Logistic Regression results.

	Outcome - Survival			Outcome - Awoke
Variable	AUC mean(sd)	Sensitivity mean(sd)	Specificity mean(sd)	AUC mean(sd)	Sensitivity mean(sd)	Specificity mean(sd)
FOUR M	0.78 (0.12)	0.61 (0.14)	0.95 (0.11)	0.82 (0.13)	0.65 (0.27)	1 (0)
FOUR B	0.88 (0.13)	0.84 (0.17)	0.85 (0.22)	0.91 (0.12)	0.88 (0.18)	0.82 (0.17)
GCS	0.88 (0.12)	0.92 (0.11)	0.85 (0.22)	0.94 (0.1)	0.96 (0.09)	0.88 (0.16)

Table 4.

Relationship of FOUR M, FOUR B, and GCS with Death using Cox Hazard Ratio Estimates.

Variable	Hazard Ratio	95 % CI	P Value
FOUR M	0.51	(0.34, 0.78)	0.002
FOUR B	0.50	(0.37, 0.67)	<0.001
GCS	0.73	(0.55, 0.97)	0.030

Discussion

Our major novel findings are that the brainstem component of the clinical exam improved over the first 72 h and maximized in both survivors and non-survivors by Day 3; the motor examination did not significantly change over time; and both the brainstem and motor examinations differed between survivors and non-survivors. These data are important as prior work has focused on the significance of a lack of pupillary or corneal response at 72 h. However, little work has focused on the association of a present pupillary or corneal response at 72 h with survival.3, 15, 18 We hypothesized that in subjects who improved, brainstem recovery would precede motor recovery because of the brainstem’s relative resistance to ischemia compared to the cerebral cortex. Our results do not support this hypothesis. Brainstem exam results showed significant improvement overtime in the entire cohort, indicating that brainstem recovery was common to both subjects who awoke and survived and subjects who did not awaken and did not survive. Therefore, based on the results of this study brainstem recovery alone does not differentiate survivors from non-survivors.

Most subjects who went on to awaken did so by Day 3. This is similar to prior work suggesting that most subjects who will awaken do so by Day 3.3, 15 The significant difference in motor exam between the survivors and non-survivors was seen starting on the first day after ROSC persists. This supports prior work demonstrating that initial illness severity is strongly associated with survival.11, 14 However, we do note that survivorship is possible even in those with no motor response on Day 1, supporting prior work cautioning against early prognostication in this patient group. The overall length of stay for non-survivors suggests that early withdrawal of support did not occur in this study.

In this cohort, brainstem and motor recovery appeared to occur simultaneously when the subject regained both functions. The neurological exams were performed daily, which limited the ability to detect minute-to-minute changes. Thus, we are not able to determine if recovery may occur over a few hours. However, our design reflects the frequency of clinical examinations provided to these patients. These results demonstrate that recovery of brainstem reflexes does not always portend an improvement in motor response. They reiterate that motor recovery remains paramount in determining whether a patient is likely to survive and have a meaningful recovery. These results are consistent with a study published in 2009 by Schefold et al., which showed that GCS was higher in OHCA patients who went on to experience good outcomes.¹⁹

There were no differences in outcome between target temperatures 33 °C and 36 °C within the entire cohort, however survivors at target temperatures of 33 °C had longer time to recovery than survivors at 36 °C. These results support Bray et al, who demonstrated patients treated at a target temperature of 36 °C were extubated a day earlier and had decreased duration of sedation compared to patients at a target temperature of 33 °C.²³ We have previously demonstrated that hypothermia alters drug metabolism, providing one potential etiology for these findings.²⁸ Target temperatures assigned to patients were based on severity of clinical status and presents the possibility of indication bias. Differences in target temperatures were included in data analysis and results to consider a pragmatic factor related to clinical practice at our institution that can influence trajectory of neurological examination.

This study was limited by a small sample size. Further enrollment and increased sample size was limited by study team availability. A target sample size was not planned prior to enrollment; our strategy was to screen and enroll as many subjects as study team availability would allow. Given the small sample size of this study, there is potential for Type II error in our results. We provided detailed and standardized examinations for all subjects to strengthen the study design. Study team availability prevented more frequent examinations. However, we believe these results accurately represent the clinical examination frequency completed in most patients resuscitated from cardiac arrest. We also note that the clinical examination is used to determine neurologic prognostication, thus placing this study at risk for a self-fulfilling prophecy. Finally, this study was conducted at a comprehensive post-arrest center with excellent long-term outcomes.² These results may not be generalized to other facilities without standardized post-cardiac arrest care.

Conclusion

The clinical examination, driven by brainstem reflexes, improved over the first 5 days, but an improved brainstem examination did not differentiate between survivors and non-survivors following resuscitation. Recovery of motor function was associated with survival.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

CRediT authorship contribution statement

Alessandra I Cardi: Investigation, Data curation, Project administration, Writing – original draft. Callie M Drohan: Investigation, Data curation. Jonathan Elmer: Investigation, Writing – review & editing. Clifton W Callaway: Investigation, Validation, Writing – review & editing. Francis X Guyette: Investigation. Ankur A Doshi: Investigation. Jon C Rittenberger: Conceptualization, Methodology, Validation, Investigation, Formal analysis, Project administration, Writing – original draft, Visualization.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary material

The following are the Supplementary data to this article: